University of Wisconsin - Madison (USA)

DETERMINANTS OF CRIME RATES
OF GERMAN CITIES

by 

Tobias Bucher

TABLE OF CONTENTS

Executive Summary

Section 1: Introduction

Section 2: Data Characteristics

Section 3: Model Selection and Interpretation

Section 4: Summary and Concluding Remarks

Section 5: Appendix

EXECUTIVE SUMMARY

Crime rates, in this report defined as the number of crimes committed per 1,000 citizens, vary considerably from city to city in Germany. As crime rates are important in order to determine the quality of a city′s administration and the overall quality of life in the area, a cross-sectional study was performed and a regression model was introduced in order to better understand the determinants of crime rates of German cities.

The study shows that crime rates can be accurately predicted by means of four explanatory variables. The number of inhabitants of the city and the percentage rate of severe thefts among the total number of crimes are positively correlated to the cities′ crime rates. The percentage rate of foreign people among the citizenry also influences the crime rate, but the relationship is somewhat more complicated since the squared term of this variable is included in the recommended regression model. We will see that, as a rule of thumb, the crime rate actually decreases as the percentage of foreigners among the total population increases as long as this rate is less than a certain value, and that the crime rate increases as the percentage rate of foreigners increases once this value is exceeded. Finally, this study shows that, other things equal, the crime rates of cities that are located in the northern part of Germany are significantly higher than the crime rates of cities in Southern Germany. The model presented in this report explains about 75.9% of the variability in the data on crime rates. Therefore, it can be used to make fairly accurate predictions of the crime rates of German cities.

SECTION 1 INTRODUCTION

A city′s crime rate, defined as the total number of crimes that were committed per 1,000 citizens within one year, is a very important figure. It does not only provide information about the actual frequency of crimes, but also represents a good clue for the quality of the city′s administration and the effectiveness of the police in protecting the citizens from criminal acts. Moreover, crime rates can be a very important factor influencing the overall quality of life in the city.

Nowadays, more and more German cities are concerned about their image and pay particular attention to delinquency prevention programs and the citizens′ increased demand for safety. Detailed police statistics are published more often than in previous years, although the time offset between the observation period and the time of the data′s public release is still very long. However, the reports published by the German police allow comparisons between all major German cities in terms of the frequency and the kinds of crimes committed within the city limits.

This study was conducted in order to better understand the determinants of crime rates of German cities and to provide predictions of the actual crime rates in a timelier manner than official reports can do. A regression model was fit to a cross-sectional data set consisting of actual crime rates of forty-three German cities and a total of nine predictor variables that I anticipated to provide some information about the variation in the cities′ crime rates. The data was obtained from various sources, among them an annual abstract of statistics and two websites. More detailed information on the sources of data is provided in Appendix A.1.

The outline of the remainder of this report is as follows. In Section 2, Data Characteristics, I will present basic summary statistics of the response variable, the crime rate, and analyze its relationship to the nine potential explanatory variables. Section 3, Model Selection and Interpretation, introduces the recommended regression model and provides in-depth analysis of the relationships between the crime rate and the explanatory variables of the model. Furthermore, I will show the practical relevance and validity of the model by making some predictions and interpreting important regression statistics of the recommended model as well as of two alternative models. Section 4, Summary and Concluding Remarks, provides a short summary of the key findings of my study, a brief discussion of the problems that I experienced during the research process, and some flaws of the model. Finally, at the conclusion of this report, the Appendix provides all relevant details to my analysis.

SECTION 2 DATA CHARACTERISTICS

[...]