Stigma in the Labor Market
Keith Finlay∗
November 13, 2014
At the end of 2011, almost seven million US residents were under the supervision of the
correction system. Many times this number have at some point been under the supervision of
the correction system. As ex-offenders are released, they face the challenge of re-entering the
labor market. This paper examines a broad set of policies that influence whether an individual’s
criminal history record is observed by potential employers. Using the first fourteen waves of the
NLSY97, there is evidence that labor market outcomes are worse for ex-offenders when their
criminal histories are easily accessible to employers. Non-offenders from highly offending groups
have significantly better labor market outcomes in the presence of open records. The net effect
of open information supports the notion that employers statistically discriminate in the absence
of criminal history data. Using data from the March CPS, there is evidence that the net effect of
open information is negative for black men, the group with the highest rate of interaction with
the US criminal justice system. Since the CPS does not sample the incarcerated population, these
estimates likely understate how greater employer access to criminal history data will expand the
white-black employment gap.
Keywords: criminal histories, asymmetric information, employer learning, statistical discrimination
JEL codes: D82, D83, J71, J78, K42
∗
Department of Economics, 206 Tilton Hall, Tulane University, New Orleans, LA 70118. The paper has benefitted
from the comments of Abbie Wozniak, Bentley MacLeod, and participants at the 2013 AEA Meetings and the 2013
NBER Summer Institute Meetings on the Economics of Crime. Seth Benzell, Whitney Ruble, Parker Smith, Kelsey
Williams, Peter Baham, Javon Mead, Melissa Tran, and Minzala Mvula provided excellent research assistance. I thank
the National Science Foundation (award SMA-1004569) and the Tulane Research Enhancement Fund for support.
1
Introduction
At the end of 2011, almost seven million US residents were under the supervision of the correction
system (Glaze and Parks, 2012). An unknown but significantly larger number of people has at
some point been under the supervision of the correction system. As ex-offenders are released, they
face the challenge of re-entering the labor market. Legitimate employment is a critical predictor
of criminal desistance, so re-entry is an important consideration for measuring the social costs of
crime. Ex-offenders face stigma in their search for work, and information about their criminal
histories might now be expected to be available in perpetuity in the age of the Internet. This paper
examines a broad set of policies that influence whether an individual’s criminal history record is
observed by a potential employer.
We can categorize these policies into those that affect the legal access to criminal history data
and those that affect the diligence of employer searches. The primary access point to criminal
history data is through online searches. Since the late 1990s, 35 states have begun sharing their
criminal history databases through public websites. This has rapidly reduced the cost of obtaining
criminal history data. While these databases are primarily composed of adult records, databases in
13 states also include juvenile offenses. Moreover, an increasing number of juvenile offenders are
transferred to adult court, and the records for these juveniles will also enter the adult databases.
Given their low employment after release, it follows that ex-offenders have difficulty finding
work. Insofar as there is a negative correlation between criminal offending and productivity in
the legitimate labor market, increased access to criminal history records is expected to worsen
their labor market opportunities. But economic theory predicts effects for non-offenders as well.
Employers have imperfect information about the criminal records of applicants, so rational employers may use observable correlates of criminality as proxies for criminality and statistically
discriminate against groups with high rates of criminal activity or conviction. In the absence of
open records, non-offenders from groups with high conviction rates would be adversely affected.
When accurate criminal history records become easier to obtain, the labor market outcomes of
these non-offenders should improve as employers can determine with greater certainty whether
1/50
applicants have criminal records.
This paper tests these hypotheses using labor market and criminal histories from the National
Longitudinal Survey of Youth 1997 (NLSY97). The criminal history variables in this survey allow
one to distinguish ex-offenders from non-offenders. I also use the criminal histories to model
employer perceptions of criminality assuming that they are based on rational expectations of the
probability of prior conviction. Using the first fourteen waves of the NLSY97, there is evidence
that labor market outcomes are worse for ex-offenders when their criminal histories are easily
accessible to employers. Non-offenders from highly offending groups have significantly better
labor market outcomes in the presence of open records. The net effect of open information supports
the conjecture that employers statistically discriminate in the absence of criminal history data.
The impacts are largest for employment and smallest for wages, with earnings reflecting a
mixture of these results. Economists entertain the idea that employers should be able to discount
in wages any risk created by asymmetric information. My results suggest that employment is a
real barrier, since open information has little effect on wages for ex-offenders or non-offenders
who look like ex-offenders. One possible explanation is that the risk created by negligent hiring
lawsuits is too large to be offset by wage reduction, especially near a binding minimum wage. A
concurrent explanation is that employers can minimize their risk of discrimination lawsuits at the
firing stage by not considering the employment of ex-offenders altogether.1
The NLSY97 has a relative small number of ex-offenders, so I augment the analysis with
repeated cross-sections from the March Current Population Survey (CPS). Ex-offenders cannot be
identified in the CPS, so I estimate net effects of more open criminal history data on demographic
groups. The predicted overall effect of more open information on subpopulations composed of
ex-offenders and non-offenders is ambiguous. Measuring the overall impact for black men is
important given their overrepresentation in the criminal justice system and its implication for the
persisting black-white employment and wage differentials. I find that wages and earnings are lower
for black men under open records (when I model openness as years since records have been open).
1
See the discussion in Donohue and Siegelman (1991).
2/50
As with the NLSY97, wage results are not significantly different from zero. Given that more men
await to be released from prison and these men are not sampled in the NLSY, these estimates likely
understate the long-run effects of criminal history information on the black-white employment gap.
This study makes two important contributions to the empirical literature on the labor market
effects of employer use of pre-employment screening technologies: it exploits exogenous changes
in the employer’s information set to identify the effect of that information and it uses observed
criminal history data to distinguish effects for less desirable applicants (offenders) from more
desirable applicants (non-offenders). The research design makes use of statutory and technological
changes in the amount of criminal history data available to employers. This strategy contrasts with
research that uses variation in employer decisions to conduct criminal background checks, since
these decisions are likely endogenous to the composition of applicant pools (Holzer et al., 2007).
The research design used in this paper is similar to one used by Autor and Scarborough (2008)
to study the diffusion of pre-employment personality tests at a national retail chain. They find
that the relative hiring of blacks did not fall after the introduction of the tests, despite the fact
that blacks in general perform worse on the tests, and they suggest that managers were effectively
statistically discriminating before the tests. Audit studies also attempt to exogenously vary the
information available to employers. For example, Pager (2003) conducts an audit study of the
effect of criminal records and finds that the callback rate for ex-offenders was less than half of the
callback rate for non-offenders.
In addition to providing evidence of statistical discrimination (Arrow, 1973; Phelps, 1972;
Aigner and Cain, 1977), this paper complements theoretical work addressing the efficiency of
stigma. Stigma can increase the legal punishment for committing crime or violating norms, and
potentially reduce the propensity to commit crime (Rasmusen, 1996). But the dynamics are likely
more complicated since stigma reduces the ability of ex-offenders to re-enter legitimate society and
thus increases recidivism (e.g., Blume, 2002). Furuya (2002) shows that the level of employment
of ex-offenders is suboptimal if the employer primarily bears the possible risk, and that sharing
the risk can increase welfare by increasing the employment of ex-offenders and thereby reducing
3/50
recidivism. Funk (2004) shows that this inefficiency can be reduced if repeat offenders receive
harsher punishment.
The results of this paper are also important for understanding the transition of ex-offenders
back into the legitimate labor force. As the flow of released prisoners increases over the next few
decades, the issue of re-entry into the legitimate labor market will force policy makers to consider
the unintended consequences of open criminal history records. Legitimate employment is a strong
predictor of criminal desistance (Sampson and Laub, 1993; Needels, 1996; Uggen, 2000; Lageson
and Uggen, 2013), so expanded use of criminal background checks has the potential to increase
recidivism and the long-term fiscal costs of criminal punishment. But there may also be some
beneficiaries from open records. All else equal, individuals who do not have criminal records
but come from highly offending groups stand to benefit from a more transparent criminal records
system.
2
Stigma and statistical discrimination
Economic theory predicts nuanced effects for non-offenders and ex-offenders that may result from
statistical discrimination by employers. If employers are averse to hiring ex-offenders, then they
have an incentive to use observable correlates of criminality or conviction as proxies for those qualities. Using these proxies, employers can classify individuals as coming from groups with low rates
of conviction (or low perceived criminality) or high rates of conviction (high perceived criminality).
In the absence of open records, one would observe an averaging of the labor market outcomes
for individuals within either group. For example, black men who are high school dropouts have
very high conviction rates. If employers statistically discriminate, then the outcomes for black
non-offenders that have not completed high school will be relatively worse than they would have
been without statistical discrimination, but ex-offenders from that group will have relatively better
outcomes. Similarly, white ex-offenders should benefit from statistical discrimination because they
come from a group with relatively low rates of conviction.
Now suppose that criminal history records become publicly available. If employers can directly
observe criminal history records, they no longer need to rely on statistical discrimination. This
4/50
will cause a separation in the labor market outcomes of ex-offenders and non-offenders within
highly offending groups. Specifically, ex-offenders should do worse and non-offenders should
have improved labor market outcomes. Suppose that perceived criminality is a index created by the
employer using observable proxy variables as a substitute for observed criminality. When criminal
records are available to employers, labor market outcomes bifurcate from the hiring policy under
closed records. Now that employers can distinguish offenders from non-offenders, the labor market
outcomes of non-offenders with high perceived criminality improve. Also note that offenders with
low perceived criminality should suffer a greater decline in labor market outcomes relative to other
offenders.
This model can be used to test the two primary hypotheses generated by the model of statistical
discrimination. First, the model predicts that the main effect of true criminality should become
more negative when employers can access criminal history records. Second, the model predicts
that non-offenders with high perceived criminality should have improved labor market outcomes
if potential employers can verify that they are non-offenders (i.e., when criminal history records
are publicly available). This hypothesis mirrors one in which ex-offenders with high perceived
criminality have relatively worse outcomes from similar non-offenders once records are open. Note
that both of these hypotheses test relative and not absolute effects of employer access to criminal
history data.
The hypotheses above focus on allocative efficiency, but reducing asymmetric information
should also increase productive efficiency. As the quality of labor matches improve, we might
expect the total wage bill to increase. This is difficult to test in a single cohort like the NLSY97,
but the total partial effect of open information might still give some guidance about productive
efficiency gains of increased employer access to criminal history data.
3
Public availability of adult and juvenile criminal history records
The US Equal Employment Opportunity Commission (1987a; 1987b; 2012) has declared that
employers may violate Title VII of the Civil Rights Act of 1964 if they broadly deny employment
to applicants with criminal records, but that employers can ban applicants who have committed
5/50
particular offenses if employers demonstrate these offenses are directly related to a job function.
Some states have more severe restrictions, but there is little evidence of active enforcement until
recently.2 In this section, I highlight the policies that affect criminal history record access.
3.1
Online sources of criminal history records
Before states began to aggregate county-level criminal history records, employers generally had to
hire third-party couriers to go to the county courthouse and do manual searches for names. Data
aggregation and automation reduces the transaction costs for employers and increases the jurisdictional scope of criminal records searches. State-level aggregation of criminal history databases
was prompted by passage of the Brady Handgun Violence Prevention Act, which mandated the
creation of a national criminal history database to be used for gun-purchase background checks
(Finlay, 2009).
Data in state criminal information systems is generated in a number of ways. The left half of
Figure 1 shows the ways that a criminal offense can become publicly known to a potential employer. The figure also documents the statutory variation used to identify the information effects.
The primary sources of criminal history data are state departments of corrections, but court cases
are sometimes available through state judicial websites. Table 1 shows when states first released
records online along with some characteristics of those repositories: whether the repositories
include only current offenders and whether background checks require a fee. Starting with earlier
compilations of online criminal history databases (Finlay, 2009; DeBacco and Greenspan, 2011;
Lee, 2012), I use the Internet Archive’s Wayback Machine, an historical archive of webpages, to
determine a date after which state criminal history data was available to the public online.3
2
There have been relatively few Title VII cases involving employment of ex-offenders, and
employers have won almost all of these cases (Connor and White, 2013). One explanation
for the small number of ex-offender employment discrimination cases is that much of this
discrimination happens at the hiring stage. Insofar as hiring discrimination is a diminished fraction
of discrimination cases (Donohue and Siegelman, 1991), it may be difficult for ex-offenders
to bring these cases to trial. The EEOC under the Obama Administration has taken a more
assertive role in charging firms who broadly exclude ex-offenders from hiring (Equal Employment
Opportunity Commission, 2013).
3
The Wayback Machine’s archives start in 1996 and can be accessed at http://web.archive.
6/50
Since the 1998, 35 states have begun sharing their criminal history databases through public
websites. Of these, 13 states require searchers to pay some cost. South Dakota is the only state
that appears to have explicit statutes against publicizing (non-sex crime) criminal history data.4
Any casual search for criminal records demonstrates the multitude of third-party vendors offering background checks. These firms acquire data by purchasing them directly from states
and counties or by scraping public online databases with a Web crawler. Since criminal history
data are part of the public record, it is difficult for state governments to ask firms to remove data
obtained by scraping. I view the third-party vendors as magnifying the access created by the state
governments, and so do not separately measure the role of third-parties but acknowledge that they
are an important part of falling transaction costs for employers.5
As the costs of obtaining criminal history have decreased, the proportion of employers conducting criminal background checks on applicants has increased. Figure 2 shows evidence of
increasing use of background checks by employers. The first two data points come from surveys
of Los Angeles employers in 1992–1994 and 2001 (Holzer et al., 2007) and the last three data
points come from national surveys conducted by the Society for Human Resource Management
(Burke, 2005; [SHRM] Society for Human Resource Management, 2010, 2012). The SHRM
surveys probably oversample large employers, who reflect its membership. But taken together,
the surveys show that approximately one third of employers always conducted checks in the early
1990s, while recently about two thirds of employers always check the criminal backgrounds of
applicants. Unfortunately, there are no surveys that identify employer state, which could be used
to directly establish the link between public provision of criminal history data and the likelihood
that employers conduct background checks.
org.
4
See SDC 23-5-7 and 24-2-20.
There is some evidence that data quality provided by third-party vendors is relatively low
(Bushway et al., 2007), which perhaps reflects the nonsystematic ways they obtain criminal history
data.
5
7/50
3.2
Criminal majority and juvenile transfer provisions
A growing body of criminal history data is generated by juvenile interactions with the criminal
justice system. Public juvenile records can be created if a state does not explicitly protect the
privacy of juveniles, if a juvenile is transferred to adult court and the state provides adult records to
the public, or if the juvenile is involved in a sex offense and the state provides sex offense records
to the public.
The proportion of juveniles taken into police custody who were transferred to criminal or adult
court doubled from 1990–2010. Figure 3 shows the disposition of juveniles taken into police
custody annually (from the FBI’s Uniform Crime Reports). Although the total proportion of
juveniles referred to adult court is low, it is the fastest growing disposition category. There is some
growth in referrals to juvenile court, and these two trends are offset by a corresponding decrease in
juveniles whose status is handled within the police department and who are subsequently released.
This figure likely understates the role of adult courts in processing juvenile felonies, since cases
handled within departments tend to be minor offenses and cases sent originally to juvenile court
may eventually find their way to adult court.
There are several different mechanisms by which states determine whether to try a juvenile as
an adult or a child. The most important of these is the age of criminal majority, at which individuals
are treated as adults. The age of criminal majority varies between 16 and 18 years of age across US
states. There are also several types of mandatory and discretionary transfer provisions, which allow
children to be moved to the adult criminal justice system if certain criteria are met. For example,
a 16 year old who commits armed robbery in Alabama is automatically transferred to adult court.
Over the last two decades, most states have added mechanisms like this that increase the likelihood
that juvenile offenders are processed in adult courts. Table 2 shows how these rules have become
more severe over the last twenty years. Juvenile sex crimes are excluded from these data because
state laws are difficult to characterize in a consistent way. As discussed below, juvenile sex crime
data is broadly available to the public because of “Megan’s laws” registries.
Criminal majority statutes, and to a lesser extent juvenile transfer provisions, have been used to
8/50
study the effect of punishment incentives on the criminal propensities of juveniles. Levitt (1998)
finds that ascending to the age of majority is associated with a significant drop in crime rates,
once changes in the relative punitiveness of adult and youth courts are taken into account. Lee
and McCrary (2009) focus on Florida from 1995 to 2002 and find very little effect around the
age of majority. Similar research designs on a variety of offender datasets have been used to find
little deterrent effect of criminal majority statutes (McCrary and Sanga, 2011) or juvenile transfer
statutes (Finlay, 2012; Cohn and Mialon, 2011; Steiner et al., 2006; Singer and McDowall, 1988).
One explanation for these results is that young people do not fully update their expectation of the
change in severity of punishment at the age of majority (Hjalmarsson, 2009).
3.2.1
General juvenile records
Since the early 1990s, states have relaxed confidential treatment of juvenile offender data, both in
terms of arrest and court records (Torbet et al., 1996). Table 1 shows which states allow juvenile
offender data to be compiled in adult criminal databases.6 Databases in 13 states also include
juvenile offenses. Moreover, an increasing number of juvenile offenders are transferred to adult
court, and the records for these juveniles will also enter the adult databases. This variable in
conjunction with the online access variable indicates whether juvenile records are broadly available
to the public online.
3.2.2
Juvenile and adult sex offenders
The US Congress mandated states to implement sex offender registries in 1994 (PL 103-322). In
1996, Congress further required states to release information to the public about sex offenders
deemed dangerous to the public (PL 104-236). Using data from Agan (2011), Table 1 shows
when states began to publish sex offender registries over the Internet. In general, sex offender
records became available earlier than general criminal records and all states had online sex offender
repositories by 2006.
6
Juvenile records availability comes from http://www.beforeyouplea.com and tests of
online repositories with recent birth dates (as of June 2013).
9/50
4
Longitudinal analysis using the NLSY97
4.1
Data
In this section, I examine the effects of more open criminal history on longitudinal labor market
outcomes in the 1997 cohort of the NLSY97. The NLSY97 includes a nationally representative
sample of all youths born between 1980 and 1974, and an over-sample of black and Hispanic
youths born in the same years. The paper uses 14 rounds of NLSY97 covering interviews from
1997through 2011. The NLSY97 has information about both the criminal activity of respondents
and their labor market outcomes. The sample period also coincides with the introduction of
Internet sites for accessing criminal history records. The criminal records policies discussed above
are matched with individual respondents using the state geocodes available in the private-release
version of the NLSY97. The regression sample consists of men and women aged at least 18 years
in the representative sample and the minority oversample.7
I use three labor market outcomes as dependent variables: employment status, the natural log
of the hourly wage, and the inverse hyperbolic sine of annual earnings.8 Employment status is
equal to one if the respondent was employed at the date of the interview. Hourly wage is the
NLSY97-created hourly wage for the primary job since the last interview. The earnings variable is
the total income from wages and salary in the calendar year before the interview.9
The NLSY97 also has extensive information on interactions with the criminal justice system.
An iterative round of questions addressed any arrests and whether they led to conviction or incarceration. I create indicator variables for whether the respondent had been convicted or incarcerated
at the time of the interview or since the date of the last interview. Individual crimes are matched
to criminal majority and juvenile transfer provisions based on the age-crime profile and the state
7
Custom sampling weights for NLSY97 respondents in any survey year come from http:
//www.nlsinfo.org/web-investigator/custom_weights.php.
√
8
The inverse hyperbolic sine, sinh−1 (x) = ln(x + x2 + 1), is an alternative to the natural
logarithm when one wants to include zeroes in the analysis (Burbidge et al., 1988; MacKinnon and
Magee, 1990). Away from zero, the derivatives of the two functions are approximately equal.
9
Wages and earnings are inflated to 2005 dollars using the All-Urban series of the Consumer
Price Index.
10/50
and year in which the crime occurred. Crimes committed above the age of criminal majority
are considered to be visible to employers when records are open. Crimes eligible for mandatory
transfer to adult court are also coded as visible to employers under open records. One concern with
self-reported data is that respondents may underreport criminal behavior, but NLSY97 respondents
have qualitatively similar age-crime profiles as those generated from larger administrative datasets
(Brame et al., 2012, 2014).
I also include a number of other variables as controls. To control for labor market experience, I
use the hours of accumulated labor market experience from age 13. Education controls include
highest grade completed and a set of dummies for highest degree received as of June 30 of
the survey year (namely, whether the individual has a general equivalency diploma (GED), a
high school diploma, an associates degree, or a bachelors or postgraduate degree). I also use
calculate the body mass index and include in the vector of observables considered by employers.
In regressions without individual fixed effects, the Armed Services Vocational Aptitude Battery
(ASVAB) test score and race, ethnicity, and gender indicators serve as controls.
The sample is restricted to respondents aged 18–31 years, although juvenile labor market and
criminal justice experience is expressed in the cumulative variables. Table 3 shows means and
standard deviations of variables for the analytic sample by age.
Although laws vary enough from state to state to make recording every nuance in their variation
impossible, I have recorded all of the common criteria for transfer, such as age, offense type, prior
criminal history, and treatment conditioned on association with gangs or firearms. These data cover
all transfer types in the District of Columbia and the 50 US states from 1997 through 2011. I merge
this statutory database with NLSY97 arrest data to examine the effect of changes in mandatory and
discretionary transfer severity on changes in arrest rates by age and offense category.
4.2
Model and identification
Define the vector of policies that affect employer access to criminal history data as A. Elements
of A are defined in such a way that higher values are associated with lower costs for accessing
criminal history data. Without loss of generality, suppose A is scalar and binary, so it is equal to
11/50
one if criminal history records are available to employers online, and zero otherwise.
Now suppose that employers use observable characteristics to make forecasts of applicant
criminality. They base this prediction on a vector of individual characteristics Z, whose elements
are easily observable and are known to be correlated with criminality or prior conviction. Suppose
employers create a regression-weighted index of variables in Z, and use this as a proxy for criminality in their hiring decisions. Since prior conviction is a low-probability event, I use a probit
model of the following form:
P [Wit = 1] = P [α0 + Zit α1 + ηit > 0] = Φ[α0 + Zit α1 + ηit ],
(1)
where Wit is a measure of criminal history for individual i in year t, ηit is an error component, and
Φ[·] is the cumulative distribution function of the standard normal distribution. After estimating
\
this regression, employers can predict a measure of perceived criminality: Cit ≡ P [W
it = 1] =
Φ[α̂0 + Zit α̂1 ].
Now consider labor market outcomes as a function of employer access to criminal history data,
employer background check intensity, and perceived criminality:
Yit = Xit β0 + β1 1[Wit = 0]1[Ait = 1] + β2 1[Wit = 1]1[Ait = 0]
(2)
+β3 1[Wit = 1]1[Ait = 1] + β4 1[Wit = 0]1[Ait = 1]Cit
+β5 1[Wit = 1]1[Ait = 0]Cit + β6 1[Wit = 1]1[Ait = 1]Cit
+β7 Cit + εit ,
where Y is a labor market outcome, X is a vector of regressors, 1[·] is the indicator function, and
εit is an error term.
Three coefficients in Equation 2 (β1 , β2 , and β3 ) measure the intersection of actual criminal
history and record openness:
12/50
Actual criminal history
Observability
closed
open
non-offender
ex-offender
Has no record, closed records (omitted)
Has record, closed records (β2 )
Has no record, open records (β1 )
Has record, open records (β3 )
These three dummies are then interacted with the perceived criminality estimate (yielding β4 , β5 ,
and β6 ). Those six parameters are sufficient to examine the effects of eliminating asymmetric
information. In particular, the effect of information on the outcomes for ex-offenders is equal to
β1 + Cβ4 . The effect of information on the outcomes of non-offenders is β3 + Cβ6 − β2 − Cβ5 .
Then, the net effect of information on the relative outcomes of non-offenders is β1 + Cβ4 − β3 −
Cβ6 + β2 + Cβ5 .
There are two primary threats to identification in this model. First, any attempt to identify
the labor market impacts of conviction or incarceration require one to account for the unobserved
heterogeneity between offenders and non-offenders (e.g., Grogger (1995); Kling (2006)). I use
two alternative strategies: exploit the panel structure of the NLSY97 and include a individual fixed
effect or include the time-invariant ASVAB score along with a dummy variable identifying if the
score is missing. Also note that the estimated perceived criminality index probably also accounts
for some of the unobserved heterogeneity.
The second threat to identification broadly concerns labor market differences in states with
more open records policies versus those with less open records policies. If these states have more
punitive punishment regimes, we might also expect the labor market outcomes of ex-offenders
to be worse. This concern is mediated by the gradual opening of criminal history records across
states, which I have identified as being primarily caused by technological hurdles in Brady Act
compliance across states (Finlay, 2009). The source of state variation in juvenile policies is,
unfortunately, less clear. Finally, note that a confound for the information effects in this model
must differentially affect non-offenders from ex-offenders conditional on perceived criminality.
By conditioning on perceived criminality, I hope to also raise the bar for confounds caused by
across-state variation in labor market conditions.
13/50
4.3
Results
I first show estimates used to construct measures of perceived criminality. Figure 4 shows kernel
density estimates of predicted probability of a respondent having ever been convicted of a serious
crime. Probit regressions were estimated separately by gender, racial, and ethnic groups.10 The
top panel shows the distributions for black, Hispanic, and all other men. The bottom panel shows
the analogous distributions for women. Men in general are more likely to be convicted of a crime.
There is some racial and ethnic differences in the distributions—black men and non-black, nonHispanic women are more likely to have convictions. The racial differential in conviction rates is
quite small compared to the racial differential in prior conviction rates (both in the population and
in this sample). The predicted probabilities of prior conviction in Figure 4 will serve as a measure
of employers’ perception of criminality given an assumption of rational expectations.
With an estimate of perceived criminality, I estimate regressions of labor market outcomes as
a function of the typical regressors, actual criminal history, perceived criminal history, and records
openness. Table 4 focuses on the parameters necessary to test hypotheses on asymmetric information and statistical discrimination.11 The 90th percentile of the perceived criminality distribution
is 0.35, and I use these in the linear combination of estimated parameters.
In Table 4, odd columns are estimated as pooled cross-sectional samples but include timeinvariant regressors. The even columns exploit the panel structure with individual fixed effects.
For individuals with a criminal record, all but one estimate of the information effect is negative,
although only the earnings estimate in the pooled model is significantly different from zero. These
estimates are consistent with somewhat worse outcomes for ex-offenders once employers have
easier access to criminal history data. For non-offenders, all but one estimate is positive and
only the pooled employment model has a statistically significant result. The net effect is also
of interest because it measures the bifurcation in the outcomes for non-offenders relative to exoffenders. These estimated effects are consistently positive and the employment and earnings
10
11
The full results are in Table A-1.
Full results can be found in Table A-2.
14/50
estimates are both statistically significant in the pooled models. The fixed effects estimates are
similar magnitudes for employment and earnings, while the wage estimates seem indistinguishable
from zero in both specifications. In economic context, open records improve the relative likelihood
of employment of non-offenders by 2.6 percentage points and the relative annual earnings of nonoffenders by 12 percent. Given the poor labor market outcomes of individuals in these high-risk
groups, these are substantial improvements.
Figure 5 explores the results as a function of the full range of perceived criminality. Recall
that most of the sample has a predicted probability of conviction below 0.4 for men and 0.2 for
women. In these ranges, the net information effects for employment and earnings are clearly
positive and statistically significantly different from zero. One might expect these lines to slope
upwards—meaning non-offenders from highly offending groups benefit the most from more open
information. These empirical models are not capable of distinguishing those difference, but a richer
dataset with more ex-offenders may allow for the identification of nonlinear effects of information.
For wages, the point estimates stay quite close to zero across the range of perceived criminality.
5
5.1
Cross-sectional analysis of the net effects of open criminal histories on black men
Data, model, and identification
One weakness of the NLSY97 data is the small sample of individuals who report interactions with
the criminal justice system. An alternative approach to understanding the effect of open criminal
history data is to take these policies to a much larger dataset in which I do not have information
on criminal behavior. Here, I use the repeated sample from the March Current Population Survey
to examine the net effects of information on the labor market outcomes of individuals in various
demographic groups. The March waves have detailed labor market data of a representative sample
of working-age individuals from 1997–2011.
With CPS data, I estimate the following labor market outcomes regressions:
yiast = Xi β + di Xist βd + Rast γ + di Rast γd + εiast ,
15/50
where yiast is a labor market outcome for individual i of age a in state s in year t. The records
variable, R, can be defined either as an indicator for whether records are open in state s in year
t or the number of years records have been open since the individual reached the state’s age of
majority. Samples are stratified by a vector of demographic identifiers di , which identify black
men, Hispanic men, white (non-black, non-Hispanic) women, black women, and Hispanic women.
A vector of controls Xist includes discretized highest grade completed, a quadratic of potential
experience, a central city indicator, a union indicator, the effective state minimum wage, the state
unemployment rate, year fixed effects, and state fixed effects. Any control that enters the regression
is also interacted with the demographic indicators, so that the parameters of interest γd do not
reflect the differences in return to characteristics across demographic groups.
I consider a number of labor market outcomes y: whether an individual is currently employed,
her log hourly wage, the inverse hyperbolic sine of her earnings, whether she receives either health
insurance or pension benefits from her employers, whether she works at a firm with more than 500
employees, and the mean education of the occupation in which the individual is employed. I also
examine the school enrollment decision.
5.2
Results
Results from a prime-age sample can be found in Table 5. In odd columns, I model open records as
a indicator. In even columns, it is measured as a duration: in how many years have criminal records
been available since the individual passed the state’s age of majority. For employment, none of the
demographic interactions with the binary records variable are statistically significantly different
from zero. When availability is modeled as a duration, there is a 0.4 percentage point reduction in
employment probabilities for black men per additional year of record openness. Given a 0.15 gap
in employment rates between white and black men in states that never opened their criminal records
(shown below in the table), ten years of record availability would be equivalent to a 27% increase
in the black-white employment gap. For wages, there is no statistically significant evidence of a
differential effect of more information for black men. For earnings, black men have approximately
2% lower earnings for each additional year of record openness (Column 6). These results for
16/50
employment and earnings but not wages, in a broad sample of working adults, are consistent with
the earlier result from the NLSY97.
I now examine a subsample that we expect to be more greatly impacted by open criminal
records: younger people with less education. For Table 6, I restrict the sample to those aged 18–35
years with at most a high school education. The effects for this younger, less skilled sample are
qualitatively similar to the ones in Table 5. Black men have lower employment rates and earnings
in states that have had greater public access to criminal history data. The employment impact for
black men is roughly twice the size here, while the earnings loss is roughly three times as large.
This suggests that open records reduce employment opportunities more rapidly for young, black
ex-offenders than they improve them for young, black non-offenders.
The next set of results come from alternative outcomes in which the theoretical effect of more
information is even less clear. In Table 7, the first two columns have as an outcome an indicator for
workers provided with health insurance or pension benefits by employers. In the next two columns,
large firm is an indicator for workers at firms with at least 500 employees. Occupational skill is
the mean of highest grade completed for the current occupation—from a 1994–1996 CPS sample.
And enrolled is an indicator for school enrollment.
For firms with employee benefits and large workforces, human resource departments are likely
to be more sophisticated and possibly more adept at finding the criminal history data of applicants,
but they might also be large enough to handle the potential risk of hiring ex-offenders. It is well
documented that black men and women are more likely to work at larger firms, possibly because
their more formal hiring processes reduce discrimination. For these outcomes there does not
appear to be a consistent effect of more open information on black men. With a binary treatment,
black men are more likely to have benefits under open records, but the sign changes in the other
specifications. An even larger effect is found for the younger, less educated sample in Table 8,
Column 1.
For occupational skill levels, open records have a mixed effect on the black men. Occupational
skill conveys something about the qualifications necessary to obtain employment in a particular
17/50
position. For the price-age sample, the duration treatment is associated with black men obtaining
employment in occupations with lower skill levels. Finally, I look at the enrollment decision. One
possible response to negative information in the labor market is to seek an alternative positive
signal. School enrollment is one such path. There is not statistically significant evidence that black
men increase or decrease school enrollment when criminal records are publicly available.
With cross sectional data, I can be less certain I are identifying the causal effect of record
openness. In addition, the CPS does not sample the currently incarcerated. Insofar as these
individuals are less likely to have positive labor market outcomes (Western and Pettit, 2000), the
estimated effects of open information should be smaller in an absolute sense then they would be if
the currently incarcerated were in the labor market and sampled by the CPS.
6
Conclusion and discussion of policy
The US has experienced a massive increase in incarceration since the 1970s that has only recently
started to slow. Researchers have begun to understand how incarceration policy affects crime
(Johnson and Raphael, 2012), but also the long-term ramifications of crime policy on the labor
market (Pager et al., 2009b,a). Millions of individuals have criminal records and the Internet makes
these data broadly available to the public. If crime is correlated with productivity in the legitimate
labor market, rational employers will statistical discriminate to avoid hiring ex-offenders. With a
simple policy dichotomy of open versus closed criminal records, greater access to criminal history
data may help non-offenders who otherwise look like offenders. But this access also furthers the
stigma experience by ex-offenders in the labor market and may therefore increase recidivism.
In this paper, I explore a broad set of policies that affect whether criminal history data is
available to employers. I find evidence that more open criminal history data is associated with
improvements in employment and earnings of non-offenders relative to ex-offenders. This supports
the hypothesis that employers statistically discriminate to avoid hiring ex-offenders. But the
evidence also refutes the hypothesis that employers are willing to lower wages to offset the risk of
hiring ex-offenders. The initial hiring decision appears to be the primary difficulty for ex-offenders
or non-offenders from highly offending groups.
18/50
Although non-offenders from highly offending groups have improved labor market outcomes
under open records, my CPS results show that these benefits of open records have not been large
enough to counteract the worse outcomes for black ex-offenders. Black men do not appear to have
broadly improved labor market outcomes under open records. And since the CPS does not sample
the institutionalized population, I argue that estimates of the net effect of open information on
black men must only get worse with no change in criminal justice, human capital, or employment
policy addressing the low employment of ex-offenders.
A range of public policies has been considered for addressing the low employment of exoffenders, especially as it affects recidivism. The most common approaches deal with the creation
and distribution of criminal history data (expulsion and sealing) or place restrictions on the use of
criminal history data during the hiring process (“ban the box”).
Most states have some statutory process for expelling criminal records. These are usually available for juvenile crimes or for those wrongly convicted. Generally, these procedures are incredibly
costly, so that it would be difficult to imagine them being used on a regular basis. An alternative
policy puts an expiration date on criminal history records, so that an individual’s rap sheet would
become hidden or expelled after a certain amount of time has passed since adjudication. But most
importantly for the context of this research, expulsion and data expiration have virtually no impact
on third-party vendors of criminal history data.
Although it has been declared discriminatory by the Equal Employment Opportunity Commission (2012, 1987a,b), some employers ban all ex-offenders from applying for jobs.12 Under the
Obama administration, the EEOC has become more active in pursing cases against employers with
these policies, arguing that they disproportionately affect black applicants (Equal Employment
Opportunity Commission, 2013).
A more active legislative response requires employers to remove job application questions
about criminal records until some later stage in the hiring process—a policy called “ban the box”.
This effort began with efforts by city council in some large municipalities and focused on public
12
Although the norm is to bar ex-offenders from employment, there are some employers who
voluntarily ignore criminal history records (e.g., Men’s Wearhouse).
19/50
sector hiring. The most significant “ban the box” legislation recently became law in Minnesota.
Starting January 1, 2014, private employers in Minnesota may not inquire or consider a criminal
record of an applicant until the applicant has been selected for an interview.13 (Minnesota law
previously had this restriction only for public positions.) This type of law still allows employers
to consider previous offenses as part of the final hiring decision if they are deemed relevant for the
position. But employers must first consider the applicant’s other qualifications in the first round of
the hiring process.
Expanding “ban the box” to private employers ignores the real concern of employers about
some ex-offenders—it treats all ex-offenders the same. Evidence in this paper suggests that
restricting access to information has broad effects on highly offending group, such as black men.
One rationale for the broad employer bans of ex-offenders is the uncertainty about potential
liability. Firms can be sued by customers or employees injured by an employee who the firm
was negligent to hire. There is considerable uncertainty about what constitutes negligence during
the hiring process and what types of harmful workplace behavior could be foreseeable given a
particular criminal record (Hickox, 2011; Peebles, 2012). Uncertainty about liability could be
reduced by standardizing the due diligence requirement during hiring or possibly capping the
damages allowable under negligent hiring if particular due diligence is met. Under most statutory
proposals addressing negligent hiring, criminal background checks would remain important during
the hiring process since they constitute a central part of due diligence.
A bolder negligent hiring proposal might be for the government to share even more information
about ex-offenders. Departments of corrections compile risk ratings for use during probation and
parole hearings. These data might be useful to employers to understand the likelihood of productive
work or workplace violence? Sharing this information with employers allows for heterogeneous
treatment and productive employment. For currently supervised offenders, wider sharing of risk
rating data may also provide a more direct incentive for criminal desistance. Natural parallel
programs could provide inmates with opportunities for human capital development.
13
See
https://www.revisor.mn.gov/bills/text.php?number=SF523&session_year=
2013&session_number=0.
20/50
Post-release employment bonds could help competing government agencies internalize the
dynamic public costs of recidivism. Like social investment bonds, these could be risk rated using
the same offender risk data discussed above. Research has shown that early employment after
release and the skill-appropriate job opportunities are critical for desistance (Schnepel, 2012). This
highlights the importance of studying how employers view the hiring of ex-offenders.
21/50
References
Agan, Amanda Y. 2011. Sex Offender Registries: Fear without Function? Journal of Law and
Economics 54(1): 207–39.
http://www.jstor.org/stable/10.1086/658483
Aigner, Dennis J. and Glen G. Cain. 1977. Statistical Theories of Discrimination in Labor Markets.
Industrial and Labor Relations Review 30(2): 175–87.
http://www.jstor.org/stable/2522871
Arrow, Kenneth J. 1973. The Theory of Discrimination. In Orley Ashenfelter and Albert Rees,
editors, Discrimination in Labor Markets. Princeton, NJ: Princeton University Press.
http://www.amazon.com/Discrimination-Labor-Markets-Orley-Ashenfelter/dp/
0691041709
Autor, David H. and David Scarborough. 2008. Does Job Testing Harm Minority Workers?
Evidence from Retail Establishments. Quarterly Journal of Economics 123(1): 219–77.
http://qje.oxfordjournals.org/content/123/1/219.abstract
Blume, Lawrence. 2002. Stigma and Social Control. Cornell University. Mimeo.
http://www.ssc.wisc.edu/econ/Durlauf/networkweb1/wpapers/es-119.pdf
Brame, Robert, Shawn D. Bushway, Ray Paternoster, and Michael G. Turner. 2014. Demographic
Patterns of Cumulative Arrest Prevalence by Ages 18 and 23. Crime Delinquency 60(3): 471–
486.
http://cad.sagepub.com/content/60/3/471.abstract
Brame, Robert, Michael G. Turner, Raymond Paternoster, and Shawn D. Bushway. 2012.
Cumulative Prevalence of Arrest from Ages 8 to 23 in a National Sample. Pediatrics 129(1):
21–27.
http://pediatrics.aappublications.org/content/129/1/21.abstract
Burbidge, John B., Lonnie Magee, and A. Leslie Robb. 1988. Alternative Transformations
to Handle Extreme Values of the Dependent Variable. Journal of the American Statistical
Association 83(401): 123–27.
http://www.jstor.org/stable/2288929
Burke, Mary Elizabeth. 2005. 2004 Reference and Background Checking Survey Report.
Alexandria, VA: Society for Human Resource Management.
http://www.shrm.org/Research/SurveyFindings/Documents/Reference%20and%
20Background%20Checking%20Survey%20Report.pdf
Bushway, Shawn, Shauna Briggs, Faye Taxman, Meridith Thanner, and Mischelle Van Brakle.
2007. Private Providers of Criminal History Records: Do You Get What You Pay For? In Shawn
Bushway, Michael A. Stoll, and David F. Weiman, editors, Barriers to Reentry? The Labor
Market for Released Prisoners in Post-Industrial America. New York: Russell Sage Foundation.
http://muse.jhu.edu/books/9781610441018
22/50
Cohn, Jacob and Hugo M. Mialon. 2011. The Impact of Juvenile Transfer Laws on Juvenile Crime.
Emory Law and Economics Research Paper No. 10-63.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1606002
Connor, Terence G. and Kevin J. White. 2013. The Consideration of Arrest and Conviction Records
in Employment Decisions: A Critique of the EEOC Guidance. Seton Hall Law Review 43(3):
Article 3.
http://scholarship.shu.edu/shlr/vol43/iss3/3
DeBacco, Dennis A. and Owen M. Greenspan. 2011. Survey of State Criminal History Information
Systems, 2010. Bureau of Justice Statistics.
https://www.ncjrs.gov/pdffiles1/bjs/grants/237253.pdf
Donohue, John J and Peter Siegelman. 1991. The Changing Nature of Employment Discrimination
Litigation. Stanford Law Review 43(5): 983–1033.
http://www.jstor.org/stable/1228893
Equal Employment Opportunity Commission. 1987a. Policy Statement on the Issue of Conviction
Records under Title VII of the Civil Rights Act of 1964. Notice N-915.
http://www.eeoc.gov/policy/docs/convict1.html
———. 1987b. Policy Statement on the Use of Statistics in Charges Involving the Exclusion of
Individuals with Conviction Records from Employment. Notice N-915-061.
http://www.eeoc.gov/policy/docs/convict2.html
———. 2012. Consideration of Arrest and Conviction Records in Employment Decisions Under
Title VII of the Civil Rights Act of 1964. Notice 915-002.
http://www.eeoc.gov/laws/guidance/arrest_conviction.cfm
———. 2013. EEOC Files Suit Against Two Employers for Use of Criminal Background Checks.
Press release published by author.
http://www.eeoc.gov/eeoc/newsroom/release/6-11-13.cfm
Finlay, Keith. 2009. Effect of Employer Access to Criminal History Data on the Labor
Market Outcomes of Ex-Offenders and Non-Offenders. In David Autor, editor, Labor Market
Intermediation, pp. 89–125. Chicago: University of Chicago Press.
http://www.nber.org/books/auto07-1
———. 2012. The Effectiveness of Criminal Majority and Juvenile Transfer Laws. Unpublished
manuscript.
http://econ.tulane.edu/kfinlay/criminal_majority.pdf
Funk, Patricia. 2004. On the Effective Use of Stigma as a Crime-Deterrent. European Economic
Review 48(4): 715–28.
http://www.sciencedirect.com/science/article/pii/S0014292103001430
Furuya, Kaku. 2002. A Socio-Economic Model of Stigma and Related Social Problems. Journal
of Economic Behavior and Organization 48(3): 281–90.
http://www.sciencedirect.com/science/article/pii/S0167268101002311
23/50
Glaze, Lauren E. and Erika Parks. 2012. Correctional Populations In The United States, 2011.
Bureau of Justice Statistics NCJ 239972.
http://bjsdata.ojp.usdoj.gov/content/pub/pdf/cpus11.pdf
Grogger, Jeffrey. 1995. The Effect of Arrests on the Employment and Earnings of Young Men.
Quarterly Journal of Economics 110(1): 51–71.
http://qje.oxfordjournals.org/content/110/1/51.abstract
Hickox, Stacy A. 2011. Employer Liability For Negligent Hiring Of Ex-Offenders. Saint Louis
University Law Journal 55: 1001–46.
http://www.slulawjournal.org/volume55/hickox_employer/
Hjalmarsson, Randi. 2009. Crime and Expected Punishment: Changes in Perceptions at the Age
of Criminal Majority. American Law and Economics Review 11(1): 209–48.
http://aler.oxfordjournals.org/content/11/1/209.abstract
Holzer, Harry J., Steven Raphael, and Michael A. Stoll. 2007. The Effect of an Applicant’s
Criminal History on Employer Hiring Decisions and Screening Practices: Evidence from Los
Angeles. In Shawn Bushway, Michael A. Stoll, and David F. Weiman, editors, Barriers to
Reentry? The Labor Market for Released Prisoners in Post-Industrial America. New York:
Russell Sage Foundation.
http://muse.jhu.edu/books/9781610441018
Johnson, Rucker and Steven Raphael. 2012. How Much Crime Reduction Does the Marginal
Prisoner Buy? Journal of Law and Economics 55(2): 275–310.
http://www.jstor.org/stable/10.1086/664073
Kling, Jeffrey R. 2006. Incarceration Length, Employment, and Earnings. American Economic
Review 96(3): 863–76.
http://www.jstor.org/stable/30034076
Lageson, Sarah and Christopher Uggen. 2013. How Work Affects Crime—And Crime Affects
Work—Over The Life Course. In Chris L. Gibson and Marvin D. Krohn, editors, Handbook of
Life-Course Criminology, pp. 201–12. Springer New York.
http://dx.doi.org/10.1007/978-1-4614-5113-6_12
Lee, Dara N. 2012. The Digital Scarlet Letter: The Effect of Online Criminal Records on Crime.
Working paper.
http://web.missouri.edu/˜leedn/OnlineRecords_DLee.pdf
Lee, David S. and Justin McCrary. 2009. The Deterrence Effect of Prison: Dynamic Theory and
Evidence. Working paper.
http://www.econ.berkeley.edu/˜jmccrary/lee_and_mccrary2009.pdf
Levitt, Steven D. 1998. Juvenile Crime and Punishment. Journal of Political Economy 106(6):
1156–85.
http://www.jstor.org/stable/2990736
24/50
MacKinnon, James G. and Lonnie Magee. 1990. Transforming the Dependent Variable in
Regression Models. International Economic Review 31(2): 315–39.
http://www.jstor.org/stable/2526842
Maguire, Kathleen, editor. 2013. Sourcebook of Criminal Justice Statistics [Online]. Bureau of
Justice Statistics.
http://www.albany.edu/sourcebook
McCrary, Justin and Sarath Sanga. 2011. Youth Offenders and the Deterrence Effect of Prison.
Mimeo.
http://dx.doi.org/10.2139/ssrn.1980326
Needels, Karen E. 1996. Go Directly to Jail and Do Not Collect? A Long-Term Study of
Recidivism, Employment, and Earnings Patterns among Prison Releasees. Journal of Research
in Crime and Delinquency 33(4): 471–96.
http://jrc.sagepub.com/content/33/4/471.abstract
Pager, Devah. 2003. The Mark of a Criminal Record. American Journal of Sociology 108(5):
937–975.
http://www.jstor.org/stable/10.1086/374403
Pager, Devah, Bruce Western, and Bart Bonikowski. 2009a. Discrimination in a Low-Wage Labor
Market: A Field Experiment. American Sociological Review 74(5): 777–99.
http://asr.sagepub.com/content/74/5/777.abstract
Pager, Devah, Bruce Western, and Naomi Sugie. 2009b. Sequencing Disadvantage: Barriers
to Employment Facing Young Black and White Men with Criminal Records. Annals of the
American Academy of Political and Social Science 623(1): 195–213.
http://ann.sagepub.com/content/623/1/195.abstract
Peebles, Katherine A. 2012. Negligent Hiring and the Information Age: How State Legislatures
Can Save Employers From Inevitable Liability. William and Mary Law Review 53(4): 1397–
1433.
http://scholarship.law.wm.edu/wmlr/vol53/iss4/7
Phelps, Edmund S. 1972. The Statistical Theory of Racism and Sexism. American Economic
Review 62(4): 659–61.
http://www.jstor.org/stable/1806107
Rasmusen, Eric. 1996. Stigma and Self-Fulfilling Expectations of Criminality. Journal of Law and
Economics 39(2): 519–43.
http://www.jstor.org/stable/725715
Sampson, Robert J. and John H. Laub. 1993. Crime in the Making: Pathways and Turning Points
through Life. Cambridge, MA: Harvard University Press.
http://www.amazon.com/Crime-Making-Pathways-Turning-through/dp/0674176057
25/50
Schnepel, Kevin T. 2012. Labor Market Opportunities and Crime: Evidence from Parolees.
Mimeo, UCSB.
http://www.econ.ucsb.edu/˜schnepel/Kevin_T._Schnepel/RESEARCH_files/Kevin_
Schnepel_JMP.pdf
[SHRM] Society for Human Resource Management. 2010. Background Checking: Conducting
Criminal Background Checks SHRM Poll. Alexandria, VA: Author.
http://www.shrm.org/Research/SurveyFindings/Articles/Pages/
BackgroundCheckCriminalChecks.aspx
———. 2012. Background Checking—The Use of Criminal Background Checks in Hiring
Decisions. Alexandria, VA: Author.
http://www.shrm.org/research/surveyfindings/articles/pages/
criminalbackgroundcheck.aspx
Singer, Simon I. and David McDowall. 1988. Criminalizing Delinquency: The Deterrent Effects
of the New York Juvenile Offender Law. Law and Society Review 22(3): 521–36.
http://www.jstor.org/stable/3053628
Steiner, Benjamin, Craig Hemmens, and Valerie Bell. 2006. Legislative Waiver Reconsidered:
General Deterrent Effects of Statutory Exclusion Laws Enacted Post-1979. Justice Quarterly
23(1): 34–59.
http://dx.doi.org/10.1080/07418820600552444
Torbet, Patricia, Richard Gable, Hunter Hurst IV, Imogene Montgomery, Linda Szymanski, and
Douglas Thomas. 1996. State Responses to Serious and Violent Juvenile Crime. Office of
Juvenile Justice and Delinquency Prevention NCJ 161565.
https://www.ncjrs.gov/pdffiles/statresp.pdf
Uggen, Christopher. 2000. Work as a Turning Point in the Life Course of Criminals: A Duration
Model of Age, Employment, and Recidivism. American Sociological Review 65(4): 529–46.
http://www.jstor.org/stable/2657381
Western, Bruce and Becky Pettit. 2000. Incarceration and Racial Inequality in Men’s Employment.
Industrial and Labor Relations Review 54(1): pp. 3–16.
http://www.jstor.org/stable/2696029
26/50
Figure 1: How criminal history data become accessible to employers and the identifying
information used in the paper
Behavior, procedure, or policy
Identifying variation
criminal
offense
AFQT score or individual fixed effects
is
offender
juvenile?
n
y
juvenile
transfer?
n
state variation in juvenile transfer laws
y
juvenile
court
accessible
juvenile
records?
state variation in age of criminal majority
adult
court
state variation in juvenile record openness
y
n
accessible
criminal
records?
n
y
sex
crime?
n
non-public
criminal
history
state variation in criminal record openness
state variation in sex offender record openness
y
public
criminal
history
27/50
1.0
Figure 2: Surveys of employer use of criminal background checks during hiring,
1990–2013
0.0
0.2
Proportion of surveyed employers
0.4
0.6
0.8
Always conduct criminal
background checks
Never conduct criminal
background checks
Los Angeles samples
SHRM samples
1990
1992
1994
1996
1998
2000
2002
Year
2004
2006
2008
2010
2012
Notes: The first two surveys of Los Angeles employers are reported in Holzer et al. (2007). The
last three surveys were conducted by the Society for Human Resource Management (Burke, 2005;
[SHRM] Society for Human Resource Management, 2010, 2012). These are national samples but
probably oversample large employers, which reflects the membership of SHRM.
28/50
Percent of juveniles taken into police custody
20
40
60
Figure 3: Percent distribution of juveniles taken into police custody by method of
disposition, US, 1990–2011
0
Referred to juvenile court
Handled within police department
and released
Referred to criminal or adult court
Referred to other police agency
Referred to welfare agency
1990
1995
2000
Year
2005
2010
Notes: Source data is from Maguire (2013). Data accessed online at http://www.albany.edu/
sourcebook/pdf/t4262011.pdf on March 1, 2013.
29/50
6
Figure 4: Kernel density estimates of the predicted probability of prior conviction,
NLSY97, ages 18–31, 1997–2011
0
Kernel density estimate
2
4
Black male
Hispanic male
Other male
.1
.2
.3
.4
.5
Predicted probability of any prior conviction
20
0
.6
.7
0
5
Kernel density estimate
10
15
Black female
Hispanic female
Other female
0
.1
.2
.3
.4
.5
Predicted probability of any prior conviction
.6
.7
Notes: Predicted probabilities of prior conviction come from estimated probit models of prior
conviction described in Equation 1. Full results can be found in Table A-1.
30/50
Figure 5: Estimated net effects of criminal history information on labor market outcomes as
a function of perceived criminality (with 95% confidence intervals), NLSY97, ages 18–31,
1997–2011
Panel B: Net information effect on log wages
−.05
−.04
Net information effect on log wages
0
.05
.1
Net information effect on employment probability
−.02
0
.02
.04
.06
.15
Panel A: Net information effect on employment
probability
0
.1
.2
.3
.4
.5
.6
Predicted probability of any prior conviction
.7
0
.1
.2
.3
.4
.5
.6
Predicted probability of any prior conviction
.7
−.1
0
Net information effect on IHS earnings
.1
.2
.3
.4
Panel C: Net information effect on IHS earnings
0
.1
.2
.3
.4
.5
.6
Predicted probability of any prior conviction
.7
Notes: Each point on the solid lines represents the linear combination of the estimated parameters in Table 4
(Columns 1, 3, and 5): β1 + Cβ4 − β3 − Cβ6 + β2 + Cβ5 . The dotted lines are 95% confidence intervals.
31/50
Figure 6: Regressions of employment status on criminal justice variables, March CPS, ages 18–55,
1997–2011
White males
Raw differences
Information estimate (binary)
Information estimate (open 5 years)
Raw differences (young, low−ed sample)
Information estimate (binary; young, low−ed sample)
Information estimate (open 5 years; young, low−ed sample)
95% confidence intervals
Black males
Hispanic males
White females
Black females
Hispanic females
−0.25
−0.20
−0.15
−0.10
−0.05
0.00
Relative difference in employment probabilities
0.05
0.10
Notes: Figure shows results from two samples: the main sample of individuals aged 18–55 years and a young, loweducation sample of individuals aged 18–35 years with at most high-school completion. All models include standard
labor market controls (discretized highest grade completed, a quadratic of potential experience, a central city indicator,
a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed
effects) and the full set of interactions of those controls with the demographic categories listed in the table. Confidence
intervals account for errors with arbitrary within-state dependence.
32/50
Figure 7: Regressions of log wages on criminal justice variables, March CPS, ages 18–55, 1997–
2011
White males
Raw differences
Information estimate (binary)
Information estimate (open 5 years)
Raw differences (young, low−ed sample)
Information estimate (binary; young, low−ed sample)
Information estimate (open 5 years; young, low−ed sample)
95% confidence intervals
Black males
Hispanic males
White females
Black females
Hispanic females
−0.50
−0.40
−0.30
−0.20
−0.10
0.00
Relative difference in log wages
0.10
0.20
Notes: Figure shows results from two samples: the main sample of individuals aged 18–55 years and a young, loweducation sample of individuals aged 18–35 years with at most high-school completion. All models include standard
labor market controls (discretized highest grade completed, a quadratic of potential experience, a central city indicator,
a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed
effects) and the full set of interactions of those controls with the demographic categories listed in the table. Confidence
intervals account for errors with arbitrary within-state dependence.
33/50
Figure 8: Regressions of IHS of income on criminal justice variables, March CPS, ages 18–55,
1997–2011
White males
Raw differences
Information estimate (binary)
Information estimate (open 5 years)
Raw differences (young, low−ed sample)
Information estimate (binary; young, low−ed sample)
Information estimate (open 5 years; young, low−ed sample)
95% confidence intervals
Black males
Hispanic males
White females
Black females
Hispanic females
−3.00
−2.50
−2.00
−1.50
−1.00
−0.50
Relative difference in IHS income
0.00
0.50
Notes: Figure shows results from two samples: the main sample of individuals aged 18–55 years and a young, loweducation sample of individuals aged 18–35 years with at most high-school completion. All models include standard
labor market controls (discretized highest grade completed, a quadratic of potential experience, a central city indicator,
a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed
effects) and the full set of interactions of those controls with the demographic categories listed in the table. Confidence
intervals account for errors with arbitrary within-state dependence.
34/50
Figure 9: Regressions of employment with job benefits on criminal justice variables, March CPS,
ages 18–55, 1997–2011
White males
Raw differences
Information estimate (binary)
Information estimate (open 5 years)
Raw differences (young, low−ed sample)
Information estimate (binary; young, low−ed sample)
Information estimate (open 5 years; young, low−ed sample)
95% confidence intervals
Black males
Hispanic males
White females
Black females
Hispanic females
−0.30
−0.20
−0.10
0.00
0.10
Relative difference in likelihood of receiving job benefits
0.20
Notes: Figure shows results from two samples: the main sample of individuals aged 18–55 years and a young, loweducation sample of individuals aged 18–35 years with at most high-school completion. All models include standard
labor market controls (discretized highest grade completed, a quadratic of potential experience, a central city indicator,
a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed
effects) and the full set of interactions of those controls with the demographic categories listed in the table. Confidence
intervals account for errors with arbitrary within-state dependence.
35/50
Figure 10: Regressions of employment by a firm with more than 500 employees on criminal
justice variables, March CPS, ages 18–55, 1997–2011
White males
Raw differences
Information estimate (binary)
Information estimate (open 5 years)
Raw differences (young, low−ed sample)
Information estimate (binary; young, low−ed sample)
Information estimate (open 5 years; young, low−ed sample)
95% confidence intervals
Black males
Hispanic males
White females
Black females
Hispanic females
−0.15
−0.10
−0.05
0.00
0.05
0.10
0.15
Relative difference in likelihood of working for large firm
0.20
Notes: Figure shows results from two samples: the main sample of individuals aged 18–55 years and a young, loweducation sample of individuals aged 18–35 years with at most high-school completion. All models include standard
labor market controls (discretized highest grade completed, a quadratic of potential experience, a central city indicator,
a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed
effects) and the full set of interactions of those controls with the demographic categories listed in the table. Confidence
intervals account for errors with arbitrary within-state dependence.
36/50
Figure 11: Regressions of mean education for occupation on criminal justice variables, March
CPS, ages 18–55, 1997–2011
White males
Raw differences
Information estimate (binary)
Information estimate (open 5 years)
Raw differences (young, low−ed sample)
Information estimate (binary; young, low−ed sample)
Information estimate (open 5 years; young, low−ed sample)
95% confidence intervals
Black males
Hispanic males
White females
Black females
Hispanic females
−1.20
−1.00
−0.80
−0.60
−0.40
−0.20
0.00
0.20
Relative difference in mean highest grade completed for occupation
Notes: Figure shows results from two samples: the main sample of individuals aged 18–55 years and a young, loweducation sample of individuals aged 18–35 years with at most high-school completion. All models include standard
labor market controls (discretized highest grade completed, a quadratic of potential experience, a central city indicator,
a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed
effects) and the full set of interactions of those controls with the demographic categories listed in the table. Confidence
intervals account for errors with arbitrary within-state dependence.
37/50
Figure 12: Regressions of school enrollment on criminal justice variables, March CPS, ages 18–
55, 1997–2011
White males
Raw differences
Information estimate (binary)
Information estimate (open 5 years)
Raw differences (young, low−ed sample)
Information estimate (binary; young, low−ed sample)
Information estimate (open 5 years; young, low−ed sample)
95% confidence intervals
Black males
Hispanic males
White females
Black females
Hispanic females
−0.10
−0.08
−0.06
−0.04
−0.02
0.00
0.02
0.04
Relative difference in school enrollment probabilities
0.06
Notes: Figure shows results from two samples: the main sample of individuals aged 18–55 years and a young, loweducation sample of individuals aged 18–35 years with at most high-school completion. All models include standard
labor market controls (discretized highest grade completed, a quadratic of potential experience, a central city indicator,
a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed
effects) and the full set of interactions of those controls with the demographic categories listed in the table. Confidence
intervals account for errors with arbitrary within-state dependence.
38/50
Table 1: The provision of criminal history data online to the public, by state
N
C
J
S
N
C
J State
S
2000
N
C
J
2005
2000 S
N
2005
2000 C
J
S
Alabama
2010
NICS, Corrections, Judicial, and Sex
offender repository timelines
(
free
not free
no histories)
2010
2005
No free general
criminal history
search (2014)
Juvenile records
included in adult
databases
2010
×
2000
2005
2010
2000
2005
2010
2000
2005
2010
2000
2005
2010
2000
2005
2010
×
2000
2005
2010
×
×
2000
2005
2010
2000
2005
2010
×
×
2000
2005
2010
2000
2005
2010
×
2000
2005
2010
×
2000
2005
2010
2000
2005
2010
2000
2005
2010
2000
2005
2010
2000
2005
2010
2000
2005
2010
×
2000
2005
2010
×
N
C
J
S
Alaska
N
C
J
S
Arizona
N
C
J
S
Arkansas
N
C
J
S
California
N
C
J
S
Colorado
N
C
J
S
Connecticut
N
C
J
S
Delaware
N
C
J
S
District of Columbia
N
C
J
S
Florida
N
C
J
S
Georgia
N
C
J
S
Hawaii
N
C
J
S
Idaho
N
C
J
S
Illinois
×
×
N
C
J
S
Indiana
N
C
J
S
Iowa
N
C
J
S
Kansas
N
C
J
S
Kentucky
Notes: Dates and database content come from the Internet Archive’s Wayback Machine (http:
//web.archive.org). Juvenile records availability comes from http://www.beforeyouplea.
com and tests of online repositories with recent birth dates (as of June 2013). Juvenile
records availability indicates whether juvenile criminal records are compiled in adult criminal
databases. This variable in conjunction with the online access variable indicates whether
juvenile records are broadly available to the public online. Sex crime registry data come from
Agan (2011).
39/50
Table 2: The means by which juveniles get to adult court
Category
Discretionary
Mandatory
Other rules
Number of Number of Median age Median age
states with states with
for rule
for rule
rule in 1997 rule in 2011
in 1997
in 2011
o
31
32
14
14
Procedure
Judicial waiver
Direct file
Murder
Robbery
Assault
Burglary
Theft
Major property
Drug possession
Drug trafficking
Statutory exclusion o
Mandatory waiver
Murder
Robbery
Assault
Burglary
Theft
Major property
Drug possession
Drug trafficking
Once an adult,
always an adult
Presumptive waiver
Blended sentencing
Criminal majority Criminal majority
29
27
17
25
18
23
18
25
30
30
21
29
22
27
22
30
14
14
14
14
14
14
14.5
14
14
14
14
14
14
14
14.5
14
27
33
15
15
19
5
1
1
1
0
2
2
21
32
10
2
1
1
3
2
4
34
14.5
14
14
17
14
16.5
16
16
—
15
16
15
17
14
16
16
15.5
—
—
—
—
—
—
—
—
—
States with
age 16
in 1997
3
States with
age 16
in 2011
3
States with
age 17
in 1997
8
States with
age 17
in 2011
10
Notes: Source is author’s calculations from statutes shown in Table A-3. Juvenile sex crimes are excluded
from these data because state laws are difficult to characterize in a consistent way. Moreover, juvenile sex
crime data is broadly available to the public because of “Megan’s Laws” databases.
40/50
41/50
Ever incarcerated
Age 18 Age 19 Age 20 Age 21
0.02
0.03
0.04
0.05
(0.15) (0.17) (0.20) (0.21)
Ever convicted
0.08
0.11
0.13
0.14
(0.28) (0.31) (0.33) (0.34)
Employed
0.55
0.62
0.66
0.69
(0.50) (0.49) (0.47) (0.46)
Hourly wage in primary job
7.65
8.50
9.27 10.16
(5.72) (5.88) (6.10) (7.01)
Annual earnings ($1000)
2.31
4.01
6.31
8.22
(3.72) (5.46) (7.51) (9.19)
Experience since age 14 (1000s hours)
0.60
0.48
3.27
4.46
(1.27) (1.37) (2.74) (3.40)
Tenure in primary job (weeks)
20.61 30.38 45.27 56.42
(39.72) (46.52) (56.42) (67.10)
Earned GED
0.01
0.03
0.04
0.06
(0.12) (0.17) (0.20) (0.23)
Earned HS diploma
0.44
0.71
0.75
0.75
(0.50) (0.45) (0.43) (0.43)
Earned AA degree
0.00
0.00
0.01
0.02
(0.01) (0.01) (0.08) (0.15)
Earned BA degree or more
0.00
0.01
0.01
0.02
(0.07) (0.07) (0.10) (0.12)
Body mass index
24.04 24.49 24.88 25.33
(4.66) (4.81) (4.89) (5.10)
BMI missing
0.04
0.05
0.06
0.07
(0.19) (0.21) (0.24) (0.25)
Lives in central city
0.31
0.31
0.35
0.38
(0.46) (0.46) (0.48) (0.48)
ASVAB
37.00 36.91 37.26 37.22
(31.85) (31.86) (32.12) (31.96)
ASVAB missing
0.19
0.19
0.20
0.19
(0.39) (0.40) (0.40) (0.39)
Female
0.50
0.49
0.50
0.50
(0.50) (0.50) (0.50) (0.50)
Black
0.26
0.26
0.26
0.26
(0.44) (0.44) (0.44) (0.44)
Hispanic
0.21
0.21
0.21
0.21
(0.41) (0.41) (0.41) (0.41)
Notes:
Age 22
0.06
(0.24)
0.15
(0.36)
0.73
(0.44)
11.32
(8.02)
9.87
(10.84)
5.55
(4.11)
63.29
(74.98)
0.07
(0.25)
0.68
(0.47)
0.04
(0.19)
0.07
(0.26)
25.75
(5.26)
0.07
(0.26)
0.39
(0.49)
37.26
(32.05)
0.19
(0.39)
0.50
(0.50)
0.26
(0.44)
0.21
(0.41)
Age 23
0.07
(0.25)
0.17
(0.37)
0.76
(0.43)
12.74
(9.06)
11.91
(12.90)
6.74
(4.92)
70.92
(83.33)
0.08
(0.28)
0.59
(0.49)
0.05
(0.21)
0.15
(0.36)
26.14
(5.31)
0.07
(0.25)
0.42
(0.49)
36.93
(31.92)
0.19
(0.39)
0.50
(0.50)
0.27
(0.44)
0.21
(0.41)
Age 24 Age 25 Age 26 Age 27 Age 28 Age 29
0.08
0.08
0.09
0.09
0.10
0.11
(0.27) (0.27) (0.28) (0.29) (0.30) (0.31)
0.18
0.19
0.20
0.20
0.20
0.22
(0.38) (0.39) (0.40) (0.40) (0.40) (0.41)
0.78
0.77
0.77
0.77
0.76
0.75
(0.42) (0.42) (0.42) (0.42) (0.43) (0.43)
13.98 14.91 15.66 16.59 17.27 17.69
(9.42) (9.81) (9.88) (10.65) (11.37) (11.39)
14.70 17.81 19.76 21.90 23.95 25.33
(15.88) (18.16) (20.45) (22.36) (24.24) (26.30)
8.02
9.20 10.27 11.41 12.50 13.24
(5.85) (6.75) (7.68) (8.63) (9.69) (10.35)
80.31 91.34 103.70 115.42 130.62 142.34
(91.74) (101.94) (113.50) (125.73) (138.45) (150.46)
0.10
0.11
0.11
0.12
0.12
0.13
(0.30) (0.31) (0.32) (0.32) (0.33) (0.34)
0.53
0.50
0.47
0.46
0.45
0.45
(0.50) (0.50) (0.50) (0.50) (0.50) (0.50)
0.05
0.06
0.06
0.06
0.07
0.07
(0.23) (0.24) (0.24) (0.24) (0.25) (0.26)
0.20
0.22
0.24
0.24
0.26
0.25
(0.40) (0.41) (0.43) (0.43) (0.44) (0.43)
26.51 26.71 27.03 27.25 27.63 27.61
(5.51) (5.53) (5.61) (5.65) (5.72) (5.65)
0.05
0.06
0.06
0.06
0.07
0.06
(0.23) (0.24) (0.23) (0.24) (0.25) (0.24)
0.43
0.43
0.43
0.42
0.42
0.41
(0.50) (0.49) (0.49) (0.49) (0.49) (0.49)
36.80 36.73 36.78 36.26 36.08 35.79
(31.94) (31.89) (31.92) (31.84) (31.76) (31.60)
0.19
0.19
0.19
0.20
0.20
0.20
(0.40) (0.39) (0.39) (0.40) (0.40) (0.40)
0.50
0.50
0.50
0.50
0.50
0.50
(0.50) (0.50) (0.50) (0.50) (0.50) (0.50)
0.27
0.27
0.27
0.27
0.27
0.28
(0.45) (0.44) (0.45) (0.45) (0.45) (0.45)
0.21
0.21
0.21
0.21
0.21
0.21
(0.41) (0.41) (0.41) (0.41) (0.41) (0.41)
Table 3: Selected descriptive statistics by age, NLSY97, ages 18–30, 1997–2011
Age 30
0.11
(0.32)
0.21
(0.41)
0.77
(0.42)
19.03
(12.77)
26.33
(28.65)
15.08
(11.15)
159.27
(167.23)
0.12
(0.33)
0.47
(0.50)
0.07
(0.25)
0.25
(0.43)
27.95
(5.61)
0.06
(0.24)
0.43
(0.50)
34.58
(31.58)
0.21
(0.41)
0.51
(0.50)
0.29
(0.45)
0.21
(0.40)
Table 4: Regressions of labor market outcomes on criminal justice variables, NLSY97, ages 18–31,
1997–2011
Employment
Log wage
IHS earnings
(1)
(2)
(3)
(4)
(5)
(6)
Has no record, open records (β1 )
0.006
0.000 −0.043*** 0.012 −0.001
−0.044
(0.005) (0.009) (0.007) (0.013) (0.017)
(0.031)
Has record, closed records (β2 )
0.006
0.000
0.005 −0.014
0.109*** 0.123
(0.012) (0.022) (0.018) (0.029) (0.039)
(0.075)
Has record, open records (β3 )
−0.025***−0.045* −0.036***−0.069*
0.019
−0.008
(0.009) (0.023) (0.013) (0.039) (0.030)
(0.079)
Perceived criminality
0.397*** 0.158** 0.008 −0.160
0.725*** −0.096
(0.035) (0.072) (0.057) (0.102) (0.123)
(0.246)
Has no record, open records×perceived criminality (β4 )
0.056** 0.047
0.099** −0.015
0.179*
0.353**
(0.026) (0.045) (0.040) (0.068) (0.095)
(0.153)
Has record, closed records×perceived criminality (β5 )
−0.216***−0.144** −0.073 −0.113 −0.717*** −0.836***
(0.041) (0.067) (0.066) (0.099) (0.142)
(0.227)
Has record, open records×perceived criminality (β6 )
−0.127***−0.018 −0.046 −0.016 −0.627*** −0.818***
(0.031) (0.062) (0.049) (0.091) (0.111)
(0.204)
Year fixed effects
×
×
×
×
×
×
Individual fixed effects
×
×
×
Information effect (with record, C = .35)
−0.000 −0.002 −0.031** −0.022 −0.059* −0.124*
H0 : β3 + .35β6 − β2 − .35β5 = 0
(0.009) (0.021) (0.015) (0.031) (0.033) (0.067)
Information effect (no record, C = .35)
0.026*** 0.017 −0.008
0.007
0.062*** 0.079*
H0 : β1 + .35β4 = 0
(0.006) (0.012) (0.010) (0.018) (0.024) (0.041)
Net information effect (C = .35)
0.026** 0.018
0.023
0.029
0.121*** 0.203**
H0 : β1 + .35β4 − β3 − .35β6 + β2 + .35β5 = 0
(0.012) (0.024) (0.019) (0.037) (0.043) (0.079)
R2
0.12
0.05
0.26
0.28
0.28
0.21
N
82,652
82,652
65,147
65,147
82,652
82,652
Mean of dep. var.
0.733
0.733
2.382
2.382
2.289
2.289
S.D. of dep. var.
0.442
0.442
0.606
0.606
1.672
1.672
Notes: Models in odd-numbered columns are estimated as pooled cross-sectional samples, while models in the even-numbered
columns use the panel with individual fixed effects. All models include the following time-varying regressors: highest grade
completed and its square; total work experience since age 13 and its square; age and its square; body-mass index; and
dummies for completion of a GED, high school diploma, AA degree, or a bachelors degree or more; and dummies for centralcity residence and missing body-mass index. Pooled cross-sectional models include the following time-invariant regressors:
ASVAB score and dummies for missing ASVAB, female sex, black race, and Hispanic ethnicity. Wages models also include
tenure in weeks in the primary job. Stars indicate statistical significance: * p < 0.05, ** p < 0.01, *** p < 0.001. Full results
can be found in Table A-2.
42/50
43/50
0.0041
(0.0122)
0.0042
(0.0081)
0.0048
(0.0048)
0.0070
(0.0090)
0.0112
(0.0082)
−0.0041*** −0.0029
(0.0014)
(0.0254)
0.0018
0.0273
(0.0015)
(0.0238)
−0.0015*
0.0078
(0.0009)
(0.0111)
−0.0012
0.0521**
(0.0017)
(0.0209)
−0.0001
−0.0165
(0.0012)
(0.0197)
−0.0033
(0.0041)
0.0036
(0.0042)
0.0014
(0.0015)
−0.0033
(0.0039)
−0.0012
(0.0036)
Employment
Log wage
(1)
(2)
(3)
(4)
−0.0039
−0.0073
(0.0036)
(0.0076)
0.0039***
0.0010
(0.0013)
(0.0021)
−0.0014
(0.1481)
0.0970
(0.0947)
0.0608
(0.0537)
0.1051
(0.0955)
0.0805
(0.1093)
−0.0196**
(0.0095)
0.0141
(0.0188)
−0.0155*
(0.0090)
−0.0059
(0.0175)
−0.0023
(0.0121)
IHS earnings
(5)
(6)
−0.0110
(0.0348)
0.0496***
(0.0129)
Raw differences from white males in closed states
Black males
−0.15
−0.15
−0.26
−0.26
−1.44
−1.44
Hispanic males
−0.03
−0.03
−0.34
−0.34
−0.48
−0.48
White females
−0.10
−0.10
−0.23
−0.23
−1.28
−1.28
Black females
−0.15
−0.15
−0.37
−0.37
−1.56
−1.56
Hispanic females
−0.23
−0.23
−0.45
−0.45
−2.73
−2.73
Specification
R2
0.14
0.14
0.44
0.44
0.14
0.14
N
1,358,458
1,358,458
186,867
186,867
1,361,836
1,361,836
Mean of dep. var.
0.76
0.76
2.72
2.72
8.40
8.40
Std. dev. of dep. var.
0.43
0.43
0.63
0.63
4.50
4.50
Notes: All models include standard labor market controls (discretized highest grade completed, a quadratic of potential experience, a
central city indicator, a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed
effects) and the full set of interactions of those controls with the demographic categories listed in the table. Standard errors in parentheses
account for errors with arbitrary within-state dependence. Stars indicate statistical significance: * p < 0.05, ** p < 0.01, *** p < 0.001.
Hispanic females
Black females
White females
Hispanic males
Interaction with open records variable above
Black males
Years since records open
Open records
Covariates
Table 5: Regressions of labor market outcomes on criminal justice variables, March CPS, ages 18–55, 1997–2011
44/50
−0.0303
(0.0191)
0.0107
(0.0123)
0.0012
(0.0088)
0.0191
(0.0162)
0.0146
(0.0105)
−0.0087**−0.0098
0.0048
(0.0036) (0.0422) (0.0098)
0.0007 −0.0185
0.0015
(0.0018) (0.0319) (0.0053)
−0.0019 −0.0290
0.0008
(0.0021) (0.0263) (0.0061)
−0.0025
0.0563 −0.0076
(0.0030) (0.0488) (0.0062)
0.0015 −0.1264***
−0.0120*
(0.0018) (0.0440) (0.0064)
Employment
Log wage
(1)
(2)
(3)
(4)
−0.0051
0.0329**
(0.0067)
(0.0162)
0.0032*
0.0002
(0.0018)
(0.0034)
−0.2255
(0.1911)
0.1648
(0.1020)
−0.0053
(0.1009)
0.2071
(0.1945)
0.1270
(0.1542)
−0.0632**
(0.0288)
−0.0099
(0.0212)
−0.0153
(0.0199)
−0.0189
(0.0315)
−0.0205
(0.0182)
IHS earnings
(5)
(6)
0.0110
(0.0666)
0.0637***
(0.0182)
Raw differences from white males in closed states
Black males
−0.19
−0.19
−0.09
−0.09
−1.78
−1.78
Hispanic males
0.05
0.05
−0.08
−0.08
0.18
0.18
White females
−0.12
−0.12
−0.24
−0.24
−1.49
−1.49
Black females
−0.21
−0.21
−0.22
−0.22
−2.10
−2.10
Hispanic females
−0.24
−0.24
−0.19
−0.19
−2.82
−2.82
Specification
R2
0.19
0.19
0.36
0.36
0.17
0.17
N
283,926
283,926
32,607
32,607
285,036
285,036
Mean of dep. var.
0.66
0.66
2.34
2.34
7.38
7.38
Std. dev. of dep. var.
0.47
0.47
0.50
0.50
4.58
4.58
Notes: All models include standard labor market controls (discretized highest grade completed, a quadratic of potential
experience, a central city indicator, a union indicator, the effective state minimum wage, the state unemployment rate, year
fixed effects, and state fixed effects) and the full set of interactions of those controls with the demographic categories listed
in the table. Standard errors in parentheses account for errors with arbitrary within-state dependence. Stars indicate statistical
significance: * p < 0.05, ** p < 0.01, *** p < 0.001.
Hispanic females
Black females
White females
Hispanic males
Interaction with open records variable above
Black males
Years since records open
Open records
Covariates
Table 6: Regressions of labor market outcomes on criminal justice variables, March CPS, ages 18–35,
individuals with at most high school completion, 1997–2011
45/50
0.0187*
(0.0093)
0.0027
(0.0141)
−0.0017
(0.0037)
0.0135*
(0.0076)
0.0014
(0.0139)
−0.0005
(0.0021)
−0.0010
(0.0009)
0.0026***
(0.0006)
0.0008
(0.0010)
−0.0010
(0.0016)
−0.0022
(0.0104)
0.0113
(0.0126)
0.0029
(0.0052)
0.0025
(0.0087)
−0.0019
(0.0111)
−0.0013
(0.0015)
−0.0022
(0.0014)
0.0006
(0.0008)
0.0006
(0.0018)
−0.0018
(0.0014)
Benefits
Large firm
(1)
(2)
(3)
(4)
0.0003
−0.0015
(0.0042)
(0.0045)
−0.0033***
−0.0014
(0.0006)
(0.0009)
0.0145
(0.0340)
0.0012
(0.0178)
0.0145
(0.0144)
−0.0028
(0.0250)
−0.0468**
(0.0188)
−0.0077**
(0.0033)
−0.0032
(0.0034)
−0.0016
(0.0029)
−0.0072*
(0.0042)
−0.0053
(0.0043)
0.0067
(0.0052)
−0.0018
(0.0042)
0.0010
(0.0021)
0.0002
(0.0040)
0.0062
(0.0046)
0.0004
(0.0012)
−0.0003
(0.0022)
−0.0010**
(0.0004)
−0.0003
(0.0013)
−0.0004
(0.0022)
Occup. skill
Enrolled
(5)
(6)
(7)
(8)
−0.0050
−0.0020
(0.0111)
(0.0021)
0.0083***
−0.0083***
(0.0025)
(0.0016)
Raw differences from white males in closed states
Black males
−0.05
−0.05
0.11
0.11
−0.66
−0.66
0.00
0.00
Hispanic males
−0.27
−0.27
−0.13
−0.13
−1.12
−1.12
−0.01
−0.01
White females
0.02
0.02
0.03
0.03
0.08
0.08
0.00
0.00
Black females
−0.05
−0.05
0.16
0.16
−0.33
−0.33
0.00
0.00
Hispanic females
−0.15
−0.15
−0.02
−0.02
−0.84
−0.84
0.00
0.00
Specification
R2
0.10
0.10
0.04
0.04
0.40
0.40
0.36
0.37
N
1,189,996
1,189,996
1,088,056
1,088,056
1,361,836
1,361,836
1,361,836
1,361,836
Mean of dep. var.
0.80
0.80
0.47
0.47
13.04
13.04
0.08
0.08
Std. dev. of dep. var.
0.40
0.40
0.50
0.50
1.60
1.60
0.27
0.27
Notes: Benefits is an indicator for those provided with health insurance or pension benefits by employers. Large firm is an indicator for workers at firms with at least 500
employees. Occupational skill is the mean of highest grade completed for the current occupation—from a 1994–1996 CPS sample. Enrolled is an indicator for school
enrollment. All models include standard labor market controls (discretized highest grade completed, a quadratic of potential experience, a central city indicator, a union
indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state fixed effects) and the full set of interactions of those controls
with the demographic categories listed in the table. Standard errors in parentheses account for errors with arbitrary within-state dependence. Stars indicate statistical
significance: * p < 0.05, ** p < 0.01, *** p < 0.001.
Hispanic females
Black females
White females
Hispanic males
Interaction with open records variable above
Black males
Years since records open
Open records
Covariates
Table 7: Regressions of labor market outcomes on criminal justice variables, March CPS, ages 18–55, 1997–2011
46/50
Occup. skill
Enrolled
(5)
(6)
(7)
(8)
0.0193
−0.0003
(0.0177)
(0.0041)
0.0057*
−0.0052***
(0.0034)
(0.0017)
0.0672*** 0.0034
0.0033
0.0040
0.0173 −0.0043
0.0105
0.0002
(0.0197)
(0.0039) (0.0187) (0.0027) (0.0322) (0.0052) (0.0119) (0.0023)
0.0032
0.0025
0.0018 −0.0003 −0.0079 −0.0020 −0.0039
0.0008
(0.0170)
(0.0016) (0.0153) (0.0024) (0.0280) (0.0042) (0.0074) (0.0015)
0.0044
0.0019
0.0040
0.0038 −0.0024 −0.0006 −0.0059 −0.0016*
(0.0091)
(0.0026) (0.0105) (0.0023) (0.0179) (0.0043) (0.0050) (0.0009)
0.0058
0.0012 −0.0156 −0.0012
0.0051
0.0001
0.0003 −0.0008
(0.0245)
(0.0030) (0.0206) (0.0031) (0.0440) (0.0063) (0.0075) (0.0014)
0.0161
−0.0024 −0.0046 −0.0025 −0.0162
0.0031
0.0008
0.0002
(0.0168)
(0.0023) (0.0154) (0.0024) (0.0276) (0.0048) (0.0074) (0.0021)
Benefits
Large firm
(1)
(2)
(3)
(4)
−0.0032
0.0116
(0.0081)
(0.0080)
−0.0069***
−0.0014
(0.0017)
(0.0017)
Raw differences from white males in closed states
Black males
−0.04
−0.04
0.14
0.14
−0.12
−0.12
−0.02
−0.02
Hispanic males
−0.28
−0.28
−0.14
−0.14
−0.44
−0.44
−0.07
−0.07
White females
0.02
0.02
0.07
0.07
0.24
0.24
0.02
0.02
Black females
−0.11
−0.11
0.17
0.17
0.08
0.08
−0.01
−0.01
Hispanic females
−0.14
−0.14
−0.01
−0.01
−0.14
−0.14
−0.06
−0.06
Specification
R2
0.08
0.08
0.05
0.05
0.11
0.11
0.38
0.38
N
221,877
221,877
200,983
200,983
285,036
285,036
285,036
285,036
Mean of dep. var.
0.64
0.64
0.41
0.41
12.17
12.17
0.13
0.13
Std. dev. of dep. var.
0.48
0.48
0.49
0.49
0.94
0.94
0.33
0.33
Notes: Benefits is an indicator for those provided with health insurance or pension benefits by employers. Large firm is an indicator for workers at firms
with at least 500 employees. Occupational skill is the mean of highest grade completed for the current occupation—from a 1994–1996 CPS sample.
Enrolled is an indicator for school enrollment. All models include standard labor market controls (discretized highest grade completed, a quadratic of
potential experience, a central city indicator, a union indicator, the effective state minimum wage, the state unemployment rate, year fixed effects, and state
fixed effects) and the full set of interactions of those controls with the demographic categories listed in the table. Standard errors in parentheses account
for errors with arbitrary within-state dependence. Stars indicate statistical significance: * p < 0.05, ** p < 0.01, *** p < 0.001.
Hispanic females
Black females
White females
Hispanic males
Interaction with open records variable above
Black males
Years since records open
Open records
Covariates
Table 8: Regressions of labor market outcomes on criminal justice variables, March CPS, ages 18–35, individuals with at most
high school completion, 1997–2011
Table A-1: Probit regressions of prior conviction on variables observation to employers, NLSY97,
ages 18–31, 1997–2011
Men
Women
(1) Black (2) Hispanic (3) Other (4) Black (5) Hispanic (6) Other
0.134**
0.274*** −0.201*** 0.343*** −0.002
0.025
(0.068)
(0.103)
(0.048)
(0.102)
(0.117)
(0.059)
HGC squared
−0.011*** −0.019***
0.003
−0.016*** −0.004
−0.005**
(0.003)
(0.004)
(0.002)
(0.004)
(0.005)
(0.002)
Earned GED
0.298*** −0.027
0.233*** −0.124*
0.520***
0.022
(0.051)
(0.065)
(0.043)
(0.071)
(0.091)
(0.053)
Earned HS diploma
−0.408*** −0.385*** −0.480*** −0.784***
0.064
−0.543***
(0.050)
(0.054)
(0.040)
(0.072)
(0.089)
(0.049)
Earned AA degree
−0.106
−0.428*** −0.612*** −0.675***
0.006
−0.495***
(0.114)
(0.143)
(0.074)
(0.147)
(0.185)
(0.084)
Earned BA degree or more
−0.322*** −0.145
−0.598*** −0.773***
0.013
−0.696***
(0.116)
(0.133)
(0.070)
(0.152)
(0.206)
(0.086)
Experience since age 14 (1000s hours) −0.024*** −0.008
0.012*** −0.012
−0.007
0.025***
(0.006)
(0.007)
(0.004)
(0.009)
(0.010)
(0.006)
Experience squared
0.000
−0.000
−0.001*** 0.001*
−0.000
−0.001***
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
Age
0.658***
0.543***
0.550*** 0.604***
0.264**
0.348***
(0.076)
(0.086)
(0.052)
(0.096)
(0.122)
(0.065)
Age squared
−0.012*** −0.009*** −0.009*** −0.011*** −0.004*
−0.006***
(0.002)
(0.002)
(0.001)
(0.002)
(0.003)
(0.001)
Body mass index
−0.023*** −0.007**
−0.018*** −0.002
0.003
−0.010***
(0.003)
(0.003)
(0.002)
(0.003)
(0.005)
(0.002)
BMI missing
−0.808*** −0.256**
−0.538*** −0.357*** −0.142
−0.273***
(0.101)
(0.116)
(0.084)
(0.117)
(0.151)
(0.085)
Lives in central city
0.063**
−0.109*** −0.004
0.160***
0.090*
0.114***
(0.031)
(0.034)
(0.022)
(0.041)
(0.048)
(0.027)
Constant
−8.639*** −8.387*** −5.601***−10.647*** −4.755*** −5.125***
(1.008)
(1.249)
(0.701)
(1.354)
(1.657)
(0.883)
Psuedo R2
0.14
0.12
0.13
0.12
0.05
0.10
N
10,650
8,802
22,085
11,415
8,720
20,980
Mean of dep. var.
0.271
0.223
0.212
0.075
0.064
0.098
S.D. of dep. var.
0.445
0.416
0.409
0.263
0.244
0.297
Notes: Models are estimated as pooled cross-sections. Stars indicate statistical significance: * p < 0.05, ** p < 0.01, ***
p < 0.001.
Highest grade completed
47/50
Table A-2: Regressions of labor market outcomes on criminal justice variables,
NLSY97, ages 18–31, 1997–2011
Employment
Log wage
IHS earnings
(1)
(2)
(3)
(4)
(5)
(6)
Highest grade completed
0.072***−0.010 −0.009 −0.104*** 0.271*** 0.024
(0.007) (0.015) (0.012) (0.022) (0.027)
(0.050)
HGC squared
−0.002*** 0.001
0.001*** 0.004***−0.009*** 0.000
(0.000) (0.001) (0.000) (0.001) (0.001)
(0.002)
Earned GED
0.015*
0.024
0.045*** 0.041*
0.104***−0.059
(0.008) (0.015) (0.012) (0.022) (0.028)
(0.051)
Earned HS diploma
0.144*** 0.055*** 0.057*** 0.009
0.387*** 0.074**
(0.008) (0.012) (0.010) (0.015) (0.023)
(0.034)
Earned AA degree
0.198*** 0.074*** 0.162*** 0.093*** 0.707*** 0.229***
(0.011) (0.018) (0.016) (0.025) (0.040)
(0.054)
Earned BA degree or more
0.241*** 0.188*** 0.249*** 0.219*** 0.938*** 0.693***
(0.011) (0.017) (0.015) (0.022) (0.037)
(0.049)
Experience since age 14 (1000s hours)
0.021*** 0.009*** 0.008*** 0.006*** 0.101*** 0.081***
(0.001) (0.001) (0.001) (0.002) (0.002)
(0.004)
Experience squared
−0.000***−0.000*** 0.000 −0.000 −0.001***−0.002***
(0.000) (0.000) (0.000) (0.000) (0.000)
(0.000)
Age
−0.030*** 0.055*** 0.067*** 0.137*** 0.480*** 0.641***
(0.012) (0.017) (0.017) (0.023) (0.039)
(0.058)
Age squared
0.000* −0.001** −0.001** −0.002***−0.009***−0.011***
(0.000) (0.000) (0.000) (0.000) (0.001)
(0.001)
Body mass index
0.001***−0.002** −0.003*** 0.001
0.004***−0.000
(0.000) (0.001) (0.000) (0.001) (0.001)
(0.002)
BMI missing
−0.011 −0.064***−0.126*** 0.027 −0.190***−0.129*
(0.012) (0.022) (0.016) (0.030) (0.039)
(0.075)
Lives in central city
0.000
0.028*** 0.006
0.017** −0.002
0.087***
(0.003) (0.005) (0.005) (0.008) (0.012)
(0.018)
ASVAB
0.000***
0.001***
0.003***
(0.000)
(0.000)
(0.000)
ASVAB missing
0.002
0.012
0.054***
(0.005)
(0.008)
(0.018)
Black
−0.064***
−0.072***
−0.341***
(0.004)
(0.006)
(0.014)
Hispanic
0.024***
0.028***
−0.037**
(0.004)
(0.006)
(0.016)
Has no record, open records (β1 )
0.006
0.000 −0.043*** 0.012 −0.001
−0.044
(0.005) (0.009) (0.007) (0.013) (0.017)
(0.031)
Has record, closed records (β2 )
0.006
0.000
0.005 −0.014
0.109*** 0.123
(0.012) (0.022) (0.018) (0.029) (0.039)
(0.075)
Has record, open records (β3 )
−0.025***−0.045* −0.036***−0.069*
0.019
−0.008
(0.009) (0.023) (0.013) (0.039) (0.030)
(0.079)
Perceived criminality
0.397*** 0.158** 0.008 −0.160
0.725***−0.096
(0.035) (0.072) (0.057) (0.102) (0.123)
(0.246)
Has no record, open records×perceived criminality (β4 )
0.056** 0.047
0.099** −0.015
0.179*
0.353**
(0.026) (0.045) (0.040) (0.068) (0.095)
(0.153)
Has record, closed records×perceived criminality (β5 )
−0.216***−0.144** −0.073 −0.113 −0.717***−0.836***
(0.041) (0.067) (0.066) (0.099) (0.142)
(0.227)
Has record, open records×perceived criminality (β6 )
−0.127***−0.018 −0.046 −0.016 −0.627***−0.818***
(0.031) (0.062) (0.049) (0.091) (0.111)
(0.204)
Tenure in primary job (weeks)
0.000*** 0.000***
(0.000) (0.000)
Year fixed effects
×
×
×
×
×
×
Individual fixed effects
×
×
×
Information effect (with record, C = .35)
−0.000 −0.002 −0.031** −0.022 −0.059* −0.124*
H0 : β3 + .35β6 − β2 − .35β5 = 0
(0.009) (0.021) (0.015) (0.031) (0.033) (0.067)
Information effect (no record, C = .35)
0.026*** 0.017 −0.008
0.007
0.062*** 0.079*
H0 : β1 + .35β4 = 0
(0.006) (0.012) (0.010) (0.018) (0.024) (0.041)
Net information effect (C = .35)
0.026** 0.018
0.023
0.029
0.121*** 0.203**
H0 : β1 + .35β4 − β3 − .35β6 + β2 + .35β5 = 0 (0.012) (0.024) (0.019) (0.037) (0.043) (0.079)
R2
0.12
0.05
0.26
0.28
0.28
0.21
N
82,652
82,652
65,147
65,147
82,652
82,652
Mean of dep. var.
0.733
0.733
2.382
2.382
2.289
2.289
S.D. of dep. var.
0.442
0.442
0.606
0.606
1.672
1.672
Notes: Models in odd-numbered columns are estimated as pooled cross-sectional samples, while
models in the even-numbered columns use the panel with individual fixed effects. Stars indicate
statistical significance: * p < 0.05, ** p < 0.01, *** p < 0.001.
48/50
49/50
§ 12-15-102
47.12.020
8-201
§ 9-27-303
Family Code-11-1-6500
19-1-103
46b-120
10-§ 901
II-§ 16-2301
XLVII-985.03
§ 15-11-2
3-31-§ 577-1
20-5-32-101
705-405/5-105
11-8-1-2
VI-232.2; XV-599.1
38-101; 38-2302;
LI-600.020
VIII-1-804
15-501-§3003
§4–202.1
XVII-119-52
712A.2
§ 260B.007
§ 43-21-151
XII-211.021
41-1-101
§ 43-245
62A.030 XII-169-B:2
2A:4A-22
§32A-1-4; §32A-2-3
11-A-U-§720.10
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of Columbia
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
Juvenile Defined
—
III-4-305
—
—
—
712A.2
—
—
—
41-5-206
§ 43-276
—
—
—
—
—
—
—
—
§ 15-11-28
—
—
—
19-2-517
46b-127
—
II-§ 16-2307
XLVII-985.557
§ 9-27-318 Welfare and Institutions2-1-2-17-707.1; Welfare
and Institutions-2-1-2-17707
13-501
—
—
Direct File
—
—
—
19-2-518
46b-127
10-§1010
II-§ 16-2307
—
—
VIII-11-862
15-503-§3101;
§ 3-8A-06
—
712A.2d
§ 260B.125
§ 43-21-157
XII-211.071
—
—
62B.390
XII-169-B:24
2A:4A-26 —
38-2347;
LI-640.01; LI-635.020
VI-232.45
—
VIII-11-857
—
—
—
—
§ 260B.125
—
—
—
—
62B.330
XII-169-B:24
2A:4A-26 —
—
LI-635.02
—
§ 15-11-30.2
§ 15-11-30.3
3-31-§571-22
—
20-5-20-508
—
705-405/5-805; 705705-405/5-805
405/5‑810
31-30-3-2; 31-30-3-3; 31-30-31-30-3-6
3-4; 31-30-3-5
19-2-518
—
10-§1010
II-§ 16-2307
XLVII-985.556
§ 9-27-318 —
Welfare and Institutions- —
2-1-2-17-707.1; Welfare
and Institutions-2-1-2-17707
8-327
47.12.100; Alaska Court
Rules, Part VI, Rule 20
§ 12-15-203
Judicial Waiver or Light Mandatory Waiver or
Presumptive
Tough Presumptive
—
—
Once an Adult Always
an Adult
11-A-H-§180.75
III-4-305
—
§ 3-8A-03
XVII-119-74
—
§ 260B.007
§ 43-21-151
—
41-5-206
—
62B.330
—
—
§32A-2-3
—
—
VI-232.8
14-15-10-3; 31-30-1-1; 3130-1-2
—
—
10-§1010
—
XLVII-985.557, XLVII985.56
§ 15-11-28
—
20-5-20-509
705-405/5-130
—
Welfare and Institutions2-1-2-17-707.1; Welfare
and Institutions-2-1-2-17707
—
—
15-503-§3101;
§4–202
—
—
§ 260B.125
§ 43-21-157
XII-211.071
—
—
—
XII-169-B:27
—
—
38-2302
—
VI-232.45A
31-30-1-2
—
—
10-§1010
II-§ 16-2307
XLVII-985.556; XLVII985.557
—
3-31-§571-22
20-5-20-509
705-405/5-130
—
Welfare and Institutions2-1-2-17-707.1; Welfare
and Institutions-2-1-2-17707
8-202; 8-302; 8-341; 13-501 8-341
47.12.030
§ 12-15-102; § 12-15-204
Statutory Exclusion
Table A-3: Statutory references for criminal majority and juvenile transfer laws
50/50
VIII-1-804
15-501-§3003
§4–202.1
XVII-119-52
712A.2
§ 260B.007
§ 43-21-151
XII-211.021
41-1-101
§ 43-245
62A.030 XII-169-B:2
2A:4A-22
§32A-1-4; §32A-2-3
11-A-U-§720.10
§ 7B-1604
14-10-01; 12.1-04-01
XXI-2152.02
10A-§ 2-1-103
34-§419C.005
42-VI-§6302; Rules of
—
Juvenile Court BehaviorRule 120
§ 14-1-3
§ 63-19-20
§26-7A-1; §26-8C-2
37-11-102; 37-1-102
§ 51.02
15-2-1
33-51-§ 5102
§ 16.1-228
26.28.010; 13.40.020
§49-1-2; §49-5-1
938.02
§ 14-6-203; §14-1-101
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
—
—
§ 14-6-203
—
—
33-52-§ 5203
§ 16.1-269.1
—
—
—
—
—
10A-§2-5-206
—
—
—
—
—
III-4-305
—
—
—
712A.2
—
—
—
41-5-206
§ 43-276
—
—
—
—
—
—
38-101; 38-2302;
LI-600.020
Kansas
Kentucky
—
VI-232.2; XV-599.1
Iowa
§49-5-10
938.18.
§ 14-6-237
§ 54.02
78A-6-703
33-52-§ 5204
§ 16.1-269.1
13.40.110
37-1-134
§ 63-19-1210
§26-11-4
§ 14-1-7; § 14-1-7.1
10A-§2-5-205
34-§419C.349; 34§419C.352
42-VI-§6355
XXI-2152.1
§ 27-20-34
—
§ 7B-2200; § 7B-2203
VIII-11-862
15-503-§3101;
§ 3-8A-06
—
712A.2d
§ 260B.125
§ 43-21-157
XII-211.071
—
—
62B.390
XII-169-B:24
2A:4A-26 —
38-2347;
LI-640.01; LI-635.020
VI-232.45
§49-5-10
—
—
—
78A-6-702
—
§ 16.1-269.1
—
—
§ 63-19-1210
—
§ 14-1-5
Rules of Juvenile Court
Behavior-Rule 394
—
—
XXI-2152.1; XXI-2152.12
§ 27-20-34
—
§ 7B-2200
VIII-11-857
—
—
—
—
§ 260B.125
—
—
—
—
62B.330
XII-169-B:24
2A:4A-26 —
—
LI-635.02
—
—
938.183
—
§ 51.03; § 8.07
78A-6-701
33-51-§ 5102
—
13.04.030
—
§ 63-19-20
§26-11-3.1
—
42-VI-§6302
10A-§ 2-5-101
14-§ 137.707
—
—
11-A-H-§180.75
—
III-4-305
—
§ 3-8A-03
XVII-119-74
—
§ 260B.007
§ 43-21-151
—
41-5-206
—
62B.330
—
—
§32A-2-3
—
—
VI-232.8
—
938.183
—
§ 54.02
78A-6-702
—
§ 16.1-271
13.40.020
37-1-134
—
§26-11-4
§ 14-1-7.3.
42-VI-§6302
10A-§ 2-5-204
Psuedo: 34-§419C.364
XXI-2152.02
§ 27-20-34
—
§ 7B-1604
—
15-503-§3101;
§4–202
—
—
§ 260B.125
§ 43-21-157
XII-211.071
—
—
—
XII-169-B:27
—
—
38-2302
—
VI-232.45A