An empirical study of socio-economic determinants of crime

Abstract

Crimes lower the quality of life for many people in all societies. The concept of crime prevention is directly related to the studies about causation and prevention methods. State statistics institutes and law enforcement agencies store various kinds and quantities of data. Law enforcement agencies can use these data to take some measures to eliminate and minimize the crime rates. Temporal and spatial variations in crime rates have been studied extensively in the literature. A variety of demographic factors including unemployment rate, education level, age composition of the population, and average income level have been suggested by different researchers as explanatory variables for observed variations in crime rates. Variables such as number of police per capita, crime clear-up rates, and incarceration population have been considered as deterrents of crime. In this thesis, using cross-sectional data associated with 81 cities in Turkey, selected types of crime (murder, theft, crimes against persons and property) are empirically analyzed. The multiple regression results indicate that crime rates in general are negatively related to clear up rates and average income level in a region seems to be positively related to crimes against property. For short term forecasting of crime rates, a trend and seasonality adjusted exponential smoothing scheme is applied to the time series data for theft rates in three major cities. The highest mean absolute percentage eiTor computed ba.sed on one-period-ahead forecast errors is less than 15 % in all cases.