COMMENTARY
Analysis of Juvenil Juvenile e Crime Effects of State State Apparatus Malvika Tyagi
T
Crime committed by juveniles in he Apprentices Act, 1850 was the very first legislation pertaining to India and the law pertaining to children which provided for the it has been the subject of debate rehabilitation of those in the age group and concern, primarily because of 10–18 years convicted for an offence. of a perceived rise of vviolent iolent This was followed by the Reformatory Schools Act, 1897, and the Madras Chilcrime cri me and the proposal (and (and dren Act, 1920, Bengal Children Act, 1922 eventual amendment ame ndment in 2015) and Bombay Children Act, 1924. of treating juvenile offenders The Children Act, 1960 was the first involved in heinous crimes as central enactment in independent India relating to children which followed the adults. adults. This Thi s evokes an enquiry United Nations Congress on the Preveninto the various aspects of tion of Crime and Treatment of Offendcrime committed by juveniles, juveniles, ers. It applied only to the union territosuch as the trend of such crime ries, since the subject matter of juvenile justice fell under the state list of the over time, the socio-economic socio-economic Indian Constitution. characteristics of juvenile The United Nations Standard Minimum offenders, and the role of the state Rules for the Administration of Juvenile in dealing with it. Justice were an opportunity as well as a
Malvika Tyagi (malvika.tyagi@gmail. (malvika.tyagi@gmail. com) com) is a doctoral student of Economics at the Department of Humanities and Social Sciences, Indian Inst itute of Technology, New Delhi. Economic & Political Weekly
EPW
DECEM BER 17, 2016
compulsion for Indian Parliament to bring a uniform juvenile justice legislation for the entire country (except for the state of Jammu and Kashmir). Hence, the Juvenile Justice Act came into being. It defined a juvenile as a boy as not having attained atta ined the age of 16 years and a girl as not having attained the age of 18 years. A delinquent juvenile juvenile was defined as a juvenile who was alleged to have committed an offence. Even though the Government of India had ratified the United Nations ConvenConvention on the Rights of the Child Chi ld in 1992, it felt the need to re-enact the existi ng law to bring in conformity this particular, as well as all other international instruments. This led to the enactment of the Juvenile Justice (Care and Protect ion of Children) Act, 2000. This act increased the age of majority f rom 16 to 18 (in the case of boys), boys), a decision which whic h has been partially reversed in 2015 by reducing the age of majority from 18 to 16 in cases of heinous crime. Although the issue of juvenile crime and the efficacy of t he state in curbing it vol lI no 51
has been debated in the last couple of years, there has not been, so far, a quantitative analysis in the Indian context to bring out the relation between the two, keeping in mind the other othe r plausible factors that lie at the root of this socio-economic problem. This is what this article attempts to do. Figures 1 and 2 (p 18) show the trend of juvenile crime over the last decade. Juvenile crime has only slightly increased, both in absolute terms and as a percentage of total crime, after the Juvenile Justice (Care and Protection of Children) Act, 2000 came into being. This means that the slight increase was wa s on account of a change in the definition of juvenile crime (as now even those above the age of 16 were considered juveniles). But, the slight upward trend even many years after the act means that juvenile crime has increased generally over the last decade. This is further seen in Figure 3 (p 18), which shows the continuously increasing proportion of arrested juveniles over the age of 16. Looking at the socio-economic characteristics of delinquents, it is seen that mostly the school dropouts and ones belonging to the poorest of families comprise the highest proportion of those committing crime (Figures 4 and 5 [p 19], respectively).
Methodology While panel data has been assimilated for assessing the response of juvenile crime to the state apparatus, the main purpose of using state-level panel data instead of all-India time series data, is to have a more robust result. While it would have been nice to have been able to use data for a longer time period and bring out any possible state/time-fixed effect impact, the lack of many control variables for a longer time period has impeded such an attempt. An ordinary least square (OLS) estimation1 of juvenile crime response to state apparatus 2 seems to be well suited, given the short time span and the minimal variation in the variables at hand. The number of crimes committed by juvenilecr imeit) has been de juveniles ( juvenilecrime fined as a fraction of juvenile population (in the age group 15–18 years) that has 17
COMMENTARY Figure 1: Number of Juvenile Delinquents
(in lakh)
2.5
2
1.5
1
0.5
0 1990
1992
1994
1996
1998
2000
2002
2004
200 6
2008
2 010
2012
Source: Constructed by author using data from the National Crime Records Bureau.
Figure 2: Percentage of Total Crime Committed by Juveniles 1.4
1.2
1
juvenilecrimeit = β 0 + β 1 convictedjuveniles i(t-1) + β 2 adutsarrestedit + β 3 police it + β 4 [Control Variable sit ] +uit
0.8
0.6
where i is the ith state, t stands for the year, and u is the error term.
0.4
0.2
Data Sources
0 1993 1994 1995 1996 1997 1998 1999 2000 20 01 2002 20 03 2004 20 05 2006 20 07 2008 2009 2010 2011 2012
Source: Same as Table 1.
Figure 3: Age Group-wise Percentage of Crime Committed by Juveniles 80 Juveniles apprehended, 16–18 70 60 50 Juveniles apprehended, 12–16 40 30 20 Juveniles apprehended, 7–12 10 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
2010 2011 2012
Source: Same as Table 1.
been apprehended for committing crime. 3 The age group-wise population data only contains the category 15–19 years. It is better to use “apprehended” instead of “convicted” as a large proportion of cases remain pending when, at the same time, the acquittal rate is very low. The state apparatus comprises three variables. The first is one-period lagged values (to rule out simultaneity as well as 18
This is more so because it is rare for juveniles to commit crime without an adult accomplice, as observed in Tyagi (2016). The third explanatory variable is the strength of police personnel as a proportion of the population of a state. It is important to control for other relevant variables, such as development indicators (Human Development Index [HDI], sex ratio, gross enrolment ratio, percentage of Scheduled Caste population, child labour), and others, such as power deficit, corruption, etc, which may have an effect on the juvenile crime rate. This is to ensure that the variation in juvenile crime rate is not the result of some unaccounted exogenous variables, especially as strong correlations are observed between them and juvenile crime. The following equation is estimated:
possible ratio bias) of the log of convicted juveniles (convicted because their incarceration time, unlike adults, is too low, so the demonstration effect is only expected to show up upon conviction). The second is adults arrested (not convicted, because disposal of adult cases is much slower than for juveniles, so the demonstration effect of adult punishment is captured well enough by the arrest).
Panel data comprising 13 states from 2003 to 2007 is compiled. The dependent variable is computed from the annual publications of the National Crime Records Bureau (NCRB). The three explanatory variables—juveniles convicted, adults arrested and strength of police personnel—are also obtained from the annual publications of the NCRB. Data for the control variables—income, youth unemployment, sex ratio and percentage of Scheduled Caste population—are obtained from the census and/or the National Sample Survey Office’s quinquennial reports. Data for gini coefficient, HDI, and child labour are obtained from Nayak et al (2010), Human Development Report 2009, and the Ministry of Labour and Employment, respectively, and interpolated. Data for corruption indicators, transmission and distribution losses, and power deficit (an indicator of the efficiency of a state) are obtained from Indiastat. 4
Results The statewise data shows that Maharashtra, Andhra Pradesh and Rajasthan have the highest juvenile crimes, followed
DECEM BER 17, 2016
vol lI no 51
EPW
Economic & Political Weekly
COMMENTARY Figure 4: Number of Juvenile Delinquents by Educational Status 16,000 Primary
Above primary but below matriculation
12,000
Illiterate
8,000 Matriculation or above 4,000
0 2001
2002
2003
2004
2005
20 06
2007
2008
2009
2010
2011
2012
2011
2012
2013
Source: Same as Table 1.
Figure 5: Number of Juvenile Delinquents by Income Groups 30,000 30000
25,000 25000
20,000 20000
15,000 15000
10000 10,000 5000 5,000 00 2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2013
Up to ` 25,000 Upto Rs.25000
25,001 to ` ` Rs.25,001 To50,000 Rs.50,000
50,001 toto ,00,000 ` ` 1Rs.1 Rs.50001 lakh
1,00,000 ` ` 2,00,000 Rs.1 lakh totoRs.2 lakh
2,00,000 to ` ` Rs.2 lakh to Rs.3,00,000 3 lakh
3,00,000 Above ` Rs. 3 lakh
Source: Same as Table 1.
Table 1: Linear Regression Dependent Variable: Juvenile crime Prob > F=0.0000; R-squared=0.82 Coefficient
Convicted juveniles(t–1)
Robust Standard Error
-0.00335***
.0018
0.00017** -12.54*
Adults arrested Police
t-statistic
P value
-1.85
0.071
.0 00
4.87
0.000
5.16
-2.43
0.019
Per capita income
6.88
.000
.56
0.579
Youth unemployment
.005*
.001
-3.43
0.002
Gini urban
4.741
3.446
1.38
0.177
Gini rural
13.401*
5.834
2.30
0.027
.189
2.33
0.25
HDI
.440
Corruption Child labour Sex ratio
.013
.009
1.29
0.204
4.261
4.091
-0.10
0.918
.040
.036
1.11
0.272
Note: Number of observations = 64. * Significant at the 5% level; ** Signific ant at the 1% level; *** Significant at the 10% level
by Gujarat and Haryana. Kerala, Punjab and West Bengal have some of the lowest recorded juvenile crimes. It is seen that a higher gini coefficient (rural), high incidence of child labour and a high corruption indicator (using the conventional transmission and distribution loss as proxy) seem to correlate with increased juvenile crime, and negatively correlate with a higher gross enrolment ratio, sex ratio and per capita
income of a state. If taken together, these variables continue to have the depicted effects on the dependent variable, although only a couple of them (sex ratio and gini–rural) remain significant. I chose to use them only as control rather tha n as explanatory variables because they involve the exercise of interpolation due to lack of data (although, since the time period is short, and these variables are slow
Economic & Political Weekly
vol lI no 51
EPW
DECEM BER 17, 2016
changing by nature, this should not be of much concer n). As can be seen in Table 1, convicted ju veniles and police strength have a discouraging (negative) effect of 0.003% and 12% respectively, while adult arrests have a positive effect (all variables are in logs). The discouraging effect of past convictions may not entirely be indicative of deterrence, and may partially be a result of the incarceration effect, that is, those who are prone to recidivism are now confined in juvenile homes, unable to commit crime. The relatively large role of the police in curbing crime is suggestive of the need to increase police visibility. The fact that adult crime might be influencing juvenile crime, as indicated by a positive relation between the two, however little the influence, is a disturbing feature and is also borne out in my ongoing fieldwork. An increase in gini coefficient (rural) and youth unemployment seem to increase juvenile crime, while it seems to be unrelated to the other control var iables, but that is not surprising given t he short time span studied in this case.
Concluding Remarks Although reported juvenile crime rate in India is not very high, it has been on the rise over the last decade, with those 16 years or more in age comprising well over half of the delinquents in total. The largest proportion of recent juvenile offenders is of those who have finished primary school, and dropped out after that, reaffirming the observations of Tyagi (2016). Until a few years back, the largest proportion of juvenile offenders was of those who were in primary school, which is in line w ith the “concentration aspect” of schooling (Jacob and Lars 2003), wherein the increase in potentially “volatile interactions” as a result of geographical concentration overshadows the “incapacitation effect” of school (engaging in constructive activities). The fact that an overwhelmingly large proportion of offenders belong to the poorest of families speaks volumes of their vulnerability in falling prey to crime, and debunks the dismissal of delinquents as merely psychologically disturbed, or depraved beings, without any socio-economic disability. 19
COMMENTARY
Some socio-economic indicators used as controls in this article, such as youth unemployment and gini coefficient, have a statistically significant positive impact on juvenile crime, but their interpolated nature and the short time span in this particular study gives only a weak result. There seems to be some evidence of the potential of the state apparatus’s deterrence on juvenile crime, but with caveats. Given that police density turns out to be important in curbing crime, the fact that both geographically as well as in terms of per capita, it is rat her low in most states, is worrying. Even though conviction of juveniles seems to curb crime (although minimally), it could be the case that a good proportion of juvenile offenders are prone to recidi vism, which means that the fall in crime is merely an affect of incarceration, as habitual offenders are held up in juvenile
20
homes. Moreover, there is the question of imperfect information. If potential delinquents are not well-informed of the consequences of their actions, or act on impulse, or are emboldened by virtue of being in the company of adults, then the extent to which conviction can curb future crime is small. The positive relation between the number of adult and juvenile offenders brings out the “schooling aspect” of cri me, wherein it is in the company of potential adult offenders or even repeat offenders that juveniles may be engaging in crime, as also reflected in my earlier article (Tyagi 2016). Notes 1
Since the nature of the data is that of a panel, a natural question that arises is why the fixed effects or the random effects technique is not used. This is because fi xed effects will not work well with data for which the withi n-cluster variat ion is minimal, or for slow-chang ing variables over time, which is the case with this data. The rationale behind the random effects
model is that, unlike the fixed effects model, the variation across entities is assumed to be random and uncorrelated with the predictor or independent variables included in the model, which is unlikely to be the case here. 2 Inspire d from Steven D Levitt (1998), who uses a similar methodology to measure the impact of punishment on juvenile crime. 3 Levitt (1998) also uses another definition of ju venile crime, the fract ion of total arre sts that comprise of juveniles, multiplied by total cognisable crime. But, as he himself admits, this might lead to a simultaneity and a ratio bias, a problem I want to avoid. 4 http://www.indiastat.com.
References Jacob, B and L Lars (2003): “Are Idle Hands the Devil’s Workshop? Incapacitation, Concentration, and Juvenile Crime,” Americ an Economi c Review, Vol 93, pp 1560–77. Levitt, Steven D (1998): “Juvenile Crime and Punishment,” Journal of Political Economy, Vol 106, No 6, pp 1156–85. Nayak, P K, S K Chattopadhyay, A V Kumar and V Dhanya (2010): “Inclusive Growth and Its Regional Dimension,” Occasional Papers, Reserve Ba nk of India, Vol 31, No 3. Tyagi, M (2016): “Understanding Juvenile Crime: Notes from the Field,” Economic & Political Weekly , Vol 51, No 2, pp 23–25.
DECEM BER 17, 2016
vol lI no 51
EPW
Economic & Political Weekly