Time Series and Panel Data Analysis
6.6 Panel Data Study
6.6.2 Results and Discussion
The result of the analysis using panel data estimation across the districts and over the given period of time employing the fixed effects model is placed below in Table 6.3
Table 6.3 Panel study results
Dependent variable-lcaseper1,00,000population
Coef. Std. Err. t P>|t|
lprisonpopulation 0.016 0.040 0.41 0.683 lchargesheetrate -0.018 0.012 -1.54 0.125
lcaseperi/o 0.657 0.031 21.44 0.000
ldropout rate -0.004 0.014 -0.26 0.792
lpercapitaddp 0.801 0.091 8.78 0.000
lvehicleperi/o -0.221 0.039 -5.69 0.000
cons -4.79 0.889 -5.39 0.000
Source: Author‘s calculation using STATA
Thus, from the results of the fixed effect model in Table 6.3, it can be seen that the coefficients PCDDP and some deterrence variables are statistically significant. It is expected that dropout rate would significantly affect criminal activities. But it is found that this variable is statistically not significant in the present study. In this context, it is worthwhile to note that the data structure of dropout rate in this study comprises of only those who dropped out before completing the lower primary level. It is obvious that this stratum of dropouts would constitute a very miniscule quantum of people without sufficient skill to compete in the legitimate job market. It would be more meaningful to use a broad based data of dropouts taking into account the individuals who failed to complete higher classes like upper primary to make it more inclusive in its coverage. However, due to lack of such detailed data, disaggregated up to district level, the present analysis had to be restricted to the lower primary dropout rate only. Again, the estimated coefficient of the per capita income variable in the district level indicates a positive influence of economic prosperity on crime level. Thus, one can see that when the district fixed effects are controlled through the fixed effect model of panel data analysis, the net effect of income variable on crime becomes positive and this is consistent with the results of panel data study by Dutta and Hussain(2009).
As expected, the coefficient of deterrence variable in terms of vehicles per I/O is negative. This means that the more the logistical support in terms of increased mobility the less will be crime committed. This would provide mobility support for the police organisation which would increase the probability of arrest and allow for efficient and prompt investigation. Logistical support has been proxied by the availability of vehicles per
investigating officer since this would mean effective deterrence over time and space. More mobility through increased availability of vehicles per investigating officer would mean more visibility of police personnel and more arrests. This would put the criminals at greater risk leading to increased expectation of higher costs of doing offences. As a consequence there would be desistence or specific deterrence from indulging in further criminal activities. Even this would lead to a deterrence effect on the potential criminals. Two important ways of crime control have been in terms of patrolling the crime prone areas and investigating the offence so as to bring the criminals to the judicial authorities for punishment. Patrolling is thus considered an effective strategy for crime control by preventing it as well as interrupting during progress of crime commission. Moreover, it creates an atmosphere of confidence amongst the people. In fact poor patrolling is despised by the local people.
Choklingam(2003) in his victimization survey of crimes in four major cities in south India had referred to the benefit of patrolling by the police in deterrence. In a terrain like the hill districts of Assam or in the villages or suburban areas where the population density is thin, through foot patrolling less purpose would be served due to the distances involved. In such spatial characteristics one would like to go for police patrol on vehicles to bring more mobility and distance coverage. Though such patrol strategy would lessen the scope of citizen contact and visibility, it would be more effective because of the speed and surprise involved in it. A patrol car is likely to deter crime and also it gives feeling to the citizens about police being present. In a state like Assam where there are terrorists threat due to the presence of rebel and insurgent groups, it is required that police is given more vehicle to cover more geographical areas through patrol and take resort to proactive policing like security checks, visiting the suspects place etc. Also vehicles would help police in quick and effective investigation processes like searching the places of suspected criminals taking interrogation statements etc. It is also important that police on vehicles control public order in time through quick mobility. The estimated crime rate function for Malaysia demonstrated that increases in the numbers of police and patrolling activities in the potential crime areas would decrease the crime in Malaysia (Tang,2011).However, in some other studies the police patrolling has been found to have no effect in crime prevention (Philipson and Posner,1996).
The deterrence variable viz. cases per I/O is showing a positive sign. This means that more the number of cases per investigating officer the more is the number of crimes committed. This is because the civil police will not be in a position to pay more attention to the investigation, thus, less probability of the criminal being arrested and
convicted of the crime committed. If more cases are to be investigated by an investigating officer there would be dilution in the effort in apprehension of offenders and investigation of the cases since there would be more pressure on the available resources in the police stations, thereby affecting the result and timely disposal of cases which in turn would affect positively on the criminal activities. All these mean that one would expect a positive relationship between crimes and cases per policeman. Lesser work load in terms of fewer IPC cases per civil policeman are seen to be associated with lower crime rates (Dutta and Hussain, 2009).
The negative sign of charge sheet rate is as expected but it is not significant.
The reason for such an observation may be because of the low conviction rates experienced in the state. As a result, although charge sheets do act as a deterrent, the failure to convict a criminal does not increase the real costs to the criminals in the long run. It is also to be noted that the positive sign of variable like prison population in the estimated equation is not found to be significant. This is almost similar to the finding in the Time series analysis presented in this chapter earlier where arrest variable had such a sign. It may be mentioned that crime rates are found to be positively related to lagged arrest and conviction rates in India (Dutta and Hussain ,2009). Corruption and malpractices prevailing in most Indian jails making it difficult for most inmates and convicts to change and reform their criminal behaviour has been cited as one reason. In fact, people convicted or sent to judicial custody over minor offences often end up being hardened criminals after getting exposed to the corrupt and criminal jail environment (Piehl and DiIulio,1995). Furthermore, arrests and convictions make it increasingly difficult for that individual to get access to legitimate means of livelihood, thereby incentivising criminal behaviour in many cases (Grogger, 1995, Holzer et al, 2003, Pager, 2003, Seiter and Kadela, 2003). In other words, there seems to be a case of a
‗crime trap‘, where it is the same set of criminals undertaking criminal activities, with even better precision and efficiency. NCRB data clearly show such a trend of recidivism, with nearly 10% of the arrested persons having previous criminal records. Thus the conclusions found in the time series and panel data analysis of the present study are closely related to what other researchers have found.