Crime is an integral part of a functioning of any society. In one way or another every member of society feels the impact of criminal activity and there are almost no utopians who believe in the possibility of eradicating this social ill. The phenomenon of crime in the form of corruption, murder, robbery etc takes place: from the wild African kingdoms to the leaders of the world economy. And, unfortunately but naturally, Ukraine is not an exception.
Ukraine as well as other post-Soviet countries experienced a devastating period of 1990s. The collapse of the Soviet Union launched processes which completely changed the lifestyle of Ukrainians. On the ruins of the planned economy under the absence of control by the state (and to a greater extent with the support of) the country was flooded with crime.
The economic recovering from the collapse of 1990s began only in 2001-2002 years but social and demographic costs are irreparable. In the everyday life of ordinary Ukrainian citizens joined and strengthened the concepts of the criminal world: "strelka", "ponyatiya", "bratva" etc. This list is almost endless. Criminal wars were the main topics of news and dozens of specialized TV programs. Literature and film industry as litmus tests showed tendencies in society: strong skinhead guys dressed in leather jackets and hung with gold chains became the embodiment of the new "Ukrainian dream". Their Law "The strongest takes all" has become the basic principle of the daily and political life in our country while formal legislation was seen as something unworthy of paper on which was written.
The areas with a reputation for banditry estates where no one can feel safe emerged on the territory of the Ukrainian cities. Troyeshchyna and Otradnyi in Kyiv, Solnechnyi and Kommunar in Dnipropetrovs'k, Syhiv in Lviv, Poskot in Odessa, Textilschik in Donetsk and HTZ in Kharkiv became symbols of the immorality, economic devastation and hopelessness. Moreover, such perceptions shifted to the national level. As a result, residents of Zaporizhia or Donetsk region automatically get branded as criminals. The Ministry of Home Affairs confirmed these stereotypes: the general crime rate in the industrialized eastern and southern regions is higher than in the west and center of the country. Such a reputation brings significant economic losses in these regions. Therefore, it is very important to find what defines this pattern. What differences between the East and the West are responsible for such a distribution of criminal activity? What the factors of social and economic development significantly affect crime rate in Ukraine. Is there a so-called neighborhood or spatial effect, and how strong is it? The purpose of this article is to answer aforementioned questions.
The thesis is structured as follows: Chapter 2 presents the literature review; the Chapter 3 and Chapter 4 describe methodology and data used in this study respectively. Then Chapter 5 provides an empirical analysis and, finally, Chapter 6 summarizes all results of this study.
In this chapter the important and pertinent literature on the crime determinates and spatial distribution of crime is analyzed and summarized. This section is divided into three parts which present two aforementioned topics. The first subsection describes most important studies which provide theoretical foundation of economic crime model. The second subsection provides review of empirical studies of crime. The last subsection describes literature on studying of spatial distribution of crime. Each subsection describes the progress in a field of study according to timeframes which means that it shows the evolution of a topic.
The first attempt to analyze crime from an economic perspective is a merit of Fleisher (1966). His study focuses on the role of income in making a decision to commit illegal act by individuals. Fleisher states that lower income increase the probability of committing crime and justifies his opinion by the fact that lower income decrease the relative cost of crime and cost of being caught. Moreover, he finds that income level of victim is also a strong crime incentive. It means that inequality in wealth distribution is an important factor that has positive effect on crime activity.
The Fleisher's paper is empirical and does not provide formal model. Becker (1968) is the pioneer of a theoretical justification of crime behavior. His crime economic model (CEM) is based on the principle of individual's rational behavior. In other words, a person acts illegally if benefits of crime are higher than its cost. According to this approach, Becker constructs function that relates the number of offense by each individual to the probability of getting caught, severity of punishment, reward of illegal act and income from legal activities. Logically, offense function negatively depends on the first three factors and positively on the latter. To generalize results for the entire population Becker assumes average values of all variable and find total offense function as a sum of individual functions. He concludes that an optimality of public crime policy is determined by expenditures on police and courts, the size of punishment and the form of punishment.
The Becker's economic crime model was expended by Ehrlich (1973). In his study he investigates income level and income inequality affect crime rate for different types of crime. His main findings are the facts that higher income inequality is associated with higher crime rates and, in the same time, higher income (Ehrlich measures it using median family income) is associated with higher crime rate, which is a contradiction to Fleisher's results. In addition, Ehrlich concludes that unemployment rate has a less significant effect on crime rate than income and its distribution.
The next steps in the development of economic crime model are a merit of Block and Heineke (1975). They shows that results obtained by Becker (1968) and Ehrlich (1973) are not general and represent special cases. They criticize previous papers for that crime is defined as a function of wealth only. Block and Heineke take into account time spent in two activity types (legal and illegal) whereas Becker and Ehrlich mostly focus on number of crimes rather than time.
Aforementioned papers represent the theoretical background of economic model of crime. It shows that crime decision is a result of maximization problem when an individual compare benefits and costs of crime.
Since the late eighties researchers' attention shifted from theoretical to empirical crime models. These models depart from cost-benefit analysis and focus on social and economic determinates of crime. The majority of studies pay attention to relationships of crime with inequality, unemployment, education and age structure of population with control for other demographic and cultural characteristics. Moreover the object of study may be different types of criminal activity such as violent crime, property crime etc. This subsection provides a summary of the most influential papers.
Already in early studies on the economics of crime unemployment is seen as one of the most important factors. Fleisher (1963) and Ehrlich (1973) believe that the majority of criminals are not employed and unemployment positively affect crime rate. However, several researches show that the situation is opposite. Particularly, Freeman (1994) finds that although trends of criminal activity and unemployment are similar, the crime-unemployment relationship is insignificant. Imrohoroglu et al (2001) find that approximately 79% of prisoners were employed prior to crime committing which contradict to classical ideas.
On the other hand, Lu Han et al (2010) and Witt et al (1998) detect a significant effect of unemployment in the studies concentrated on England and Wales.
To sum up, the relationship between crime and unemployment is ambiguous and depends on types of study (time-series, cross-section or panel) as well as a specification of model. In particular, the formulation of unemployment may result in different conclusions: the total unemployment or segmented by age groups. In general, age structure of population is closely related to crime.
According to Freeman (1994), age distribution within population is connected to criminal activities through inequality in earning. It is a well-known fact income of youth is lower than income of other age groups and, therefore, young people are more prone to illegal activities. Such a pattern is called the age-crime profile and was found by Quetele (1831) who showed that crime rate increases rapidly during teen age with a peak in the mid twenties and then falls. Siu Fai Leung (1992) analyzes the age-crime profile from economic point of view and concludes that crime intensity cannot be found just from it. This conclusion can be considered as an argument in support of Freeman's idea about income nature of this relationship.
The same situation is in the relationship between crime and education. The significant number of studies shows that criminals tend to have lower education level and come from disadvantage groups of population than non-criminals. An education affects crime intensity in different ways. Firstly, it increases returns to legal activities through improvement in individual's skills and, logically, increases opportunity costs of illegal act. Moreover, education affects individuals' preferences through psychological aspect. This effect is called "civilization effect" and means that a person who received higher education is less likely to commit crime because of physiological restrictions.
All aforementioned factors such as age distribution, unemployment and education in one way or another are connected with criminal's income level. But, according to Fleisher (1966), income level of another, victim side of crime also matters. Therefore, income distribution or inequality in distribution of income is an object of interest in enormous number of studies. However, the results vary substantially with types of studies and types of crime.
Fanjzylber et al (1998, 2002, 2002) find positive effect of inequality on robbery and homicide rates in panel analysis at country level. Kelly (2000) receives positive effect on violent crime, assault, robbery and burglary; negative effect on rape and no significant effect on property crime, murder, larceny and car crime. Soares (2004) detects that inequality positively affect thefts and negatively affect burglary.
After presenting the literature on theoretical foundations of economic crime model and its empirical application, this chapter switches to the second subtopic and concentrates on spatial models of crime distribution.
Spatial econometrics is special subfield of econometrics that focuses on spatial or so-called "neighborhood" effects in regression analysis. This methodology is widely and successfully used in sociology, regional sciences and different subfields of economics.
Although the roots of the spatial econometrics go into seventies (Ord 1975, Paelinck and Klaassen 1979, Cliff and Ord 1981), Anselin has a considerable influence on its development. In the long list of his studies he classifies the main types of spatial effects (Anselin 1988), specifies regression models (Anselin 1988, Anselin et al 1996) and justifies its testing and estimation methodology (Anselin 1980, 1986). Today spatial econometric analysis is booming and is used in a variety of research directions and, particularly, in studying an economics of crime.
The main idea of spatial crime studies is a diffusion of crime between spatial units. It means that crime rates of region are likely to be higher if neighbor experiences high illegal activity. Moreover, the concept of "region" varies from suburban areas to international level.
This section consists of two parts. The first part describes the methodology used in an analysis of crime determinates for country and regional level. The second part provides the description of spatial model of crime distribution.
According to empirical studies discussed in the chapter 2, the crime model can be specifies in the form of equation 1 and estimated by Ordinary Least Squares with controls for fixed and random effects:
where Y is a vector of crime rate, X is a matrix of explanatory variables, Î² are regression coefficients and Îµ is error term.
In empirical studies different independent variables are used with focus on socio-economic and socio-demographic characteristics. Based on these studies the set of explanatory variables of aforementioned model is specified as follows:
Economic development of region: the level of economic well-being of region is considered as one of the most influential factors of crime activity. Empirical studies show that economic development negatively affects participation in illegal acts;
Income level of region: generally, these characteristics are separated into two types which are inequality in income distribution and total level of income of population. From theoretic point of view, both characteristics are positively related with crime rates, which is confirmed by empirics;
Unemployment: according to Becker's CEM participation in legal activities can be treated as an opportunity cost to crime. Therefore, crime rates and unemployment move in the same direction.
Share of young people: generally, young people are more likely to commit crime;
Urbanization: a concentration of population, especially in urban areas, increases an illegal activity.
Education: In the literature, education level is considered as a factor that leads to decrease in crime rates. Education directly raises opportunity costs of crime committing and affect crime participation in several indirect ways.
Police functioning: in Becker's CEM the probability of being caught and severity of punishment are important determines of crime committing decision. Empirics uses police expenditures, number of police officers per 100,000 people and detection rate as a measure of police functioning;
Culture and morality: cultural and moral level of population has significant influence on crime activity. In empirical studies single mother rate, share of children born to single mothers and share of children born to mothers aged up to 18 are common proxies of culture and morality.
The general specification of spatial autoregressive models (SAR) can be represented as system of equations 3 (Anselin 1988).
where Y is a crime level, X is a set of explanatory variables, W is a weights matrix, Î² are regression coefficients, Î» and are spatial coefficients and Îµ and u are error terms.
However, the majority of crime studies focuses on two special cases of SAR, namely, spatial lag model (SLM) and spatial error model (SER) whose general forms represented by equation 4 and 5, respectively.
One of the based principles of spatial econometrics is a definition of spatial relationship. In the other word, the key question is a construction of weights matrix. There are two groups of methods and the simplest is based on contiguity approach. It means that
The second method takes into account a distances between regions. It means that
The data comes from the enterprise statistics from the State Statistics Committee of Ukraine. The dataset covers 24 Ukrainian regions, AR Crimea, Kyiv and Sevastopol in the period between 2001 and 2002 years. It includes economic and geographic characteristics of region. Some of these characteristics are used to construct necessary explanatory variables. The final dataset contains following variables:
Crime rates: total crime, rapes, murders, extortions, robbery and thief rates in region. It's measured as a number of crimes per 100.000 people;
Urbanization: urbanization level. It's measured as a ration of urban population to total population, %;
Morality: share of children born to single mothers, %;
Gross regional product and growth: indicators of economic development of region (mln UAH and %).
Detection rate: measure of police efficiency. It's constructed as a ratio of the number of detected crimes to the number of registered crimes, %.
Young: the share of people aged 14-24 years.
Inequality level: describe a distribution of wealth between different groups of population. It's constructed as a ratio of income of 10% richest people to income of 10% poorest people, %.
Poverty rate: the share of people those incomes are lower than cost of living, %.
Unemployment: the share of unemployed people, %.
Education: describe educational level of region. It's constructed as a ration of the number of students at universities and ptus to the number of people aged 14-24 years.
Longitude and Latitude: geographical coordinates of region capital.
The descriptive statistics of these variables is presented in Table 1.
Table 1. Descriptive statistics of the main variables
Gross regional product
The main purpose of this thesis is to explain differences between Western and Eastern-Southern regions. Table 2 compares main characteristics of these subgroups.
Table 2. Descriptive statistics of the main variables, subgroups
Gross regional product
This section presents empirical estimation of model and analysis of results. At the first stage the model (2) is estimated with Generalized Least Squares with control for random and fixed effects. The second stage provides results of spatial model estimation.
The table 3 presents the results of "first stage" estimations.
Table 3. Regression results: GLS, fixed effect and random effect estimation.
Gross regional product
As it was expected, there is a significant difference between western and southern-eastern region. The coefficient of West dummy is significant and negative while the coefficient of East dummy is positive. For GLS regression Morality, Unemployment, Poverty, Urbanization and Detection rate are also significant. Its coefficients have theoretically expected signs. The situation is a bit different for Fixed effect and Random effect estimations. The economic growth and income distribution become significant. On the other hand Education and Urbanization shifts signs to theoretically unexpected. The effect of police efficiency on criminal activity has the largest magnitude among other variables.