Juvenile crime and punishment: evidence from Japan

Abstract

Over the last decade, juvenile crime has become a serious social problem in Japan. The Juvenile Law was revised in 2001 to impose harsher punishment on juvenile offenders. This revision makes it possible to impose criminal punishment on 14- and 15-year-old criminal offenders, while those offenders aged 16–19 have always faced criminal punishment, both before and after the revision. Using this revision as a natural experiment, this study conducts a difference-in-differences estimation to examine the effect of punishment on juvenile crime. The analysis provides evidence that punishment can deter juvenile crime. In addition, this research examines the criminal behaviour of 13-year-olds, who face no change in punishment, but who soon will in the near future. The results suggest that the revision also had a negative impact on the criminal behaviour of these younger offenders.

Acknowledgements

I would like to thank Charles Y. Horioka, Daiji Kawaguchi, Fumio Ohtake and especially Kazuo Ogawa for their helpful comment and suggestions. I am especially grateful to Kazuo Ogawa. All remaining errors are, of course, my own.

Notes

¹ The number of juvenile arrests did, however, remain constant, due to a decrease in the number of children in the population. Nevertheless, the rate of juvenile arrests clearly increased during the 1990s.

² For example, some studies have yielded the result that criminals respond to various criminal sanctions: arrest rates (Levitt, 1998b), conviction rates, sentence length (Mustard, 2003), numbers of police officers (Levitt, 1997) and levels of surveillance by police officers (Tella and Schargrodsky, 2004). Other research has suggested that economic conditions, such as employment (Carmichael and Ward, 2000; Raphael and Winter-Ember, 2001), wage levels, unemployment (Gould et al ., 2002) and inequality (Kelly, 2000) affect participation in criminal activity. Freeman (1999) surveyed a great deal of research in this area.

³ In particular, there has been little empirical research focusing on the effect of punishment on juvenile crime. Indeed, Levitt (1998a) has stated that, ‘To the best of my knowledge, however, there are no previous systematic empirical analyses of the response of juvenile crime to sanctions.’ (p. 1159).

⁴ Many researchers have increasingly paid attention to social interaction (or peer effects), and some studies have demonstrated the importance of social interaction for youth activities. For instance, Case and Katz (1991) and Leung (2004) found that family and peer influences affect juvenile activities. In addition, Sacerdote (2001) showed that peer effects among college roommates have an influence on grade point average. Similarly, Gaviria and Raphael (2001) obtained the result that social interaction is important at the school level with respect to juvenile behaviour, and Kawaguchi (2004) has provided a number of findings relating to the impact of social interaction on youths' smoking behaviour. Finally, Manski (1993, 2000) have pointed to the difficulties of identifying social interaction.

⁵ Criminal punishment has deterrence and incapacitation effects. Although both effects have negative effects on crime, the mechanism differs. While deterrence effects increase the expected cost of crime, incapacitation effects directly impose restrictions on criminal behaviours, like imprisonment. The law revision analyzed in this study provided a sudden change in punishment for adolescents in the treatment groups. Since incapacitation is generally slower than deterrence to reduce crime, the immediate reduction of juvenile crime, which this research has made an attempt to identify, can be interpreted as a deterrent effect.

⁶ In this analysis, I initially hypothesized that 13-year-olds, as a group, would be unaffected by the revision; however, the results that emerged, which are discussed subsequently, indicated that the opposite was true. To facilitate straightforward exposition, this article presents an analysis in which youths aged 13 and below are included in the treatment group.

⁷ Of course, the possibility that youths aged 13 and below had become more likely to commit crime should also be considered. The application of harsh punishment, following the revision, depends on the age at which juvenile offenders commit crimes, rather than the age at which they are arrested. Thus, if juvenile crime has no path dependency and is inter-temporally substitutable, then it may have been optimal for adolescents under the age of 13 to commit more crime before reaching the age of 14. As indicated in a later section, this article does not find evidence of an increase in the crime committed by the younger group.

⁸ Becker and Murphy (1988) argued that, ‘For example, a person may become permanently addicted to heroin or liquor as a result of peer pressures while a teenager, or of extraordinary stress while fighting in Vietnam.’ (p. 691). Finally, parents may attempt to deter their children's delinquencies in response to the revision of the Juvenile Law.

⁹ Profound changes in the way Japan dispenses criminal justice are either forthcoming or under consideration. Many people are ready to accept changes, even to welcome some of them.’ in The Japan Times (20 September 2000).

¹⁰ As Besley and Case (2000) has argued, analyses that rely on policy changes may suffer from an endogeneity of policy (simultaneity). Since most policy changes are conducted to solve certain problems, such changes often occur around the same time that problems occur. However, this simultaneity problem is likely to have little influence on the present research, for two reasons. First, the policy change in this case, which was the revision of the Juvenile Law, occurred in 2001, while the period during which juvenile crime became a serious problem in Japan occurred in the middle of the 1990s. Second, although serious offences triggered the revision of the Juvenile Law, less serious offences, in large part, comprise the data used in this article. Felonious offences (homicide, robbery, arson and rape) accounted for only about 1–2% of all juvenile crime.

¹¹ The amendment was divided into three main parts. First, it imposed harsh punishment on young offenders aged 14–19. Next, the procedure of trials was revised so as to clarify the facts of the cases. Before the revision, concern over the feelings of juvenile criminals had been given priority over that of conducting a thorough investigation into the matter. Finally, victims and their family gained easier access to the trials of suspects.

¹² In The Japan Times (1 November 2000), the subheading of the page discussing the Revised Juvenile Law was ‘Age of criminal liability lowered to 14.’

¹³ Specifically, two points were revised for offenders aged 16–19. One was that all juveniles aged 16 and above who were accused of intentionally killing their victims should, in principle, be sent to the public prosecutor instead of to Family Court. Prior to this change, the question of whether they were sent to the public prosecutor depended on the judgment of the Family Court. The other revised point required that the extent of commutation become more restricted. For example, a life sentence for a 16- or 17-year-old offender was not necessarily reduced to a definite term, while it had been so reduced under the old Juvenile Law (Iijima, 2001).

¹⁴ Heckman et al . (1999) pointed to the possibility that ignorance of indirect effects on both treatment and control groups might mislead researchers into obtaining biased estimates. The effect of social interaction is one such example: the outcome in the control groups may have been influenced by the outcome in the treatment group, or vice-versa. This research does not consider the indirect effects and this omission is undoubtedly a limitation of this study.

¹⁵ Lochner and Moretti (2004) pointed out that education may influence these two points.

¹⁶ Mocan et al . (2003) shed light on this possibility and constructed a model in which an individual considered the accumulation of two kinds of human capital, each of which was related to legal and illegal activities, respectively.

¹⁷ This article ignores cohort effects. One of the main reasons for this omission is that all cohorts were born in the same generation and were, at most, less than ten years apart. In addition, the unobserved heterogeneity among age groups in each prefecture is not controlled for, as inclusion of this heterogeneity would make estimation results unstable.

¹⁸ The Hausman test suggests that once control variables are included in the estimated equation the differences between the random-effects estimates and the fixed-effects estimates are statistically significant.

¹⁹ The figures are available from the author upon request.

²⁰ The data source was the Summary of Protection and Guidance of Youth (SPGY) (Shyonen No Hodo Oyobi Hogo No Gaikyo), which is published annually by the National Police Agency (NPA) in Japan. The most famous data on crimes in Japan are the Criminal Statistics (CS) (Hanzai Tokei Syo), published annually by the NPA. As far as juvenile crime is concerned, the information contained in the CS is less extensive than that in the SPGY. Thus, the SPGY played the primary role in this research.

²¹ At the time of writing, the latest data available were for 2002. The beginning of the sample period was taken as 1999, to avoid systematic differences, which are more likely to arise as the difference between the ‘before’ and ‘after’ periods increases.

²² Cherry and List (2002) has pointed out that research using aggregated data suffers from an aggregation bias, because deterrent effects have heterogeneous impacts across crime types. It is undeniable that the aggregation bias may undermine the estimates in this research, because data on crime types are unavailable. However, since the National Police Agency edits the data, the characterization of crimes by type depends on the interpretation of the police officers and may vary, according to time and circumstances. For instance, some police officers may have considered certain types of robbery to be larceny. Since the analysis using aggregated data avoids the problem of ambiguous sorting, it is not reasonable to argue that research using data on crime types would provide better results than that using aggregated data.

²³ Summary statistics for the population are not reported here for the sake of brevity, but are available from the author upon request.

²⁴ The ratio of daytime population to nighttime population was obtained from the Population Census, the number of population by age obtained from the Current Population Estimates, the number of police officers obtained from the Survey on Wages of Local Government Employees, the clearance rate obtained from the Criminal Statistics and the unemployment rate obtained from the Labour Force Survey. Wage was calculated using data from the Basic Survey on Wage Structure that is conducted annually by the Ministry of Labour. Work hours are broken down into regular work hours and overtime. The data on income consists of three parts: the regular wage, overtime payments and bonuses.

²⁵ Daytime population refers to nighttime population plus inflow of commuters and persons attending school from other place minus outflow of commuters and persons attending school to other place. Ratio of daytime population to nighttime population refers to daytime population per 100 nighttime population.

²⁶ Because of a lack of appropriate data, this research does not directly control for incapacitation effects, which are caused by the imprisonment of offenders. The Supreme Court annually publishes data on juvenile trials but does not report judgments for juvenile offenders by prefecture, and produces little detailed information according to age. Thus, incorporation of only the clearance rate controls for part of incapacitation effects.

²⁷ Wage was calculated from the Basic Survey on Wage Structure. The Basic Survey on Schools is the data source for both the ratio of job offers to applicants and the ratio of students to teachers.

²⁸ In Japan, youths in this age category are less likely to work, because most of them attend school. In fact, the advancement rate in high schools was between 96.9 and 97.0% during the period considered in this study.

²⁹ The 15-year-old youth cohort consisted of junior high school students and graduates, most of whom went on to high school. Thus, both kinds of variables related to school are used for 15-year-olds. In addition, almost all students in high school graduate from their schools when they are 18 years old. Thus, the variables related to school are not used for 19-year-olds. Although these data reflect job market conditions for new graduates of junior high school or high school, youths may have considered their future careers, and these therefore may have been affected.

³⁰ The estimates of the coefficients on wages do not reflect multicollinearity among the covariates. In fact, different specifications with regard to wages provide similar results.

³¹ Kalb and Williams (2003) took into account the gender difference in juvenile delinquency. They showed that the recidivism rate was higher among males than females.

³² Full results are available from the author upon request.

³³ Results are available from the author upon request.