+ All Categories
Home > Documents > Punishment and Deterrence of Crime

Punishment and Deterrence of Crime

Date post: 06-Apr-2018
Category:
Upload: somenumber91823
View: 221 times
Download: 0 times
Share this document with a friend

of 37

Transcript
  • 8/3/2019 Punishment and Deterrence of Crime

    1/37

    1

    Do Harsher Punishments Deter Crime?

    Perceptions and Behavior Around the Age of Criminal Majority

    Randi Hjalmarsson

    +

    University of MarylandDecember 2006

    PRELIMINARY AND INCOMPLETE

    Abstract

    This paper uses individual survey data (NLSY97) to assess: (i) whether perceived punishmentseverity varies discontinuously at the age of criminal majority and (ii) whether changes inperceptions at the age of criminal majority deter crime. For the sample of males, I find that the

    perceived chance of jail increases by approximately six percentage points, on average, when theindividual becomes an adult in the eyes of the courts: note that this effect is over and above theeffect of ageing in general. Using whether or not the individual has reached the age of criminalmajority as an instrument for perceived punishment severity yields evidence of deterrence forauto thefts, thefts of goods worth more than $50, and drug sales. Changes in perceptions at theage of criminal majority reduce a males propensity to steal a car by 50 percent and hispropensity to commit a theft of more than $50 or to sell drugs by approximately 20 percent. Incontrast, evidence of deterrence is not observed when considering assaults or arrests rather thancrimes.

    + Any remaining errors are my own. I can be contacted at the following address. Email: [email protected], Cell:203-435-2954, Office: 301-405-4390, Mailing Address: University of Maryland, School of Public Policy, 4131 VanMunching Hall, College Park, Maryland 20742.

  • 8/3/2019 Punishment and Deterrence of Crime

    2/37

    2

    I. Introduction

    The economics of crime literature is based on the premise that an individual will commit

    a crime if the expected benefits or rewards from that crime are greater than the expected costs

    (Becker, 1968). Thus, individuals should be deterred from crime as the expected probability or

    severity of punishment increases. Papers in the economics literature that test this hypothesis

    typically use official measures of arrest, conviction, and incarceration rates as well as sentence

    lengths to measure the probability or severity of punishment.1

    Underlying these papers,

    therefore, is the assumption that individuals are knowledgeable both of these official measures as

    well as any changes in laws that may affect them. Perhaps the more relevant measure, however,

    is the individuals perception of the certainty and severity of punishment; for instance, an

    individual cannot be deterred by an increase in the severity of punishment if he is not aware of it.

    Lochner (2005) brought this issue, one commonly considered by criminologists, to the

    attention of the economics literature. Using two individual surveys, he assessed (i) the

    determinants of an individuals perceived probability of arrest and (ii) whether the perceived

    probability of arrest has a deterrent effect on criminal activity. He finds that the heterogeneity in

    beliefs are not well explained by individual background or neighborhood characteristics, but that

    perceptions do respond to changes in an individuals own criminal and arrest history as well as to

    that of their siblings. In addition, Lochner finds that youth with a lower perceived probability of

    arrest are significantly more likely to commit a crime, consistent with deterrence theory.

    In this paper, I extend the literature in two ways. First, I analyze the determinants of

    individual perceptions of punishment severity; Lochner (2005) limits his analysis to perceptions

    of arrest. In particular, I focus on how perceptions of punishment severity respond to an

    1 A vast and continually growing number of papers are included in this literature. See, for instance, Grogger (1991),Levitt (1997, 1998), Myers (1983), Tauchen, Witte and Griesinger (1994), and Witte (1980).

  • 8/3/2019 Punishment and Deterrence of Crime

    3/37

    3

    individual reaching the age of criminal majority.2 Objectively speaking, punishments are much

    harsher for individuals tried as adults compared to juveniles. For example, Lee and McCrary

    (2005) estimate that adult incarceration lengths are 3.1 to 7.4 times as long as juvenile

    incarceration lengths in Florida. Thus, the first question studied in this paper is whether

    individual perceptions of sentence severity reflect the increased propensity to be incarcerated and

    the longer sentence lengths faced by adults. Do individuals update their beliefs, and by how

    much, when they become adults in the eyes of the courts? Anecdotal evidence of such belief

    updating is provided in a 1983 paper by Glassner et. al. One interviewee said: I try to be as

    careful as much as I can these days. Cause you know, I know I can go to jail, cause they

    changed the law. You can go to jail at sixteen. In fact, more than 60% of subjects being

    interviewed in a medium sized city in New York State indicated that there are big differences in

    their perceptions of how criminals are treated by the adult and juvenile justice systems.

    Second, this paper studies whether increased perceptions of punishment severity at the

    age of majority deter crime. This question is similar in concept to that studied by both Lee and

    McCrary (2005) and Levitt (1998). Lee and McCrary utilize the discontinuity in expected

    punishment that occurs when juveniles reach the age of majority in Florida and detailed

    administrative arrest data to test for a deterrence effect of punishment severity. They find no

    evidence of a systematic drop in arrest rates at an individuals eighteenth birthday. One would

    only expect to find such a drop in arrest rates, however, if individuals perceptions of sanction

    severity also change discontinuously at the age of majority. Changes in perceptions cannot be

    directly observed by Lee and McCrary; rather, they must rely on changes in individual arrest

    2 The age at which individuals are processed in the adult criminal justice system rather than the juvenile justicesystem is the age of criminal majority and is determined independently by each state. In 38 states, the age ofcriminal majority is 18; an 18-year old will be tried in the criminal courts. The age of criminal majority is 17 in tenstates and 16 in three states.

  • 8/3/2019 Punishment and Deterrence of Crime

    4/37

    4

    behavior to make inference about perceptions. Likewise, Levitt (1998) also does not directly

    observe whether perceptions change at the age of criminal majority. But, in contrast to Lee and

    McCrary, Levitt (1998) finds evidence of deterrence. He concludes that crime rates change

    significantly at the age of criminal majority. 3

    There is a fairly extensive criminology literature that is concerned with the role of

    perceptions; i.e. how perceptions of the criminal justice system are formed and how such

    perceptions affect crime rates. This literature includes studies of the deterrent effects of both the

    perceived certainty and perceived severity of punishment.4 Recent papers also address the

    determinants of changes in perceptions (Pogarsky, Piquero, and Paternoster, 2004; Matsueda,

    Kreager, and Huizinga, 2006). But, to the best of my knowledge, neither the economics literature

    nor the criminology literature has previously assessed whether perceptions change

    discontinuously at the age of criminal majority and whether this change in perceptions deters

    crime.

    This study ueses self-reported survey data, specifically the National Longitudinal Survey

    of Youth 1997 (NLSY97). Using survey data of this sort yields a number of advantages.

    Respondents are surveyed in multiple rounds about their perceptions of punishment severity as

    well as self-reported criminal activity, allowing for the identification of within individual

    changes in perceptions and behavior. In contrast to both Lee and McCrary (2005) and Levitt

    (1998), I can directly observe whether perceptions change at the age of criminal majority and I

    can test for a deterrence effect on criminal behavior, i.e. rather than arrest behavior.

    3 Note that while both Lee and McCrary (2005) and Levitt (1998) are looking for changes in behavior around theage of criminal majority, they use very different identification strategies and data sets. As described in the text, Leeand McCrary look for discontinuities in an individuals arrest behavior immediately after he turns 18. Levitt (1998)uses state level panel data and examines how cohort specific arrest rates change at the age of criminal majority.

    4 For instance, see Klepper and Nagin (1989), Paternoster (1987), and Horney and Marhsall (1992).

  • 8/3/2019 Punishment and Deterrence of Crime

    5/37

    5

    The empirical analysis is conducted in two stages. I first assess how an individuals

    perceptions of punishment severity change at the age of criminal majority. Basically, an

    individuals perceived chance of being sent to jail if they are arrested for stealing a car is

    regressed on his age and whether he has reached the age of adult jurisdiction; individual fixed

    effects are also included to control for unobservable heterogeneity. A discontinuous change in

    perceptions is observed for the entire sample and for males. For males, the perceived chance of

    jail increases by six percentage points, on average, when the individual becomes an adult in the

    eyes of the courts, over and above the effect of ageing in general. This estimate is not sensitive

    to the inclusion of a large set of controls. In addition, evidence that this change in perceived

    punishment severity can really be attributed to reaching the age of criminal majority is provided

    in a number of falsification tests; for instance, I consider whether there is a discontinuity in

    perceptions at an age other than the age of criminal majority.

    The second stage of the empirical analysis tests whether changes in perceived punishment

    severity at the age of criminal majority deter crime. Specifically, criminal activity in period t+1

    is regressed on perceived punishment severity and individual fixed effects; whether or not the

    individual has reached the age of criminal majority is used to instrument for perceived

    punishment severity. Evidence of deterrence is seen for auto thefts, thefts of goods worth more

    than $50, and drug sales. Changes in perceptions at the age of criminal majority reduce a males

    propensity to steal a car by 50 percent and his propensity to commit a theft of more than $50 or

    to sell drugs by approximately 20 percent. In contrast, evidence of deterrence is not observed

    when considering assaults or arrests rather than crimes.

    Thus, this paper finds evidence that increases inperceivedpunishment severity at the age

    of criminal majority deter some types of criminal activity. What are the policy implications of

  • 8/3/2019 Punishment and Deterrence of Crime

    6/37

    6

    such a result? These findings do not imply that handing out more severe punishments, e.g. more

    and longer jail sentences, will decrease crime. Deterrence would only occur if individuals update

    their beliefs regarding punishment severity. Rather, it seems to me that the relevant policy

    implication is to increase the dispersion of information about the justice system to the

    population. These results indicate that we could potentially deter crime by making the

    population more aware of the differences in punishment severity across the juvenile and adult

    justice systems.

    The remainder of the paper proceeds as follows. Section II presents the empirical design

    and Section III describes the data, focusing on the measurement of perceived punishment

    severity. Section IV analyzes whether perceived punishment severity varies discontinuously at

    the age of criminal majority and Section V tests whether these changes in perceptions deter

    crime. Section VI concludes.

    II. Empirical Methodology

    The empirical analysis is conducted in two stages. The first stage assesses how an

    individuals perceptions change at the age of criminal majority. The basic specification that is

    taken to the data is presented in equation (1), where i, s, and trepresent the individual, state of

    current residence, and survey round, respectively.

    (1)0 1

    _ist ist ist ist i ist

    P Age Adult Jur X = + + + + +

    Perceived punishment severity, P, is regressed on an individuals age at the time of each survey

    round as well as a dummy variable indicating whether the individual has reached the age of adult

  • 8/3/2019 Punishment and Deterrence of Crime

    7/37

    7

    jurisdiction,Adult_Jur, in his state of residence. For instance, consider a state where the age of

    criminal majority is 18. For an individual who is 17 at the time of the survey and living in such a

    state, Adult_Jur would equal zero. However, when the individual is 18 (i.e. in the following

    survey round),Adult_Jurequals one. Thus, 1, the coefficient on this dummy variable, captures

    the effect of becoming an adult in the eyes of the courts on an individuals perception of

    punishment severity, over and above the effect of ageing in general. Individual fixed effects,,

    are included and control for individual characteristics that are fixed over time. Moreover,

    including fixed effects implies that the identification of1 is based on within individual changes

    in perceptions. In contrast, cross sectional data would identify 1 by comparing perceived

    punishment severity across individuals, some who have reached the age of criminal majority and

    some who have not. Lastly, some specifications also control for observed time-varying

    individual characteristics,X.

    The second stage of the analysis tests whether changes in perceived punishment severity

    at the age of criminal majority deter crime. As depicted in equation (2), self-reported criminal

    activity, C, in period t + 1 is regressed on perceived punishment severity in period t.

    (2) 1 0 1ist ist ist i ist C P Z + = + + + +

    Crime in period t + 1 is used both to eliminate the issue of simultaneity, i.e. that criminal activity

    determines perceived punishment severity, and to ensure that we are studying criminal activity

    after the individual has reached the age of criminal majority. Once again, individual fixed

    effects are included as well as a vector of time-varying individual characteristics, Z. The

    individual fixed effects control for fixed unobservable characteristics that are correlated with

  • 8/3/2019 Punishment and Deterrence of Crime

    8/37

    8

    both an individuals perceptions of punishment severity as well as his self-reported criminal

    activity. Whether or not the individual has reached the age of criminal majority, Adult_Jur, is

    used to instrument for perceived punishment severity, P. Thus, equation (1) is essentially the

    first-stage equation. Using such an instrumental variables approach only identifies a local

    treatment effect; specifically, it captures the effects of changes in perceptions at the age of

    majority on criminal activity.

    III. Data

    The National Longitudinal Survey of Youth 1997 (NLSY97) consists of 8,984

    individuals who were between the ages of 12 and 16 as of December 31, 1996. This paper

    focuses on survey questions concerned with an individuals perceptions of the criminal justice

    system as well as his self-reported criminal activity. An individuals perception of punishment

    severity, the primary variable of interest in this paper, is measured by the following survey

    question. Suppose you were arrested for stealing a car, what is the percent chance that you

    would serve time in jail?5 Because this question is only asked during the first five survey

    rounds, a panel data set of five years is created. Additionally, the created data set includes

    measures of: (i) individual demographic characteristics, (ii) individual interactions with the

    justice system, such as arrest and incarceration, (iii) family background, and (iv) geographic

    characteristics. This last category includes the respondents state of residence, which was

    obtained from the NLSY97 Geocoded data. This variable is essential to the analysis, as it allows

    the appropriate age of criminal majority to be merged into the data. Recall that the age of

    5 Respondents are also asked the percent chance that they would be arrested if they stole a car, the chance that theywould be arrested and released without charge, and the chance that they would be arrested and released after payinga fine.

  • 8/3/2019 Punishment and Deterrence of Crime

    9/37

    9

    criminal majority is not 18 in all states. During the first survey round, 59 percent of respondents

    lived in a state where the age of criminal majority is 18; but, approximately 29 percent and 12

    percent, respectively, resided in states with ages of criminal majority equal to 16 and 15.

    Table 1 provides selected summary statistics for each of the five survey rounds; Panel I is

    for the entire sample and Panel II is restricted to the sample of males. The average age in the

    sample is 14.3 years in the 1997 survey round and increases to 19.0 in 2001. Thus, just 5.1

    percent of the sample has reached the age of criminal majority in 1997 while 87.3 percent would

    be tried as adults in 2001. These statistics are quite comparable when looking at the sample of

    males. In contrast, much higher rates of arrest and incarceration are seen for males relative to the

    entire sample. 10 percent of males were arrested at least once prior to the first survey round

    while more than 30 percent were arrested prior to the 2001 round; for the entire sample, the

    comparable numbers are 7.5 and 22, respectively. Likewise, 6.4 percent of males were

    incarcerated prior to 2001 but just 4.1 percent of the entire sample. According to self-reports, by

    the time of the 2001 survey round, almost 5 percent of males stole a car, 25 percent sold drugs,

    40 percent committed an assault, and 21 percent stole something worth more than $50.

    The shaded rows in Panels I and II of Table 1 indicate the average perceivedchance of

    going to jail conditional on being arrested for stealing a car. In the first survey round, the sample

    believes that there is a 45 percent chance, on average, of going to jail. The average belief for

    males is slightly higher. The perceived chance of jail increases in each year for both samples. By

    2001, the entire sample perceives the chance of jail to be 52.7 percent while the male sample

    believes the chance of jail is 56.6 percent. Thus, for males, the average perceived chance of jail

    increases by more than 10 percentage points during these five years. In contrast, there is little

    change over time in both the perceived chance of arrest and the perceived chance of arrest and

  • 8/3/2019 Punishment and Deterrence of Crime

    10/37

    10

    release; for males, both decrease by about one percentage point from the first to last year. While

    perceptions of punishment severity increase over time, i.e. as the sample ages, one cannot

    identify from the statistics presented thus far whether this increase is being driven by individuals

    reaching the age of criminal majority.

    Figure 1 depicts a histogram of the perceived chance of jail. There are three mass points.

    Approximately 20 percent of the sample reports that there is no chance of going to jail while

    another 20 percent thinks that the chance of jail is certain. Lastly, slightly more than 20 percent

    of the sample believes that there is a fifty-fifty chance of jail.

    Table 2 begins to explore whether the perceived chance of jail varies with whether the

    respondents are considered juveniles or adults by the criminal justice system. The first column

    of the table presents the average perceived chance of jail when looking at all survey rounds

    simultaneously; each row presents a different sub-sample. On average, males believe that the

    chance of jail when stealing a car is 51.7 percent. However, juvenile males believe that the

    chance is 47.9 percent while adult males believe there to be a 56.1 percent chance; this

    represents an approximately eight percentage point difference between juveniles and adults.

    Males who have interacted with the justice system believe the chance of jail to be greater.

    Arrested males believe there to be a 58.3 percent chance of jail while incarcerated males believe

    there to be a 70 percent chance. Within these groups, however, one can still observe an

    approximately eight percentage point difference in perceptions between juveniles and adults.

    IV. Do Perceptions Vary Discontinuously at the Age of Majority?

    Graphical Analysis

  • 8/3/2019 Punishment and Deterrence of Crime

    11/37

    11

    Before estimating equation (1), I graphically assess how perceptions change at the age of

    criminal majority. Figure 2 plots the average perceived chance of jail if arrested for stealing a

    car versus an individuals age relative to the age of criminal majority. An individuals relative

    age is zero when it is the individuals first year in the adult criminal justice system; this is

    depicted by the vertical line in Figure 2. Depending on the individuals state of residence, a

    relative age of zero corresponds to individuals who are actually 18, 17, or 16. Average perceived

    chance of jail is plotted for both the entire sample and just the sample of males.

    First, one can observe that males perceive the chance of jail to be slightly higher than the

    entire sample at all relative ages. One can also observe a discontinuity in perceptions when both

    samples reach the age of criminal majority. Average perceived chance of jail increases by 3.9

    percentage points for the entire sample and by 5.5 percentage points for males. Males who are

    one year younger than the age of criminal majority, on average, perceive the chance of jail to be

    49.4 percent; the perceived chance of jail jumps to 54.9 at the age of majority. Note that the

    average perceived chance of jail for males who were five years younger than the age of majority

    is 48.1 percent (as compared to 49.4 percent just one year prior to the age of majority).

    Figure 3 plots the average perceived chance of arrest for stealing a car versus relative age

    for the sample of males. However, in contrast to the perceived chance of jail, a discontinuity in

    the perceived chance of arrest is not observed.

    Regression Analysis

    Though the remainder of the analysis is restricted to the sample of males, results are

    qualitatively similar when using the entire sample, as has already been observed in Figure 2.

    Table 3 presents the results of estimating equation (1) for the sample of males. Additional

  • 8/3/2019 Punishment and Deterrence of Crime

    12/37

    12

    controls, X, are included in each column; all specifications presented in Table 3 include

    individual fixed effects. The first column of Table 3 controls only for the individuals age and

    whether he is an adult in the eyes of the courts. On average, the perceived chance of jail

    increases by 1.2 percentage points per year as the sample gets older. Over and above that,

    however, the perceived chance of jail for stealing a car is approximately 6.0 percentage points

    higher for males who have reached the age of criminal majority relative to males who are still

    juveniles. Column 2 adds in controls for self-reported criminal activity (i.e. thefts of less than

    $50, thefts of more than $50, assault, and selling drugs) and column 3 controls for interactions

    with the justice system (i.e. arrest, charge, conviction, and incarceration). The inclusion of these

    variables has virtually no effect on the coefficient corresponding to Adult_Jur. Likewise,

    controlling for state fixed effects and time varying individual characteristics, such as whether the

    individual lives in a metropolitan statistical area or in a two parent household, has no effect on

    the coefficient of interest. Adult males still believe that the chance of jail if arrested for stealing

    a car is, on average, six percentage points greater than that perceived by juvenile males.6

    Robustness: Falsification Tests

    Table 4 provides some evidence that this change in perceptions is really attributable to

    individuals reaching the age of criminal majority. Specifically, Table 4 presents the results of

    estimating equation (1) for a number of sub-samples when using false age of majority dummy

    variables; each specification includes individual fixed effects and the full set of controls. For

    6 It is not surprising that controlling for state fixed effects has no impact on the coefficient corresponding to whetherthe individual has reached the age of criminal majority. When individual fixed effects are included, there is littlevariation left in the state dummies; i.e. the only variation is for individuals who move. But, one should note that statefixed effects are quite important when individual fixed effects are omitted and identification comes from acrossindividuals. States that set the age of criminal majority to 16 may also have a number of other laws or policies thataffect an individuals perception of punishment severity.

  • 8/3/2019 Punishment and Deterrence of Crime

    13/37

    13

    instance, columns (1) through (4) restrict the analysis to the sample of 1,451 males from states

    where the age of criminal majority is 17. The first column presents the results when age and

    Adult_Jur are included; note that in this case, Adult_Jur is equivalent to a dummy variable

    indicating if the individual is 17 years old or older. Thus, as seen in column (1), the perceived

    chance of jail significantly increases by 4.5 percentage points when the actual age of criminal

    majority is reached. The next three columns present the results of using false age of majorities.

    Column (2) assumes that the age of majority is 16 and tests for a significant change in

    perceptions at the age of 16 (i.e. one year before such a change should theoretically be observed).

    However, the perceived chance of jail does not significantly change around this false age of

    majority. Likewise, a change in perceptions is not observed when treating both 18 and 19 as the

    age of majority in states where the age of criminal majority is actually 17.

    Columns (5) through (7) of Table 4 repeat this analysis for individuals from states where

    the age of majority is 18. Using the appropriate age of majority, in column (5), one observes a

    6.5 percentage point increase in the perceived chance of jail. Using 17 or 19 as the age of

    criminal majority, in columns (6) and (7) respectively, yields just a 2 percentage point increase in

    the perceived chance of jail; these coefficients are however significant. Lastly, column (8) uses

    the entire sample and acts if the age of majority is equal to 19 years old. In this case, no

    significant change in the perceived chance of jail is observed; in fact, the coefficient on a dummy

    variable indicating whether the individual is 19 or older is equal to 0.11.

    Thus, there is evidence that one really can attribute the change in the perceived chance of

    jail to an individual having reached the age of criminal majority.

    Heterogeneity

  • 8/3/2019 Punishment and Deterrence of Crime

    14/37

    14

    Before turning to the question of whether this increase in perceived punishment severity

    at the age of criminal majority deters crime, I explore the change in perceptions a bit more

    closely. Does the magnitude of the change vary across different sub-samples of the data? Each

    row of Table 5 presents the results of estimating equation (1) for a different sub-sample of males;

    only the coefficient on whether the individual has reached the age of criminal majority is

    displayed. The first row corresponds to the entire sample of males, i.e. the baseline results, and

    indicates a 6 percentage point increase in the perceived chance of jail at the age of criminal

    majority. Rows (2) through (5) assess whether this effect varies with criminal experience. Two

    measures of criminal experience are considered: whether an individual reports being arrested and

    whether an individual reports committing any crimes in a survey round prior to reaching the age

    of criminal majority. Becoming an adult in the eyes of the court does not affect perceived

    punishment severity for individuals who were previously arrested, but increases the perceived

    chance of jail by 6.4 percentage points for individuals who were not arrested. When considering

    criminal activity rather than arrest, becoming an adult significantly increases perceived

    punishment severity for both criminals and non-criminals; the point estimates, however, are

    larger for the non-criminals. Why does reaching the age of adult jurisdiction have less of an

    influence on the criminally experienced than the non-criminally experienced? This is perhaps

    counterintuitive, as one might figure experienced criminals to be more knowledgeable of the

    justice system. One possible explanation is that individuals who are criminally experienced, and

    particularly those who have been arrested, become eligible for the adult courts at a younger age

    than the age of criminal majority. For instance, in many states, juveniles who commit serious

    enough crimes can be transferred to the adult courts. A number of states also have a policy that

    says once an adult, always an adult.

  • 8/3/2019 Punishment and Deterrence of Crime

    15/37

    15

    Rows (6) through (8) address the role played by an individuals initial level of

    perceptions.7

    Row (6) considers the 1,264 males who believe the chance of jail to be less than 50

    percent in the survey round immediately prior to reaching the age of criminal majority. For these

    individuals, becoming an adult increases the perceived chance of jail by approximately 37

    percentage points on average. In contrast, the perceived chance of jail decreases 24 percentage

    points upon becoming an adult for individuals who initially believe the chance to be greater

    than 50 percent. In some sense, these individuals have no where to go but down; it is, however,

    somewhat surprising that this decrease is observed rather than a stagnation in perceptions.

    Lastly, becoming an adult results in a 6.1 percentage point increase in the perceived chance of

    jail for the 667 males who initially believe the chance of jail to be 50 percent.

    Rows (9) through (11) display the results when grouping states according to the age of

    criminal majority. The coefficients onAdult_Jurare 6.4, 4.4, and 8.7 in states where the age of

    majority is equal to 18, 17, and 16, respectively. Thus, the point estimate is largest in those states

    with the lowest age of criminal majority, i.e. 16. One possible explanation is that states with

    lower ages of majority also have harsher punishments, which are better known to the public.

    This story is not completely supported by the data, however, as the estimated effect is larger in

    the age 18 states than the age 17 states.

    Lastly, rows (12) through (15) consider how the relative punitiveness of the adult justice

    system to the juvenile system affects the relationship between becoming an adult and perceived

    punishment severity. Similar to Levitt (1998), I measure relative punitiveness of the adult

    7 For instance, Pogarsky, Piquero, and Paternoster (2004) find that the manner in which new information affects theperceived certainty of punishment depends on the initial level of perceived certainty, i.e. before the new informationis received.

  • 8/3/2019 Punishment and Deterrence of Crime

    16/37

    16

    system to the juvenile system by calculating the ratio of adult incarceration rates to juvenile

    incarceration rates, as depicted in equation (3), for the initial year of the survey.8

    (3) s sss s

    # Incar Adults # Arrested AdultsRelative Punitiveness =

    # Incar Juveniles # Arrested Juveniles

    A high measure of relative punitiveness indicates that the adult system is quite harsh relative to

    the juvenile system. Thus, one would expect that reaching the age of criminal majority has a

    larger impact on the perceived chance of jail in states with a high relative punitiveness measure.

    I create four groups of states that are defined by the quartiles of the relative punitiveness

    measure. Somewhat surprisingly, the estimated effect of becoming an adult in the eyes of the

    courts on the perceived chance of jail is fairly homogeneous across these four groups of states.

    V. Do Changes in Perceptions Around the Age of Majority Deter Crime?

    Deterrence of Auto Theft?

    The previous section provides evidence that males, on average, update their beliefs

    regarding the chance of jail if arrested for stealing a car when they reach the age of criminal

    majority. Does this change in the perceived chance of jail deter criminal behavior? I first

    address this question for auto thefts, as this is the crime category to which the survey question

    regarding perceptions specifically refers. Table 6 presents the results of estimating equation (2)

    when the dependent variable is whether the respondent reports stealing a car in period t + 1; all

    8 Adult and juvenile arrest counts come from the FBIs annual publication, Crime in the United States 1997.However, arrest data in 1997 (or in an adjacent year) were not found in this publication for the District of Columbia,Florida, Kansas, New Hampshire and Vermont. Adult incarceration statistics are found in the Prison and JailInmates at Midyear 1997 Bureau of Justice Statistics Bulletin. The 1997 number of juveniles incarcerated comesfrom the Census of Juveniles in Residential Placement Databook.

  • 8/3/2019 Punishment and Deterrence of Crime

    17/37

    17

    specifications include individual fixed effects. When instrumenting for the perceived chance of

    jail with whether the individual has reached the age of criminal majority, the estimated

    coefficient on the perceived chance of jail is significant and equal to -0.001. That is, if the

    perceived chance of jail increases by one percentage point upon reaching the age of criminal

    majority, then the individual is 0.1 percentage point less likely to steal a car in the next period.

    As can be seen in columns (1) though (6), this estimate is not sensitive to the inclusion of a

    vector of controls. Column (7), however, indicates that the deterrence effect is not observed

    when the perceived chance of jail is not instrumented for with having reached the age of criminal

    majority. That is, there is only evidence of a local effect, i.e. that the change in perceptions at

    the age of majority deters auto theft.

    How big is this deterrence effect? On average, the perceived chance of jail increases by

    six percentage points at the age of criminal majority: thus, this translates into a 0.6 percentage

    point decrease in an individuals propensity to steal a car. Given that 1.2 percent of males report

    stealing a car, this deterrence effect is actually quite large.9 Changes in perceptions at the age of

    criminal majority reduce an individuals propensity to steal a car by 50 percent.

    Deterrence of Other Crimes?

    NLSY97 only asks the respondents perceived chance of jail if arrested for stealing a car,

    as opposed to being arrested for committing other crimes. However, it is possible that an

    individuals perceived chance of jail changes at the age of criminal majority for other crimes in a

    manner similar to that for auto theft. Table 7 estimates equation (2) for three additional crime

    categories: thefts worth more than $50, drug sales, and assaults. Odd numbered columns

    instrument for the perceived chance of jail while even numbered specifications do not.

    9 This estimate is calculated by looking at observations for all males in all survey rounds.

  • 8/3/2019 Punishment and Deterrence of Crime

    18/37

    18

    Columns (1) and (3) indicate that increases in the perceived chance of jail at the age of

    criminal majority deter thefts and drug sales, with significant point estimates of -0.002 and -

    0.003, respectively. Once again, evidence of deterrence is not observed when using all of the

    variation in perceptions rather than just the variation at the age of majority. In addition, evidence

    of deterrence is not found for assaults.

    The deterrence effects estimated for thefts and drug sales are sizeable, though not as large

    as that seen for auto thefts. Specifically, the six percentage point increase in the perceived

    chance of jail at the age of majority decreases an individuals propensity to commit a theft of

    more than $50 by 1.2 percentage points and an individuals propensity to sell drugs by 1.8

    percentage points. The average propensities of a male to commit a theft worth more than $50

    and to sell drugs are 6.1 and 8.5 percent, respectively. Thus, reaching the age of majority

    reduces an individuals propensity to commit such a theft or sell drugs by approximately 20

    percent. While this deterrence effect is lower than that observed for auto thefts, this could be due

    to the fact that the survey questions specifically targeted auto thefts.

    Heterogeneity in Deterrent Effects

    The results presented thus far indicate that the increased perceived chance of jail at the

    age of criminal majority deters auto thefts, thefts of more than $50, and drug sales. These

    findings however are based on the entire sample of males, including individuals who have never

    participated in any kind of criminal activity. Table 8 presents the results for the four categories

    of crimes when considering sub-samples of males who are more or less criminally experienced.

    The first row presents the baseline deterrent effects estimated for the entire sample in each of the

    four crime categories. The next two rows restrict the sample to males who were and were not

  • 8/3/2019 Punishment and Deterrence of Crime

    19/37

    19

    arrested prior to the age of criminal majority. The last two rows consider males who do and do

    not report committing any types of crime prior to the age of criminal majority.

    Though the point estimates are larger in the arrested than the non-arrested sample, they

    are also more imprecise as the sample size is much smaller. However, when looking at self-

    reported criminal activity (of any type), the deterrent effects are approximately double those seen

    for the baseline specification. Of course, no evidence of deterrence is seen when looking at the

    sample of individuals who report no criminal activity prior to the age of criminal majority. Thus,

    it appears to be individuals who report criminal activity, but who are not arrested, who are

    driving the estimated deterrent effects.

    Can These Results Be Reconciled with Lee and McCrary (2005)?

    To isolate the deterrence impact of more severe sentences, Lee and McCrary (2005) use

    administrative felony arrest data in Florida, where the age of criminal majority is 18. They do not

    find a significant drop in arrests when individuals turn 18. That is, they do not find evidence that

    more severe punishments deter crime. In contrast, I find that reaching the age of criminal

    majority significantly deters auto theft, thefts of more than $50, and drug sales. Can these two

    sets of findings be reconciled?

    One factor to note is that Lee and McCrary restrict their analysis to the study of index

    crimes. Thus, the crime categories studied in both papers are not completely comparable; non-

    index crimes, such as drug sales and petty theft, are included in the NLSY97 analysis. Thus, it

    may simply be the case that increased punishment severity deters only relatively minor crimes.

    Another potential explanation of the differential results may be that Lee and McCrary

    study arrests while I study self-reported criminal activity. It could be the case that increased

  • 8/3/2019 Punishment and Deterrence of Crime

    20/37

    20

    punishment severity deters crime, but that this deterrence effect is not reflected in arrests. The

    relationship between self-reported criminal activities and arrest in the NLSY97 sample is perhaps

    not as strong as one would expect. For instance, the proportions of individuals who report

    committing any crime or a serious crime decrease in every survey round.10 For instance, as can

    be seen in Table 1, the proportion of males who report committing any crime since the date of

    the last interview decreases monotonically from 56.8 percent to 22.8 percent. Likewise, the

    proportion who report committing a serious crime decreases from 29.7 percent to 16.4 percent.

    In contrast, the proportion of individuals who are arrested decreases slightly from the first period

    to the last (10 to 9.4 percent), but this does not occur in a monotonic manner. For the most part,

    the same proportion of individuals is arrested in each survey round. In addition, the correlation

    coefficient between arrest and whether the individual reports any crime is just 0.25.

    To assess whether the change in perceptions deters arrests, I estimate equation (2) when

    the dependent variable is whether the individual is arrested in period t + 1, rather than whether

    he commits a crime. As can be seen in Table 8, there is no evidence that increases in perceived

    punishment severity at the age of criminal majority impact arrests. Thus, looking at just arrests

    would lead one to conclude that increased punishment severity does not deter crime, while one

    would reach the opposite conclusion when studying criminal activity directly.

    Robustness: Distinguishing Deterrence from Incapacitation

    A challenge commonly encountered in the economics of crime literature is the ability to

    distinguish between deterrence and incapacitation, i.e. the decrease in crime that results from

    10 The any crime category includes destruction of property, thefts of less than $50, thefts of more than $50, otherproperty crimes, assault, drug sales, and car theft. The any serious crime category is restricted to include thefts ofmore than $50, assault, drug sales, and car theft. Note that for the first survey round, these variables capture whetherthe individual committed these crimes at any time prior to the date of the first survey (i.e. not just in the last year).

  • 8/3/2019 Punishment and Deterrence of Crime

    21/37

    21

    isolating offenders from society (National Research Council, 1978). The results presented thus

    far in this paper indicate that increases in the perceived chance of jail at the age of criminal

    majority deter auto thefts, thefts of more than $50, and selling drugs. Is it possible that these

    estimated deterrence effects are actually capturing incapacitation? That is, are individuals

    committing less crime just after reaching the age of majority because they are incarcerated and

    unable to commit crimes in society?

    Table 10 presents evidence that incapacitation is not driving the estimated deterrence

    effects. The first three columns consider the case of auto theft. Column (1) presents the results

    of estimating equation (2), yielding a significant coefficient of -.001 on the perceived chance of

    jail. Included in this initial specification is whether the individual is incarcerated in period t.

    Controlling for such incarceration would not, however, capture incapacitation in period t+1, i.e.

    when criminal activity is being measured. Column (2) adds in a control for whether the

    individual is incarcerated at the time of the interview. This coefficient is negative and

    significant, suggesting that incarceration does have an incapacitation effect. However, its

    inclusion does not impact the estimated coefficient on the perceived chance of jail. Column (3)

    controls for whether the individual is incarcerated at all during period t+1. The resulting

    coefficient is actually significant and positive, perhaps capturing the fact that individuals who

    report auto thefts in t+1 are also more likely to be incarcerated in that period. But, once again,

    controlling for this measure does not impact the estimated deterrence effect of increasing

    perceived punishment severity. Thus, one can infer from these results that incarceration at the

    time of the interview or during the next period does not vary discontinuously at the age of

    criminal majority.

  • 8/3/2019 Punishment and Deterrence of Crime

    22/37

    22

    The remaining columns of Table 10 replicate this analysis for the other crime categories:

    thefts of more than $50, selling drugs, and committing an assault. Incarceration at the time of the

    interview has a negative coefficient associated with it in all specifications, though it is not always

    significant. So, there does appear to be some incapacitation. However, controlling for such

    incarceration or incarceration in period t+1 does not impact the estimated deterrent effects of

    changes in perceptions at the age of criminal majority for any of the crime categories.

    VI. Conclusions and Policy Implications

    This paper studies the effects of reaching the age of criminal majority on both perceptions

    of punishment severity and criminal activity. On average, males believe that the chance of going

    to jail if arrested for stealing a car increases by approximately six percentage points at the age of

    criminal majority. Does this increase in perceived punishment severity deter crime? Evidence of

    deterrence is seen for auto theft, thefts of more than $50, and drug sales. However, such

    evidence is not observed when considering assaults or looking at arrests rather than criminal

    activity. Likewise, Lee and McCrary (2005) find no evidence that harsher punishments deter

    crime when looking at arrests.

    Finding that individuals are deterred from crime when they reach the age of criminal

    majority does not imply that handing out more severe punishments, e.g. more and longer jail

    sentences, will decrease crime. Such deterrence would only occur if individuals update their

    beliefs regarding punishment severity. Rather, the more relevant policy implication may be to

    provide the population of potential criminals with more knowledge of the differences in

    punishment severity across the juvenile and adult justice systems.

  • 8/3/2019 Punishment and Deterrence of Crime

    23/37

    23

    References:

    Becker, Gary. (1968) Crime and Punishment: An Economic Approach, Journal of PoliticalEconomy, 76: 169-217.

    Glassner, Barry, Margret Ksander, Bruce Berg, and Bruce Johnson. (1983) A Note on theDeterrent Effect of Juvenile vs. Adult Jurisdiction, Social Problems, 31(2).

    Grogger, Jeffrey (1991) Certainty vs. Severity of Punishment,Economic Inquiry, 29: 297-301.

    Horney, Julie and Ineke Marshall (1992) Risk Perceptions Among Serious Offenders: The Roleof Crime and Punishment, Criminology, 30(4).

    Klepper, Steven and Daniel Nagin (1989) The Deterrent Effect of Perceived Certainty andSeverity of Punishment Revisited, Criminology, 27(4).

    Lee, David and Justin McCrary (2005) Crime, Punishment, and Myopia, NBER WorkingPaper 11491.

    Levitt, Steven (1997) Using Electoral Cycles in Police Hiring to Estimate the Effect of Policeon Crime,American Economic Review, 87(3).

    Levitt, Steven (1998) Juvenile Crime and Punishment, The Journal of Political Economy,106(6).

    Lochner, Lance (2005) Individual Perceptions of the Criminal Justice System,NBER WorkingPaper #9474. Forthcoming American Economic Review.

    Matsueda, Ross, Derek Kreager, and David Huizinga (2006) Deterring Delinquents: A RationalChoice Model of Theft and Violence,American Sociological Review, 71: 95-122.

    Myers, Samuel (1983) Estimating the Economic Model of Crime: Punishment vs. DeterrentEffects, Quarterly Journal of Economics, 98: 157-66.

    The National Research Council. (1978) Deterrence and Incapacitation: Estimating the Effects ofCriminal Sanctions on Crime Rates Eds. Alfred Blumstein et. al., National Academy ofSciences, Washington, D.C.

    Paternoster, Raymond (1987) The Deterrent Effect of the Perceived Certainty and Severity of

    Punishment: A Review of the Evidence and Issues,Justice Quarterly, 4(2).

    Pogarsky, Greg, Alex Piquero, and Ray Paternoster (2004) Modeling Change in Perceptionsabout Sanction Threats: The Neglected Linkage in Deterrence Theory, Journal of QuantitativeCriminology, 20(4).

    Tauchen, Helen, Ann Dryden Witte, and Harriet Griesinger (1994) Criminal Deterrence:Revisiting the Issue with a Birth Cohort,Review of Economics and Statistics, 76(3).

  • 8/3/2019 Punishment and Deterrence of Crime

    24/37

    24

    Trumbull, William (1989) Estimations of the Economic Model of Crime Using Aggregate andIndividual Level Data, Southern Economic Journal, 94: 423-4369.

    Witte, Ann Dryden (1980) Estimating the Economic Model of Crime with Individual Data,

    Quarterly Journal of Economics, 94(1).

  • 8/3/2019 Punishment and Deterrence of Crime

    25/37

    25

    Table 1. Selected Summary Statistics

    Survey Round1997Mean

    1998Mean

    1999Mean

    2000Mean

    2001Mean

    Panel I: Entire Sample

    % Chance Arrest 58.7 57.5 61.3 59.7 58.3% Chance Arrest and Release 32.8 31.4 33.7 33.5 34.0% Chance Arrest and Jail 45.0 46.5 50.0 51.8 52.7Age 14.3 16.0 16.9 18.0 19.0% in Adult Jurisdiction 5.1 29.2 48.1 69.4 87.3% Male 51.2 51.2 51.1 51.1 50.7% Black 25.9 26.3 25.9 26.2 25.7% Hispanic 21.0 21.2 21.0 20.9 21.2% Ever Arrested 7.5 12.9 15.9 19.4 22.1% Ever Incarcerated 0.5 1.7 2.4 3.4 4.1% Ever Steal Car 1.3 2.2 2.6 2.8 3.2

    % Ever Sell Drugs 6.6 11.1 14.2 16.9 19.0% Ever Assault 18.3 25.2 28.9 31.1 32.6% Ever Steal > $50 7.5 10.9 13.2 14.9 15.9% Any Crime 48.7 29.9 24.7 21.3 18.0% Any Major Crime 23.8 18.8 16.1 14.6 12.7% Any Arrest 7.5 7.0 6.3 6.9 6.3

    Panel II: Males% Chance Arrest 58.8 56.4 60.4 58.6 57.8% Chance Arrest and Release 30.5 28.9 30.3 29.3 29.6

    % Chance Arrest and Jail 46.4 48.4 52.2 55.5 56.6Age 14.3 15.9 16.9 18.0 18.9% in Adult Jurisdiction 5.0 29.0 47.6 69.7 87.7% Black 25.3 25.8 25.4 25.5 24.8% Hispanic 21.0 21.1 20.9 20.9 21.2% Ever Arrested 10.0 17.0 21.7 26.4 30.1% Ever Incarcerated 0.7 2.6 3.6 5.3 6.4% Ever Steal Car 1.7 3.1 3.8 4.1 4.6% Ever Sell Drugs 8.5 13.8 18.3 22.0 24.5% Ever Assault 23.5 31.2 35.5 38.1 40.2% Ever Steal > $50 9.9 14.3 17.5 19.4 20.6

    % Any Crime 56.8 35.9 29.3 27.1 22.8% Any Major Crime 29.7 23.1 20.0 19.0 16.4% Any Arrest 10.0 9.8 9.3 10.3 9.4

    Summary statistics are presented for those individuals for whom % chance of arrest and jail is not missing. Variablesbeginning % Ever indicate the percentage of the sample that was ever arrested, for instance; i.e. prior to therelevant survey round. Variables beginning % Any indicate the percentage of the sample that was arrested, forinstance, at least once since the date of the last interview.

  • 8/3/2019 Punishment and Deterrence of Crime

    26/37

    26

    Table 2. Average Perceived Chance of Jail for Different Sub-Samples of Males

    All Rounds 1998 Round 2000 Round

    All Males 51.7 % 48.4 % 55.5 %

    Juvenile Jurisdiction 47.9 46.5 51.7Adult Jurisdiction 56.1 53.2 57.3

    Ever Arrested Males 58.3 54.0 61.7Juvenile Jurisdiction 53.7 51.1 60.1Adult Jurisdiction 61.1 59.2 62.2

    Ever Incarcerated Males 70.0 68.9 71.5Juvenile Jurisdiction 63.6 62.0 64.9Adult Jurisdiction 72.4 78.1 72.8

    Males Who Ever Stole a Car 58.7 55.6 65.4Juvenile Jurisdiction 54.5 55.6 63.5Adult Jurisdiction 61.7 55.6 65.9

  • 8/3/2019 Punishment and Deterrence of Crime

    27/37

    27

    Table 3. The Determinants of Perceived Chance of Jail for Males

    (1) (2) (3) (4) (5)

    Age 1.227** 1.195** 1.123** 1.100** 1.108**(0.196) (0.203) (0.203) (0.204) (0.205)

    Adult_Jur 5.978** 6.051** 6.013** 6.017** 5.978**(0.797) (0.798) (0.798) (0.801) (0.803)

    any_stlls50 -0.568 -0.830 -0.775 -0.848(0.826) (0.828) (0.829) (0.832)

    any_stlmor50 -0.425 -1.643 -1.566 -1.464(1.229) (1.246) (1.247) (1.251)

    any_assault 0.786 0.516 0.412 0.382

    (0.843) (0.846) (0.848) (0.851)

    any_sdrug -1.602 -2.204* -2.149* -2.122(1.081) (1.089) (1.091) (1.095)

    any_arrest 1.756 1.834 1.801(1.516) (1.519) (1.525)

    any_charge 1.180 1.135 1.138(2.113) (2.117) (2.126)

    any_conv 4.587* 4.702* 4.552*(2.204) (2.209) (2.216)

    any_incar 3.554 3.614 3.802(2.625) (2.629) (2.638)

    msa 0.331(1.843)

    live_2parents 0.150

    (0.951)

    Individual Fixed Effects YES YES YES YES YESState Fixed Effects NO NO NO YES YESObservations 20205 20160 20126 20126 20043# of individuals 4589 4589 4589 4589 4589Standard errors in parentheses. * significant at 5%; ** significant at 1%. Individual fixed effects are included in allspecifications.

  • 8/3/2019 Punishment and Deterrence of Crime

    28/37

    Table 4. Robustness Check Is There a Discontinuity in Perceived Chance of Jail at False Age Cutoffs?

    (1) (2) (3) (4) (5) (6)

    States in which the Age of Criminal Majority is 17States in which the Ag

    Majority is

    Age 1.671** 2.061** 2.214** 2.720** 0.956** 1.615** (0.405) (0.378) (0.365) (0.329) (0.260) (0.289)

    Age >= 16? 2.961(1.525)

    Age >= 17? 4.542** 2.142*(1.521) (1.076)

    Age >= 18? 2.161 6.459**(1.471) (1.037)

    Age >= 19? -1.176 (1.519)

    Observations 5977 5977 5977 5977 11765 11765

    Number ofIndividuals

    1451 1451 1451 1451 2756 2756

    Standard errors in parentheses. * significant at 5%; ** significant at 1%The dependent variable in each specification is the perceived chance of a jail if arrested for stealing a car. All specificationsfull set of varying individual controls, and state fixed effects. For states in which the age of criminal majority is 17, Age Likewise, for states where the age of criminal majority is 18, Age >= 18? is equivalent toAdult_Jur.

  • 8/3/2019 Punishment and Deterrence of Crime

    29/37

    29

    Table 5. Heterogeneity in the Effect of Age of Criminal Majority on Perceived Punishment Severity

    Row SampleCoefficientonAdult_Jur

    StandardError

    Number ofIndividuals

    Number ofObservations

    (1) All Males (Baseline) 5.978** 0.803 4589 20043

    Heterogeneity by Criminal Experience

    (2) Arrested Prior to Age of Criminal Majority 3.732 2.096 817 3568

    (3) Not Arrested Prior to Age of Criminal Majority 6.430** 0.906 3543 15564

    (4) Report Crime Prior to Age of Criminal Majority 5.621** 1.005 2896 12872

    (5)Dont Report Crime Prior to Age of CriminalMajority

    7.042** 1.455 1464 6260

    Heterogeneity by Initial Perceptions

    (6)Perceived Chance Jail Less than 50% in Year

    Prior to Age of Majority36.771** 1.399 1264 5806

    (7)Perceived Chance Jail Equal to 50% in YearPrior to Age of Majority

    6.114** 1.728 667 3055

    (8)Perceived Chance Jail Greater than 50% in YearPrior to Age of Majority

    -24.152** 1.386 1278 5937

    Heterogeneity by State Age of Criminal Majority

    (9) States with Age of Criminal Majority = 18 6.419** 1.040 2755 11722

    (10) States with Age of Criminal Majority = 17 4.426** 1.529 1451 5940

    (11) States with Age of Criminal Majority = 16 8.730** 2.450 601 2381

    Heterogeneity by Relative Punitiveness of Juvenile to Adult Systems

    (12)Low (Quartile 1) ratio of adult to juvenileincarceration rates

    5.877** 1.538 1335 5492

    (13)Moderate (Quartile 2) ratio of adult to juvenileincarceration rates

    6.999** 1.787 1074 4273

    (14)Medium (Quartile 3) ratio of adult to juvenileincarceration rates

    4.687** 1.707 1071 4394

    (15) High (Quartile 4) ratio of adult to juvenileincarceration rates

    6.384** 1.544 1383 5634

    All specifications included the full set of controls and individual fixed effects. The dependent variable is perceived chanceof jail if arrested for stealing a car. * significant at 5%; ** significant at 1%. Relative punitiveness of juvenile to adult justice systems is measured as the ratio of adult incarceration rates to juvenile incarceration rates. The 25th percentileratio is 1.66, the median ratio is 2.47, and the 75th percentile ratio is 3.22. Note that the higher the ratio, the moredrastically punishment differs between the juvenile and adult systems.

  • 8/3/2019 Punishment and Deterrence of Crime

    30/37

    30

    Table 6. Do Changes in the Perceived Chance of Jail Around the Age of Criminal Majority DeterAuto Thefts?

    (1) (2) (3) (4) (5) (6) (7)

    PerceivedChance of Jail -0.001** -0.001* -0.001* -0.001* -0.001* -0.001* -0.000

    (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

    Age -0.000 -0.000 -0.000 -0.001 -0.000 -0.002**(0.001) (0.001) (0.001) (0.001) (0.001) (0.001)

    Any_assault 0.001 0.001 0.000 -0.001 -0.001(0.003) (0.003) (0.003) (0.003) (0.003)

    Any_stlmor50 -0.017** -0.017** -0.017** -0.017** -0.016**(0.004) (0.004) (0.005) (0.005) (0.004)

    Any_Sell

    Drugs -0.001 -0.001 -0.001 -0.001 0.002(0.004) (0.004) (0.004) (0.004) (0.004)

    Any_arrest 0.007 0.007 0.008 0.005(0.006) (0.006) (0.006) (0.005)

    Any_charge -0.010 -0.010 -0.011 -0.011(0.008) (0.008) (0.008) (0.008)

    Any_conv 0.028** 0.028** 0.028** 0.021*(0.009) (0.009) (0.009) (0.008)

    Any_incar -0.037** -0.036** -0.030** -0.034**(0.011) (0.011) (0.011) (0.010)

    MSA 0.007 0.008(0.008) (0.007)

    Live With 2Parents

    0.002 0.002

    (0.004) (0.004)

    Enrolled 0.003 0.005(0.003) (0.003)

    State FE NO NO NO NO YES YES YES2SLS YES YES YES YES YES YES NOObservations 15361 15361 15326 15296 15296 15182 15182Number ofindividuals

    4401 4401 4400 4397 4397 4395 4395

    Standard errors in parentheses. * significant at 5%; ** significant at 1%. The sample is restricted tomales. All specifications except for the last column instrument for the perceived chance of jail withwhether the individual has reached the age of criminal majority. The dependent variable in eachspecification is whether the individual self-reports committing an auto theft in period t+1.

  • 8/3/2019 Punishment and Deterrence of Crime

    31/37

    31

    Table 7. Do Changes in the Perceived Chance of Jail Around the Age of Criminal Majority DeterThefts of Items of More than $50, Drug Sales, and Assaults?

    (1) (2) (3) (4) (5) (6)Dependent Variable: Steal > $50 Drug Sales Assaults

    Perceived Chance of Jail -0.002* -0.000* -0.003** -0.000 0.000 -0.000

    (0.001) (0.000) (0.001) (0.000) (0.001) (0.000)

    Age -0.004 -0.008** 0.011** 0.004** -0.020** -0.020**(0.003) (0.001) (0.003) (0.002) (0.003) (0.002)

    Any_assault 0.010 0.010 0.016* 0.017* -0.162** -0.162**(0.007) (0.006) (0.008) (0.007) (0.009) (0.009)

    Any_stlmor50 -0.166** -0.163** 0.060** 0.064** 0.041** 0.041**(0.009) (0.009) (0.012) (0.010) (0.012) (0.012)

    Any_Sell Drugs 0.029** 0.034** -0.161** -0.153** -0.005 -0.005

    (0.009) (0.008) (0.011) (0.010) (0.012) (0.011)Any_arrest 0.013 0.008 0.036* 0.029* 0.003 0.003

    (0.012) (0.011) (0.015) (0.013) (0.016) (0.016)

    Any_charge -0.023 -0.024 -0.025 -0.027 0.015 0.015(0.017) (0.016) (0.021) (0.019) (0.023) (0.023)

    Any_conv 0.061** 0.047** 0.030 0.010 0.024 0.025(0.019) (0.017) (0.024) (0.021) (0.026) (0.024)

    Any_incar -0.039 -0.047* -0.007 -0.019 -0.073* -0.073*(0.022) (0.020) (0.027) (0.025) (0.029) (0.029)

    MSA 0.010 0.010 -0.001 -0.000 0.038 0.038(0.016) (0.015) (0.020) (0.018) (0.022) (0.022)

    Live With 2 Parents -0.005 -0.005 -0.003 -0.002 0.000 0.000(0.008) (0.007) (0.010) (0.009) (0.011) (0.011)

    Enrolled 0.009 0.013* 0.027** 0.032** 0.024** 0.024**(0.007) (0.006) (0.009) (0.008) (0.009) (0.009)

    State FE YES YES YES YES YES YES2SLS YES NO YES NO YES NOObservations 15115 15115 15107 15107 15118 15118

    Number of individuals 4386 4386 4384 4384 4386 4386Standard errors in parentheses. * significant at 5%; ** significant at 1%. The sample is restricted to males. Oddnumbered columns instrument for the perceived chance of jail with whether the individual has reached the age of criminalmajority. Even numbered columns are estimated using OLS. The dependent variable in each specification is listed at thetop of the relevant column and is measured as of period t+1.

  • 8/3/2019 Punishment and Deterrence of Crime

    32/37

    32

    Table 8. Heterogeneity: Coefficient on Perceived Chance of Jail for Selected SubsamplesCharacterizing Criminal Experience

    (1) (2) (3) (4)

    Dependent Variable =AutoTheft

    Steal >$50

    DrugSales Assault # obs.

    Baseline Entire Sample -0.001* -0.002* -0.003** 0.000 15,361(0.000) (0.001) (0.001) (0.001)

    Arrested Prior to Age of CriminalMajority -0.016 -0.025 -0.028 0.002 2,695

    (0.021) (0.028) (0.032) (0.013)

    Not Arrested Prior to Age of CriminalMajority -0.001* -0.002* -0.002 -0.001 11,823

    (0.000) (0.001) (0.001) (0.001)

    Report Crime Prior to Age of CriminalMajority -0.002* -0.004* -0.006** 0.000 9,790

    (0.001) (0.002) (0.002) (0.002)

    Dont Report Crime Prior to Age ofCriminal Majority -0.000 0.001 -0.000 -0.001 4,728

    (0.000) (0.001) (0.001) (0.002)

    Standard errors in parentheses. * significant at 5%; ** significant at 1%. The sample is restricted to males. Each cellpresents the coefficient that corresponds to the perceived chance of jail variable in separate regressions. In eachspecification, the perceived chance of jail is instrumented for with whether the individual has reached the age of criminalmajority. The dependent variable in each specification is listed at the top of the relevant column and is measured as of

    period t+1. The full set of controls, including individual fixed effects, are included in each specification. The regressionsonly differ by the sub-sample included and the dependent variable.

  • 8/3/2019 Punishment and Deterrence of Crime

    33/37

    33

    Table 9. Is Evidence of Deterrence Observed When Looking at Arrest Rather than Crime?

    (1) (2) (3) (4) (5) (6) (7)

    PerceivedChance of Jail -0.001 -0.001 -0.001 -0.001 -0.001 -0.000 -0.000

    (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.000)

    Age 0.002 0.002 0.003 0.002 0.003 0.003(0.003) (0.003) (0.003) (0.003) (0.003) (0.002)

    Any_assault 0.010 0.022** 0.022** 0.022** 0.022**(0.008) (0.008) (0.008) (0.008) (0.008)

    Any_stlmor50 0.005 0.036** 0.035** 0.038** 0.039**(0.011) (0.012) (0.012) (0.012) (0.011)

    Any_Sell Drugs 0.008 0.028* 0.028* 0.027* 0.027*(0.011) (0.011) (0.011) (0.011) (0.011)

    Any_arrest -0.167** -0.169** -0.173** -0.173**(0.015) (0.015) (0.015) (0.015)

    Any_charge 0.015 0.015 0.024 0.023(0.021) (0.021) (0.021) (0.021)

    Any_conv 0.012 0.012 0.003 0.001(0.024) (0.024) (0.024) (0.023)

    Any_incar -0.108** -0.111** -0.108** -0.109**(0.027) (0.027) (0.027) (0.027)

    MSA -0.002 -0.002(0.020) (0.020)

    Live With 2Parents

    0.013 0.013

    (0.010) (0.010)

    Enrolled 0.014 0.015(0.009) (0.008)

    State FE NO NO NO NO YES YES YES2SLS YES YES YES YES YES YES NOObservations 15295 15295 15261 15231 15231 15118 15118

    Number ofindividuals

    4394 4394 4392 4389 4389 4386 4386

    Standard errors in parentheses. * significant at 5%; ** significant at 1%. The sample is restricted to males. Allspecifications except for the last column instrument for the perceived chance of jail with whether the individual hasreached the age of criminal majority. The dependent variable in each specification is whether the individual self-reportsbeing arrested in period t+1.

  • 8/3/2019 Punishment and Deterrence of Crime

    34/37

    Table 10. Can Deterrence be Distinguished from Incapacitation?

    (1) (2) (3) (4) (5) (6) (7) (8) (9)

    Auto Theft Steal > $50 Drug Sales

    PerceivedChance of Jail -0.001* -0.001* -0.001* -0.002* -0.002* -0.002* -0.003** -0.003** -0.003**

    (0.000) (0.000) (0.000) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)

    Any_Incar t -0.030** -0.019 0.017 -0.039 -0.025 0.029 -0.007 0.018 0.044 (0.011) (0.012) (0.012) (0.022) (0.024) (0.024) (0.027) (0.030) (0.030)

    Incarcerated atTime ofInterview -0.040* -0.043* -0.050 -0.054 -0.088* -0.089*

    (0.017) (0.017) (0.036) (0.035) (0.044) (0.044)

    Any_Incar t + 1 0.139** 0.213** 0.105** (0.008) (0.016) (0.020)

    State FE YES YES YES YES YES YES YES YES YES 2SLS YES YES YES YES YES YES YES YES YES Observations 15182 15182 15182 15115 15115 15115 15107 15107 15107 # Individuals 4395 4395 4395 4386 4386 4386 4384 4384 4384

    Standard errors in parentheses. * significant at 5%; ** significant at 1%. The analysis is restricted to the sample of males. Eaincludes the full set of controls as well as individual and state fixed effects. The dependent variable in each specification is lisand is measured as of period t+1.

  • 8/3/2019 Punishment and Deterrence of Crime

    35/37

    35

    Figure 1. Distribution of Beliefs about the Chance of Going to Jail when Arrested for Stealing aCar

    0

    .05

    .1

    .15

    .2

    Fraction

    0 10 20 30 40 50 60 70 80 90 100

    % Chance of Jail

    Note Figure is based on observations from all survey rounds.

  • 8/3/2019 Punishment and Deterrence of Crime

    36/37

    36

    Figure 2. Perceived Chance of Jail If Arrested for Stealing a Car Relative to the Age of CriminalMajority

    0

    5

    10

    15

    20

    25

    30

    35

    40

    45

    50

    55

    60

    %c

    hanceofjailwhenstealca

    r

    -5 -4 -3 -2 -1 0 1 2 3 4

    Relative Age

    Entire Sample Males

    Note - Relative Age of 0 indicates first year individual is in adult system

  • 8/3/2019 Punishment and Deterrence of Crime

    37/37

    Figure 3. Males Perceived Chance of Arrest for Stealing a Car Relative to the Age of CriminalMajority

    0

    5

    10

    15

    20

    25

    30

    35

    40

    45

    5055

    60

    %c

    hanceofarrestwhenstealcar

    -5 -4 -3 -2 -1 0 1 2 3 4

    Relative Age

    Note - Relative Age of 0 indicates first year individual is in adult system


Recommended