Assignment 01

1. Introduction
The objective of this report is to analyze crime rates in different counties and to find the factors
that contribute to the crime rate in these countries. The data used in this report is the panel data
related to the crime of 90 observational units (counties) in the US from 1981 to 1987. To
understand the crime rate using the given data, the following empirical approach is suggested.

2. Empirical Approach
Since the given data is a panel, either a fixed effect or random effect model can be used (Porter et
al., 2019). Hausman test will be used to choose the random effect or fixed effect model. This test
is used to evaluate the consistency of the estimator, Moreover, it helps to evaluate the
correspondence of the statistical model to the data (Aït-Sahalia & Xiu, 2019). The difference
between the fixed effect and random will be examined by using the Hausman test. For this
purpose crimes committed per person for the years 1981 to 1987 in 90 counties in the US are
taken as the dependent variable and the independent variable includes the probability of arrest,
probability of conviction, probability of prison sentence, police per capita, and tax revenue per
capita.

To meet the assumption of the regression model such as homoscedasticity of the

residual, the log-log fixed effect regression model is used (Donthi et al., 2019).

The form regression equation is as follows.

log Y i=α + α 1 logX 1 i+ … α 5 logX 5i (1)

Where log Y i represents the log of the crimes committed per person for the years 1981 to 1987

logX 1i is the log of the probability of arrest.

logX 2i is the log of the probability of conviction.

logX 3i is the log of the probability of a prison sentence.

logX 4 i is the log of the police per capita.

logX 5i is the log of the tax revenue per capita.

3. Descriptive statistics and correlation analysis of the data

A brief description of the data is shown in the table below. The average value of the crime
committed per person is 0.032, whereas the average probability of arrest is 0.307, and the
probability of conviction is 0.689. This indicates that the Probability of conviction and prison
sentence is larger than the probability of arrest. The minimum value of the Crimes committed per
person is 0.002 while the maximum value is 0.164.

Table 1: Descriptive Statistics of the data

Crimes Prob. of arrest Prob. of Prob. of Police/ Tax

committed/ conviction the prison capita revenue/capita
person sentence
Mean 0.032 0.307 0.689 0.426 0.002 30.239
Standard Deviation 0.018 0.171 1.690 0.087 0.003 11.455
Minimum 0.002 0.059 0.068 0.149 0.000 14.303
Maximum 0.164 2.750 37.000 0.679 0.036 119.762

The results of the correlation of the log-transformed variables are shown in the below table. * is
indicating that the dependent variable Crimes committed per person is significantly correlated
with all independent variables. So, this specifies the strong association between the independent
and dependent variables.

Table 2: Correlation Analysis of the selected Variables

Crimes Prob. of Prob. of Prob. of Police/ Tax

committed/ arrest conviction a prison capita revenue/capita
person sentenc
e
Crimes committed/ 1.000
person
Prob. of arrest -0.4498* 1.000
Prob. of conviction -0.4569* -0.097 1.000
Prob. of a prison 0.1805* -0.021 -0.1581* 1.000
sentence
Police/ capita 0.1395* 0.053 0.2730* -0.056 1.000
Tax revenue/capita 0.1386* -0.049 0.024 -0.103* 0.2474* 1.000

4. Model Result
The result of the fixed effect and random effect models are shown in the table below.

Table 3: Fixed effect model and random effect model results

Fixed Effect Model Random Effect Model

Crimes committed/ Coef. Std. Err. T P>t Coef. Std. Err. T P>t
person
Prob. Of arrest -0.71 0.037 19.4 0 -0.713 0.037 -19.460 0.000
Prob. Of conviction -0.538 0.026 -20.5 0 -0.544 0.026 -20.730 0.000
Prob. Of prison 0.262 0.067 3.9 0 0.253 0.067 3.800 0.000
sentence
Police/ capita 0.338 0.031 10.9 0 0.344 0.031 11.130 0.000
Tax revenue/capita 0.182 0.06 3.1 0.002 0.111 0.051 2.180 0.029
_cons -3.07 0.318 -9.7 0 -2.815 0.293 -9.610 0.000

The null Hypothesis of the Hausman test is that the Random effects (RE) are better than Fixed
effects (FE). Obtain a chi-square value of 86.02 with a p-value of 0.00, indicating that the fixed
effect model is better. So the fixed effect model will be chosen.

Table 4: Hausman Test Result

Hausman Test
Chi2 (5) 86.02
Prob > Chi2 0.00

From the fixed effect model, it is found that the p-value of the Regression coefficient
is significant (less than 0.05) for each of the independent variables, indicating the
significant impact of the selected factor/ independent variable on the Crimes
committed per person. However positive sign is representing the positive impact and
the negative sign is representing the negative impact of the individual variable.
Table 5: Summary of the fixed effect model and random effect model

Random
Fixed Effects
Effects
Number of Observations 110 630
Number of Groups 17 7
R-Squared 0.571 0.57
P-value 0 0
F-Statistics / Wald Chi2 (5) 164.51 821

5. Justification of the Model

The result of the regression summary shows an R-squared value of 0.571, indicating
that 57.1% of the variations in the dependent variable are explained by independent
variables. Furthermore, to examine the reliability of the regression result, the
assumption of homoscedasticity is tested by using Breusch-Pagan / Cook-Weisberg
test. Two statistics are used chi-square and F-statistic are used. The null hypothesis of
the Constant variance is accepted for each statistic (p-value greater than 0.05) and it is
concluded that variance around the regression line is the same for all values of the
predictor variable

Table 6: Breusch-Pagan / Cook-Weisberg test result

Statistic Value P-value

Chi2(1) 3.64 0.0564
F(1 , 628) 1.53 0.2173

The result from the regression model shows that the probability of arrest and the probability of
conviction have a negative impact on the crimes committed per person. These findings suggest
reasonable results; it is common phenomenon that strict rules and regulations implemented by
police are always effective in reducing the crime rate. Physiologically people are afraid of
arresting or living under custody in any suspicious matter, this makes them not commit any
crime. Similarly, declaration of the jury of a judge declaring someone guilty of a criminal
offense (conviction) is also one of the positive factors in decreasing the crime rate (Doleac,
2020).

Moreover, the result of the study shows that the probability of a prison sentence has a positive
impact on the crimes committed per person. Imprisonment has a huge impact on the mental
health of the prisoners, which is due to the separation from loved ones and feelings of loneliness
as well as stress, delusions, paranoia, and depression which lead to a negative impact on mental
health (Haney, 2018). Consequently, this results in a decline in fear of the law, so prisoners
become more likely to commit crimes again. Similar to the probability of prison sentence, police
per capita, and tax revenue per capita also have a positive impact on the crimes committed per
person. High marginal tax rates can lead to a decrease in the saving rate and consequently is can
decrease the unemployment rate if serious attention is not paid. Increasing the unemployment
rate can potentially increase the crime rate (Mittal et al., 2019).

6. Implementation of the strategy

The finding of this economic approach for the crime data suggests many implementations
regarding the laws and regulations. Based on the result of this report, it is suggested that crime
rate must be reduce, it is important for the concerned authorities to follow the strict rules and
regulations related to citizen protection. As the model suggest that probability of a prison
sentence has a positive impact on the crimes committed per person, so, authorities must ensure
mental health and positivity among the prisoner. A sustainable judicial system in the country
must be shaped regarding the overall impact of the laws and regulations on society and
individuals. Police should make more arrests on every occasion or incident when a crime is
committed to reduce crime levels in the area. Moreover, more awareness programs or
volunteering must be done to educate the youth.
References

Porter, C. O., Outlaw, R., Gale, J. P., & Cho, T. S. (2019). The use of online panel data in
management research: A review and recommendations. Journal of Management, 45(1),
319-344.

Aït-Sahalia, Y., & Xiu, D. (2019). A Hausman test for the presence of market microstructure
noise in high frequency data. Journal of Econometrics, 211(1), 176-205.

Donthi, R., Praveen, J. P., Prasad, S. V., Mahaboob, B., & Venkateswarlu, B. (2019, December).
Numerical techniques of nonlinear regression model estimation. In AIP Conference
Proceedings (Vol. 2177, No. 1, p. 020082). AIP Publishing LLC.

Doleac, J. L. (2020). Encouraging desistance from crime. Available at SSRN 3825106.

Haney, C. (2018). The psychological effects of solitary confinement: A systematic

critique. Crime and Justice, 47(1), 365-416.

Mittal, M., Goyal, L. M., Sethi, J. K., & Hemanth, D. J. (2019). Monitoring the impact of
economic crisis on crime in India using machine learning. Computational
Economics, 53(4), 1467-1485.

Appendix
use "C:\Users\user\Downloads\[Link]"

summarize crmrte prbarr prbconv prbpris polpc taxpc

correlate lcrmrte lprbarr lprbconv lprbpris lpolpc ltaxpc

xtreg lcrmrte lprbarr lprbconv lprbpris lpolpc ltaxpc, re

estimate store re

xtreg lcrmrte lprbarr lprbconv lprbpris lpolpc ltaxpc, fe

estimate store fe
hausman fe re

estat hettest

estat hettest, fstat