crime statistics research paper

• Open access
• Published: 01 December 2021

More crime in cities? On the scaling laws of crime and the inadequacy of per capita rankings—a cross-country study

• Marcos Oliveira   ORCID: orcid.org/0000-0003-3407-5361 1 , 2  

Crime Science volume  10 , Article number:  27 ( 2021 ) Cite this article

28k Accesses

12 Citations

70 Altmetric

Metrics details

Crime rates per capita are used virtually everywhere to rank and compare cities. However, their usage relies on a strong linear assumption that crime increases at the same pace as the number of people in a region. In this paper, we demonstrate that using per capita rates to rank cities can produce substantially different rankings from rankings adjusted for population size. We analyze the population–crime relationship in cities across 12 countries and assess the impact of per capita measurements on crime analyses, depending on the type of offense. In most countries, we find that theft increases superlinearly with population size, whereas burglary increases linearly. Our results reveal that per capita rankings can differ from population-adjusted rankings such that they disagree in approximately half of the top 10 most dangerous cities in the data analyzed here. Hence, we advise caution when using crime rates per capita to rank cities and recommend evaluating the linear plausibility before doing so.

Introduction

In criminology, it is generally accepted that crime occurs more often in more populated regions. In one of the first works of modern criminology, Balbi and Guerry examined the crime distribution across France in 1825, revealing that some areas experienced more crime than others (Balbi and Guerry, 1829 ; Friendly, 2007 ). To compare these areas, they realized the need to adjust for population size and analyzed crime rates instead of raw numbers. This method eliminates the linear effect of population size on crime numbers and has been used to measure crime and compare cities almost everywhere—from academia to news outlets (Hall, 2016 ; Park and Katz, 2016 ; Siegel, 2011 ). However, this approach overlooks the potential nonlinear effects of population and, more importantly, exposes our limited understanding of the population–crime relationship.

Though different criminology theories expect a relationship between population size and crime, they tend to disagree on how crime increases with population (Chamlin and Cochran, 2004 ; Rotolo and Tittle, 2006 ). These theories predict divergent population effects, such as linear and superlinear crime growth. Despite these theoretical disputes, however, crime rates per capita are broadly used by assuming that crime increases linearly with the number of people in a region. Crucially, crime rates are often deemed to be a standard means of comparing crime in cities.

Yet the widespread adoption of crime rates is arguably due more to tradition (Boivin, 2013 ) rather than its ability to remove the effects of population size. Many urban indicators, including crime, have already been shown to increase nonlinearly with population size (Bettencourt et al., 2007 ). When we violate the linear assumption and use rates, we deal with quantities that still have population effects, thus introducing an artifactual bias into rankings and analyses.

Despite this inadequacy, we only have a limited understanding of the impact of nonlinearity on crime rates. Although previous works have investigated population–crime relationships extensively (Alves et al., 2013a ; Bettencourt et al. 2010 ; Chang et al. 2019 ; Gomez-Lievano et al., 2012 ; Hanley et al., 2016 ; Yang et al., 2019 ), they have failed to quantify the impact of nonlinear relationships on rankings and restricted their analyses to either specific offenses or countries. The lack of comprehensive systematic studies has limited our knowledge on how the linear assumption influences crime analyses and, more critically, has prevented us from better understanding the effect of population on crime.

In this work, we analyze burglaries and thefts in 12 countries and investigate how crime rates per capita can misrepresent cities in rankings. Instead of assuming that the population–crime relationship is linear, we estimate this relationship from data using probabilistic scaling analysis (Leitão et al., 2016 ). We use our estimates to rank cities while adjusting for population size, and we then examine how these rankings differ from rankings based on rates per capita. In our results, we find that the linear assumption is unjustified. We show that using crime rates to rank cities can lead to rankings that considerably differ from rankings adjusted for population size. Finally, our results reveal contrasting growths of burglaries and thefts with population size, implying that different crime dynamics can produce distinct features at the city level. Our work sheds light on the population–crime relationship and suggests caution in using crime rates per capita.

Crime and population size

Different theoretical perspectives predict the emergence of a relationship between population size and crime. Three main criminology theories expect this relationship: structural, social control, and subcultural (Chamlin and Cochran, 2004 ; Rotolo and Tittle, 2006 ). In general, these perspectives agree that variations in the number of people in a region have an impact on the way people interact with one another. These theories, however, differ in the types of changes in social interaction and how they can produce a population–crime relationship.

From a structural perspective, a higher number of people increases the chances of social interaction, which increases the occurrence of crime. Two distinct rationales can explain such an increase. Mayhew and Levinger ( 1976 ) posit that crime is a product of human contact: more interaction leads to higher chances of individuals being exploited, offended, or harmed. They claim that a larger population size raises the number of opportunities for interaction at an increasing rate, which would lead to a superlinear crime growth with population size (Chamlin and Cochran, 2004 ). In contrast, Blau ( 1977 ) implies a linear population–crime relationship. He posits that population aggregation reduces spatial distance among individuals, thereby promoting different social associations such as victimization. At the same time, as conflictive association increases, other integrative associations also increase, leading to a linear growth of crime (Chamlin and Cochran, 2004 ). Notably, the structural perspective focuses on the quantitative consequences of population growth.

The social control perspective advocates that changes in population size have a qualitative impact on social relations, which weakens informal social control mechanisms that inhibit crime (Groff, 2015 ). From this perspective, crime relates to two aspect of a population: size and stability. A larger population size leads to higher population density and heterogeneity—not only do individuals have more opportunities for social contacts, but they are also often surrounded by strangers (Wirth, 1938 ). This situation makes social integration difficult and promotes a high anonymity, which encourages criminal impulses and harms a community’s ability to socially constrain misbehavior (Freudenburg, 1986 ; Sampson, 1986 ). Similarly, from a systemic viewpoint, any change (i.e., increase or decrease) in population size can have an impact on crime numbers (Rotolo and Tittle, 2006 ). From this viewpoint, the understanding is that regular and sustained social interactions produce community networks with effective mechanisms of social control (Bursik and Webb, 1982 ). Population instability, however, hinders the construction of such networks. In communities with unstable population size, residents avoid socially investing in their neighborhoods, which hurts community organization and weakens social control, thus increasing misbehavior and crime (Miethe et al., 1991 ; Sampson, 1988 ).

Both social-control and structural perspectives solely focus on individuals’ interactions without considering their private interests. These perspectives pay little attention to how unconventional interests increase with urbanization (Fischer, 1975 ) and how these interests relate to misbehavior.

In contrast, the subcultural perspective advocates that population concentration brings together individuals with shared interests, which produces private social networks built around these interests, thereby promoting social support for behavioral choices. Fischer ( 1975 ) posits that population size has an impact on the creation, diffusion, and intensification of unconventional interests. He proposes that large populations have a sufficient number of people with specific shared interests, thus enabling social interaction and lead to the emergence of subcultures. The social networks surrounding a subculture bring normative expectations that increase the likelihood of misbehavior and crime (Fischer, 1975 ,, 1995 ).

These three perspectives—structural, social control, and subcultural—expect that a higher number of people in an area leads to more crime in that area. In the case of cities, we know that population size is indeed a strong predictor of crime (Bettencourt et al., 2007 ) . The existence of a population–crime relationship implies that we must adjust for population size to analyze crime in cities properly.

Crime rates per capita

In the literature, the typical solution for removing the effect of population size from crime numbers is to use ratios such as the following:

This ratio is often used together with a multiplier that contextualizes the quantity (e.g., crime per 100,000 inhabitants; Boivin, 2013 ). However, even though crime rates are popularly used, they present at least two inadequacies. First, the way in which we define population affects crime rates. The common approach is to use resident population (e.g., census data) to estimate rates, but this practice can distort the picture of crime in a place: crime is not limited to residents (Gibbs and Erickson, 1976 ), and cities attract a substantial number of non-residents (Stults and Hasbrouck, 2015 ). Instead, researchers suggest using ambient population (Andresen, 2006 ,, 2011 ) and accounting for criminal opportunities, which depends on the type of crime (Boggs, 1965 ; Clarke, 1984 ; Cohen et al., 1985 ; Harries, 1981 ).

Second, Eq. ( 1 ) assumes that the population–crime relationship is linear. The rationale behind this equation is that we have a relationship of the form

which means that crime can be linearly approximated via population. Given this linear assumption, when we divide crime by population in Eq. ( 1 ), we are trying to cancel out the effect of population on crime. This assumption implies that crime increases at the same pace as population growth. However, not all theoretical perspectives agree with this type of growth, and many urban indicators, including crime, have been shown to increase with population size in a nonlinear fashion (Bettencourt et al., 2007 ).

Cities and scaling laws

Much research has been devoted to understanding urban growth and its impact on indicators such as gross domestic product, total wages, electrical consumption, and crime (Bettencourt et al., 2007 , 2010 ; Bettencourt, 2013 ; Gomez-Lievano et al., 2016 ). Bettencourt et al. ( 2007 ) have found that a city’s population size, denoted by N , is a strong predictor of its urban indicators, denoted by Y , exhibiting the following relationship:

This so-called scaling law tells us that, given the size of a city, we expect certain levels of wealth creation, knowledge production, criminality, and other urban aspects. This expectation suggests general processes underlying urban development (Bettencourt et al., 2013 ) and indicates that regularities exist in cities despite their idiosyncrasies (Oliveira and Menezes, 2019 ). To understand this scaling and the urban processes better, we can examine the exponent \(\beta\) , which describes how an urban indicator grows with population size.

Urban scaling laws and rates per capita. The way in which urban indicators increase with population size depends on the class of the indicator. A Social aspects, such as crime and total wages, increase superlinearly with population size, whereas infrastructural indicators (e.g., road length) increase sublinearly. B  In nonlinear scenarios, rates per capita still depend on population size

Bettencourt et al. ( 2007 ) presented evidence that different categories of urban indicators exhibit distinct growth regimes. They showed that social indicators grow faster than infrastructural ones (see Fig.  1 A). Specifically, social indicators, such as the number of patents and total wages, increase superlinearly with population size (i.e., \(\beta > 1\) ), meaning that these indicators grow at an increasing rate with population. In the case of infrastructural aspects (e.g., road surface, length of electrical cables), an economy of scale exists. As cities grow in population size, these urban indicators increase at a slower pace with \(\beta < 1\) (i.e., sublinearly). In both scenarios, because of nonlinearity, we should be careful with per capita analyses.

When we violate the linear assumption of per capita ratios, we deal with quantities that can misrepresent an urban indicator. To demonstrate this, we use Eq. ( 3 ) to define the per capita rate C of an urban indicator as follows:

which implies that rates are independent from population only when \(\beta\) equals to one—when \(\beta \ne 1\) , population is not cancelled out from the equation. In these nonlinear cases, per capita rates can inflate or deflate the representation of an urban indicator depending on \(\beta\) (see Fig.  1 B). This misrepresentation occurs because population still has an effect on rates. By definition, we expect that per capita rates are higher in larger cities when \(\beta > 1\) , whereas when \(\beta < 1\) , we expect larger cities to have lower rates. When we use rates to compare cities in nonlinear situations, we introduce an artifactual bias. To compare cities properly, previous works have proposed scaled-adjusted indicators that account for population size (Alves et al., 2013a ; Bettencourt et al., 2010 ), supporting the need for population adjustment but failing to quantify the impact of the linear assumption on rankings of urban indicators.

More crime in cities?

In the case of crime, researchers have found a superlinear growth with population size. Bettencourt et al. ( 2007 ) showed that serious crime in the United States exhibits superlinear scaling with exponent \(\beta \approx 1.16\) , and some evidence has confirmed similar superlinearity for homicides in Brazil, Colombia, and Mexico (Alves et al., 2013b ; Gomez-Lievano et al., 2012 ). Previous works have also found that different kinds of crime in the United Kingdom and in the United States present nonlinear scaling relationships (Chang et al., 2019 ; Hanley et al., 2016 ; Yang et al., 2019 ). Remarkably, the existence of these scaling laws of crime suggests fundamental urban processes that relate to crime, independent of cities’ particularities.

This regularity manifests itself in the so-called scale-invariance property of scaling laws. It is possible to show that Eq. ( 3 ) holds the following property:

where \(g(\kappa )\) does not depend on N  (Thurner et al., 2018 ). From a modeling perspective, this relationship reveals two aspects about crime. First, we can predict crime numbers in cities via a populational scale transformation \(\kappa\)  (Bettencourt et al., 2013 ). This transformation is independent of population size but depends on \(\beta\) , which tunes the relative increase in crime such that \(g(\kappa ) = \kappa ^\beta\) . Second, Eq. ( 4 ) implies that crime is present in any city, independent of size. This implication arguably relates to the Durkheimian concept of crime normalcy in that crime is seen as a normal and necessary phenomenon in societies, provided that its numbers are not unusually high (Durkheim, 1895 ). Broadly speaking, the scale-invariance property tells us that crime in cities is associated with population in a somewhat predictable fashion. Crucially, this property might give the impression that such regularity is independent of crime type.

However, different types of crime are connected to social mechanisms in different ways (Hipp and Steenbeek, 2016 ) and exhibit unique temporal (Miethe et al., 2005 ; Oliveira et al., 2018 ) and spatial characteristics (Andresen and Linning, 2012 ; Oliveira et al., 2015 , 2017 ; White et al., 2014 ). It is plausible that the scaling laws of crime depend on crime type. Nevertheless, the literature has mostly focused on either specific countries or crime types. Few studies have systematically examined the scaling of different crime types, and the focus on specific countries has prevented us from better understanding the impact of population on crime. Likewise, the lack of a comprehensive systematic study has limited our knowledge about the impact of the linear assumption on crime rates. We still fail to understand how per capita analyses can misrepresent cities in nonlinear scenarios.

In this work, we characterize the scaling laws of burglary and theft in 12 countries and investigate how crime rates per capita can misrepresent cities in rankings. Instead of assuming that the population–crime relationship is linear, as described in Eq. ( 2 ), we investigate this relationship under its functional form as follows:

Specifically, we examine the plausibility of scaling laws to describe the population–crime relationship. To estimate the scaling laws, we use probabilistic scaling analysis, which enables us to characterize the scaling laws of crime. We use our estimates to rank cities while accounting for the effects of population size. Finally, we compare these adjusted rankings with rankings based on per-capita rates (i.e., with the linear assumption).

We use data from 12 countries to investigate the relationship between population size and crime at the city level (see the appendix for data sources). Specifically, we examine annual data from Belgium, Canada, Colombia, Denmark, France, Italy, Mexico, Portugal, South Africa, Spain, the United Kingdom, and the United States (see Table  1 ). In this work, we characterize how crime increases with population size in each country, focusing on burglary and theft. We analyze both crimes in all considered countries, except Mexico, Portugal, and Spain, where we only have data for one type of offense.

The population–crime relationship in 12 countries. Different criminology theories expect a relationship between population size and crime, predicting divergent population effects, such as linear and superlinear crime growth. Despite these theoretical disputes, however, crime rates per capita are broadly used by assuming that crime increases linearly with population size

The scaling laws of crime in cities

To assess the relationship between crime Y and population size N (see Fig.  2 ), we model \(\mathrm{P}(Y|N)\) using probabilistic scaling analysis (see the Methods section). In our study, we examine whether this relationship follows the general form of \(Y \sim N^\beta\) . First, we estimate \(\beta\) from data, and we then evaluate the plausibility of the model ( \(p>0.05\) ) and the evidence for nonlinearity (i.e., \(\beta \ne 1\) ). Our results reveal that Y and N often exhibit a nonlinear relationship, depending on the type of offense.

The scaling laws of crime. We find evidence for a nonlinear relationship between crime and population size in more than half of the data sets. In most considered countries, theft exhibits superlinearity, whereas burglary tends to display linearity. In the plot, the lines represent the error bars for the estimated \(\beta\) of each country–crime for two consecutive years; circles denote a lack of nonlinearity plausibility; triangles represent superlinearity, and upside-down triangles indicate sublinearity

In most of the considered countries, theft increases with population size superlinearly, whereas burglary tends to increase linearly (see Fig.  3 ). Precisely, in 9 out of 11 countries, we find that \(\beta\) for theft is above one; our results indicate linearity for theft (i.e., absence of nonlinear plausibility) in Canada and South Africa. In the case of burglary, we are unable to reject linearity in 7 out of 10 countries; in France and the United Kingdom, we find superlinearity, and in Canada, sublinearity. In almost all considered data sets, these estimates are consistent over two consecutive years in the countries for which we have data for different years (see Appendix  1 ).

Our results suggest that the general form of \(Y\sim N^\beta\) is plausible in most countries, but that this compatibility depends on the offense. We find that burglary data are compatible with the model ( \(p>0.05\) ) in 80% of the considered countries. In the case of theft, the superlinear models are compatible with data in five out of nine countries. We note that in Canada and South Africa, where we are unable to reject linearity for theft, the linear model also lacks compatibility with data.

We find that the estimates of \(\beta\) for each offense often have different values across countries—for example, the superlinear estimates of \(\beta\) for theft range from 1.10 to 1.67. However, when we analyze each country separately, we find that \(\beta\) for theft tends to be larger than \(\beta\) for burglary in each country, except for France and the United Kingdom.

In summary, we find evidence for a nonlinear relationship between crime and population size in more than half of the considered data sets. Our results indicate that crime often increases with population size at a pace that is different from per capita. This relationship implies that analyses with a linear assumption might create distorted pictures of crime in cities. To understand such distortions, we must examine how nonlinearity influences comparisons of crime in cities, when linearity is assumed.

Bias in crime rates per capita. When crime increases nonlinearly with population size, we have an artifactual bias in crime rates. The linearity in Portugal makes rates independent of size (left). However, in Denmark (right), because of the superlinear growth, we expect larger cities to have higher crime rates, but not necessarily more crime than expected. For example, though Aalborg and Solrød have similar theft rates, less crime occurs in Aalborg than expected for cities of the same size, based on the model, whereas Solrød is above the expectation

The inadequacy of crime rates and per capita rankings

We investigate how crime rates of the form \(C = Y/N\) introduce bias in the comparisons and rankings of cities. To understand this bias, we use Eq. ( 3 ) to rewrite crime rate as \(C \sim N^{\beta - 1}\) . This relationship implies that crime rate depends on population size when \(\beta \ne 1\) . For example, in Portugal and Denmark, this dependency is clear when we analyze burglary and theft numbers (see Fig.  4 ). In the case of burglary in Portugal, linearity makes C independent of population size. In Denmark, since theft increases superlinearly, we expect rates to increase with population size. In this country, based on data, the expected theft rate of a small city is lower than the rates of larger cities. We must account for this tendency in order to compare crime in cities; otherwise, we introduce bias against larger cities.

To account for the population–crime relationship found in data, we compare cities using the model P ( Y | N ) as the baseline. We compare the number of crimes in a city with the expectation of the model. For each city i with population size \(n_i\) , we evaluate the z score of the city with respect to \(P(Y|N=n_i)\) . The z score indicates how much more or less crime a particular city has in comparison to cities with a similar population size, as expected by the model. These z scores enable us to compare cities in a country and rank them while accounting for population size differences. In contrast, crime rates per capita only adjust for population size in the linear scenario. This approach is similar to previously proposed indicators that adjust for population size (Alves et al., 2013a ; Bettencourt et al., 2010 ). In our case, the adjustment also accounts for the variance. We denote this kind of analysis as a comparison adjusted for the population–crime relationship.

For example, in Denmark, the theft rate in the municipality of Aalborg ( \(\approx 0.0186\) ) is almost the same as in Solrød ( \(\approx 0.0188\) ). However, less crime occurs in Aalborg than expected for cities of a similar size, while crime in Solrød is above the model expectation (see Fig.  4 B). This disagreement arises because of the different population sizes. Since Aalborg is more than 10 times larger than Solrød, we expect rates in Aalborg to be larger than in Solrød. When we account for this tendency and evaluate their z scores, we find that the z score of Aalborg is \(-2.47\) , whereas in Solrød the z  score is 2.43.

Such inconsistencies have an impact on the crime rankings of cities. The municipality of Aarhus, in Denmark, for example, is ranked among the top 12 cities with the highest theft rate in the country. However, when we account for population–crime relationship using z scores, we find that Aarhus is only at the end of the top 54 rankings.

The inadequacy of per capita rankings. Per capita ranking can differ substantially from rankings adjusted for population size, depending on the scaling exponent. In Italy and Denmark, for example, A theft ranks (top) diverge considerably more than the ranks for burglary (bottom). Data points represent cities’ positions in the rankings. B In nonlinear cases, these rankings diverge, as measured via rank correlation

To understand these variations systematically, we compare rankings based on crime rates with rankings that account for the population–crime relationship (i.e., adjusted rankings). Our results reveal that these two rankings create distinct representations of cities. For each considered data set, we rank cities based on their z scores and crime rates C , and we then examine the change in the rank of each city. According to our findings, the positions of the cities can change substantially. For instance, in Italy, half of the cities have theft rate ranks that diverge in at least 11 positions from the adjusted ranking (Fig.  5 A). This disagreement means that these rankings disagree for approximately half of the top 10 most dangerous cities.

We evaluate these discrepancies by using the Kendall rank correlation coefficient \(\tau\) to measure the similarity between crime rates and adjusted rankings in the considered countries. We find that these rankings can differ considerably but converge when \(\beta \approx 1\) . The \(\tau\) coefficients for the data sets range from 0.6 to 1.0, exhibiting a dependency on the type of crime; or more specifically, on the scaling (Fig.  5 B). As expected, as \(\beta\) approaches 1, the rankings are more similar to one another. For example, in Italy, in contrast to theft, the burglary rate ranking of half of the cities only differs from the adjusted ranking in a maximum of two positions (Fig.  5 A).

Discussion and conclusion

Despite its popularity, comparing cities via crime rates without accounting for population size has a strong assumption that crime increases at the same pace as the number of people in a region. Though previous works have widely investigated the population–crime relationship, they have failed to quantify the impact of nonlinear relationships on rankings and restricted their analyses to either specific offenses or countries. In this work, we analyze crime in different countries to investigate how crime grows with population size and how the widespread assumption of linear growth influences cities’ rankings.

First, we analyzed crime in cities from 12 countries to characterize the population–crime relationship statistically, examining the plausibility of scaling laws to describe this relationship. Then, we used our estimates to rank cities and compared how those rankings differ from rankings based on rates per capita.

Our results showed that the assumption of linear crime growth is unfounded. In more than half of the considered data sets, we found evidence for nonlinear crime growth—that is, crime often increases with population size at a different pace than per capita. This nonlinearity introduces a population effect into crime rates, influencing rankings. We demonstrated that using crime rates to rank cities substantially differs from ranking cities adjusted for population size.

These findings imply that using crime rates per capita—though deemed a standard measure in criminal justice statistics—can create a distorted view of cities’ rankings. For example, in superlinear scenarios, we expect larger cities to have higher crime rates. In this case, when we use rates to rank cities, we build rankings whereby large cities are at the top. But, these cities might not experience more crime than what we expect from places with a similar population size. It is an artifactual bias introduced by population effects still present in crime rates.

Such effects arise from nonlinear population effects that persist in rates due to the linear assumption. This assumption is more than just a statistical subtlety. By assuming linearity, we essentially overlook cities’ context: we ignore the actual impact of population size on crime and how this impact depends on crime type, country, and aggregation units, among other things. For instance, our results indicate that in thefts, linearity is an exception rather than the rule. The indiscriminate use of crime rates neglects significant population–crime interactions that should be considered in order to compare crime in cities properly.

As a result of this inadequacy, we advise caution when using crime rates per capita to compare cities. We recommend evaluating linear plausibility before comparing crime rates. In general, we suggest comparing cities via the z scores computed using the approach (Leitão et al., 2016 ) discussed in the manuscript, thereby avoiding crime rates. It is important to emphasize that this inadequacy in rates is relevant only when comparing cities of different population sizes. In analyses without comparisons, a place’s crime rate can be seen as a rough indicator that contextualizes crime numbers relative to population size. Additionally, when cities have the same size, comparing crime rates boils down to comparing raw crime numbers.

In summary, in this work, we shed light on the population–crime relationship. The linear assumption is exhausted and expired. We have resounding evidence of nonlinearity in crime, which disallows us from unjustifiably assuming linearity. In light of our results, we also note that the scaling laws are plausible models only for half of the considered data sets. Better models are thus needed—in particular, models that account for the fact that different crime types relate to population size differently. More adequate models will help us better understand the relationship between population and crime.

Limitations

Our work presents limitations related to the way in which we define population, crime, and cities. First, we note that crime rates depend on how we define population; in our study, we define it as the resident population (i.e., census data). However, crime is not limited to residents (Gibbs and Erickson, 1976 ), and cities attract a significant number of non-residents (Stults and Hasbrouck, 2015 ). We highlight that this limitation is not specific to our study, and crime rates are often measured using resident population. Previous works have suggested using ambient population and accounting for the number of targets (Andresen, 2006 , 2011 ; Boggs, 1965 ). Collecting this data, however, is challenging, especially when dealing with different countries. Future research should investigate crime rates and scaling laws using other definitions of population, particularly using social media data (Malleson and Andresen, 2016 ; Pacheco et al., 2017 ).

Second, scaling analyses depend on the definition of what constitutes a city (Arcaute et al., 2014 ). In the literature, definitions include legal divisions (e.g., counties, municipalities) and data-driven delineations based on population density and economic interactions (Cottineau et al., 2017 ). It is possible that different city definitions yield divergent scaling regimes for the same urban indicator (Louf and Barthelemy, 2014 ). In our work, we only have access to crime data regarding specific aggregation units, and we thus define cities based on official legal divisions by using census data. City definitions in our analysis consequently depend on the country. We emphasize that we investigate whether per capita rankings are justified under a given city definition. Nevertheless, we believe that even though the use of other city definitions might change our quantitative results, our qualitative results are robust: the inadequacy of crime rates is independent of city definitions. When analyzing different definitions of cities, future research should examine scaling divergences as an opportunity to understand the population–crime relationship better.

Finally, cross-national crime analyses have methodological challenges due to international differences in crime definitions, police and court practices, and reporting rates, among other things (Takala and Aromaa, 2008 ). Although we avoid direct comparisons of countries’ absolute crime numbers in our work, we compare their growth exponents. In this comparison, we assume that cross-national differences have a negligible impact on how crime increases with population, particularly regarding the crime types we analyzed. We understand that some offenses (e.g., sexual assault, drug trafficking) are more sensitive to cross-national comparisons than the offenses we analyzed here (Harrendorf et al., 2010 ; Harrendorf, 2018 ). Collecting high-quality international comparative data could help future works in disentangling cross-national differences.

Probabilistic scaling analysis

We use probabilistic scaling analysis to estimate the scaling laws of crime. Instead of analyzing the linear form of Eq. ( 3 ), we use the approach developed by Leitão et al. ( 2016 ) to estimate the parameters of a distribution Y | N that has the following expectation:

that is, N scales the expected value of an urban indicator (Bettencourt et al., 2013 ; Gomez-Lievano et al., 2012 ; Leitão et al., 2016 ). Note that this method does not assume that the fluctuations around \(\ln y\) and \(\ln x\) are normally distributed (Leitão et al., 2016 ). Instead, we compare models for \(\mathrm{P}(Y|N)\) that satisfy the following conditional variance:

where typically \(\delta \in [1,2]\) , since urban systems have been previously shown to exhibit non-trivial fluctuations around the mean—the so-called Taylor’s law (Hanley et al., 2014 ). To estimate the scaling laws, we maximize the log-likelihood

since we assume \(y_i\) as an independent realization from \(\mathrm{P}(Y|N)\) . In this work, we use an implementation developed by Leitão et al. ( 2016 ) that maximizes the log-likelihood with the “L-BFGS-B” algorithm. We model \(\mathrm{P}(Y|N)\) using Gaussian and log-normal distributions in order to analyze whether accounting for the size-dependent variance influences the estimation. In the case of the Gaussian, the conditions from Eq. ( 5 ) and Eq. ( 6 ) are satisfied with

whereas in the case of the log-normal distribution,

In the log-normal case, note that, if \(\delta = 2\) , then the fluctuations are independent of N ; thus this would be the same as using the minimum least-squares approach (Leitão et al., 2016 ). With this framework, we compare models that have fixed \(\delta\) against models wherein \(\delta\) is also included in the optimization process. In the case of the Gaussian, we have fixed \(\delta =1\) and free \(\delta \in [1,2]\) , whereas in the case of the log-normal, we have fixed \(\delta =2\) and free \(\delta \in [1,3]\) . In this framework, p -values represent a statistic testing two crucial aspects of the modelling: sample independence and model compatibility with data. The statistic consists of the D’Agostino \(K^2\) test together with Spearman’s rank correlation of residuals, which evaluates compatibility and independence, respectively (Leitão et al., 2016 )

Finally, we compare each of the four models individually against the linear alternative (with fixed \(\beta = 1\) ), to test the nonlinearity plausibility. With the fits of all types of crime and countries, we measure the Bayesian information criterion ( \(\mathrm{BIC}\) ), defined as

where k is the number of free parameters in the model and lower \(\mathrm{BIC}\) values indicate better data description. The \(\mathrm{BIC}\) value of each fit enables us to compare the models’ ability to explain data.

Availability of data and materials

All data and source code are available at https://github.com/macoj/scaling_laws_of_crime/ .

Alves, L. G. A., Ribeiro, H. V., Lenzi, E. K., & Mendes, R. S. (2013a). Distance to the scaling law: A useful approach for unveiling relationships between crime and urban metrics. PLoS ONE, 8 (8), e69580.

Article   Google Scholar  

Alves, L. G. A., Ribeiro, H. V., & Mendes, R. S. (2013b). Scaling laws in the dynamics of crime growth rate. Physica A: Statistical Mechanics and Its Applications, 392 (11), 2672–2679.

Andresen, M. A. (2006). Crime measures and the spatial analysis of criminal activity. The British Journal of Criminology, 46 (2), 258–285.

Andresen, M. A. (2011). The ambient population and crime analysis. The Professional Geographer, 63 (2), 193–212.

Andresen, M. A., & Linning, S. J. (2012). The (in)appropriateness of aggregating across crime types. Applied Geography, 35 (1–2), 275–282.

Arcaute, E., Hatna, E., Ferguson, P., Youn, H., Johansson, A., & Batty, M. (2014). Constructing cities, deconstructing scaling laws. Journal of The Royal Society Interface, 12 (102), 20140745.

Balbi, A., & Guerry, A.-M. (1829). Statistique comparée de l’état de l’instruction et du nombre des crimes dans les divers arrondissements des Académies et des Cours Royales de France . Paris: Everat.

Bettencourt, L. M. A. (2013). The origins of scaling in cities. Science, 340 (6139), 1438–1441.

Bettencourt, L. M. A., Lobo, J., Helbing, D., Kühnert, C., & West, G. B. (2007). Growth, innovation, scaling, and the pace of life in cities. Proceedings of the National Academy of Sciences, 104 (17), 7301–7306.

Bettencourt, L. M. A., Lobo, J., Strumsky, D., & West, G. B. (2010). Urban scaling and its deviations: Revealing the structure of wealth, innovation and crime across cities. PLoS ONE, 5 (11), e13541.

Bettencourt, L. M. A., Lobo, J., & Youn, H. (2013). The hypothesis of urban scaling: Formalization, implications and challenges. Technical Report arXiv:1301.5919 [physics.soc-ph], arXiv.

Blau, P. (1977). Inequality and heterogeneity: A primitive theory of social structure (1st ed.). Free Press.

Google Scholar  

Boggs, S. L. (1965). Urban crime patterns. American Sociological Review, 30 (6), 899.

Boivin, R. (2013). On the use of crime rates. Canadian Journal of Criminology and Criminal Justice, 55 (2), 263–277.

Bursik, R. J., & Webb, J. (1982). Community change and patterns of delinquency. American Journal of Sociology, 88 (1), 24–42.

Chamlin, M. B., & Cochran, J. K. (2004). An excursus on the population size–crime relationship. Western Criminology Review, 5 (2), 119–130.

Chang, Y. S., Kim, H. E., & Jeon, S. (2019). Do larger cities experience lower crime rates? A scaling analysis of 758 cities in the U.S. Sustainability, 11 (11), 3111.

Clarke, R. V. (1984). Opportunity-based crime rates: The difficulties of further refinement. The British Journal of Criminology, 24 (1), 74–83.

Cohen, L. E., Kaufman, R. L., & Gottfredson, M. R. (1985). Risk-based crime statistics: A forecasting comparison for burglary and auto theft. Journal of Criminal Justice, 13 (5), 445–457.

Cottineau, C., Hatna, E., Arcaute, E., & Batty, M. (2017). Diverse cities or the systematic paradox of urban scaling laws. Computers, Environment and Urban Systems, 63 , 80–94.

Durkheim, É. (1895). Les règles de la méthode sociologique . Paris: Flix Alcan.

Fischer, C. S. (1975). Toward a subcultural theory of urbanism. American Journal of Sociology, 80 (6), 1319–1341.

Fischer, C. S. (1995). The subcultural theory of urbanism: A twentieth-year assessment. American Journal of Sociology, 101 (3), 543–577.

Freudenburg, W. R. (1986). The density of acquaintanceship: An overlooked variable in community research? American Journal of Sociology, 92 (1), 27–63.

Friendly, M. (2007). A.-M. Guerry’s Moral statistics of France : Challenges for multivariable spatial analysis. Statistical Science   22 (3), 368–399.

Gibbs, J. P., & Erickson, M. L. (1976). Crime rates of American cities in an ecological context. American Journal of Sociology, 82 (3), 605–620.

Gomez-Lievano, A., Patterson-Lomba, O., & Hausmann, R. (2016). Explaining the prevalence, scaling and variance of urban phenomena. Nature Human Behaviour, 1 (1), 0012.

Gomez-Lievano, A., Youn, H., & Bettencourt, L. M. A. (2012). The Statistics of Urban Scaling and Their Connection to Zipf’s Law. PLoS ONE, 7 (7), e40393.

Groff, E. R. (2015). Informal Social Control and Crime Events. Journal of Contemporary Criminal Justice, 31 (1), 90–106.

Hall, J. (2016). Flint is the most dangerous city in America—But it has nothing to do with the water crisis. The Independent .

Hanley, Q. S., Khatun, S., Yosef, A., & Dyer, R.-M. (2014). Fluctuation scaling, Taylor’s law, and crime. PLoS ONE, 9 (10), e109004.

Hanley, Q. S., Lewis, D., & Ribeiro, H. V. (2016). Rural to urban population density scaling of crime and property transactions in English and Welsh Parliamentary Constituencies. PLOS ONE, 11 (2), e0149546.

Harrendorf, S. (2018). Prospects, problems, and pitfalls in comparative analyses of criminal justice data. Crime and Justice, 47 (1), 159–207.

Harrendorf, S., Heiskanen, M., & Malby, S. (Eds.). (2010). International statistics on crime and justice , Volume 64. Helsinki: European Institute for Crime Prevention and Control, affiliated with the United Nations.

Harries, K. D. (1981). Crime rates and environmental denominators. Journal of Environmental Systems, 11 (1), 3–15.

Hipp, J. R., & Steenbeek, W. (2016). Types of crime and types of mechanisms. Crime & Delinquency, 62 (9), 1203–1234.

Leitão, J. C., Miotto, J. M., Gerlach, M., & Altmann, E. G. (2016). Is this scaling nonlinear? Royal Society Open Science, 3 (7), 150649.

Louf, R., & Barthelemy, M. (2014). Scaling: Lost in the smog. Environment and Planning B: Planning and Design, 41 (5), 767–769.

Malleson, N., & Andresen, M. A. (2016). Exploring the impact of ambient population measures on London crime hotspots. Journal of Criminal Justice, 46 , 52–63.

Mayhew, B. H., & Levinger, R. L. (1976). Size and the Density of Interaction in Human Aggregates. American Journal of Sociology, 82 (1), 86–110.

Miethe, T., McCorkle, R., & Listwan, S. J. (2005). Crime profiles: The anatomy of dangerous persons, places, and situations (3rd edn.). Oxford University Press.

Miethe, T. D., Hughes, M., & McDowall, D. (1991). Social change and crime rates: An evaluation of alternative theoretical approaches. Social Forces, 70 (1), 165–185.

Oliveira, M., Barbosa-Filho, H., Yehle, T., White, S., & Menezes, R. (2015). From criminal spheres of familiarity to crime networks. In Complex Networks VI , pp. 219–230. Springer.

Oliveira, M., Bastos-Filho, C., & Menezes, R. (2017). The scaling of crime concentration in cities. PLOS ONE, 12 (8), e0183110.

Oliveira, M., & Menezes, R. (2019). Spatial concentration and temporal regularities in crime. arXiv preprint arXiv:1901.03589.

Oliveira, M., Ribeiro, E., Bastos-Filho, C., & Menezes, R. (2018). Spatio-temporal variations in the urban rhythm: the travelling waves of crime. EPJ Data Science, 7 (1), 29.

Pacheco, D. F., Oliveira, M., & Menezes, R. (2017). Using social media to assess neighborhood social disorganization: A case study in the United Kingdom. In Proceedings of the thirtieth international Florida artificial intelligence research society conference, FLAIRS 2017, Marco Island, Florida, USA, May 22–24, 2017. , pp. 341–346.

Park, H., & Katz, J. (2016). Murder rates rose in a quarter of the nation’s 100 largest cities. The New York Times .

Rotolo, T., & Tittle, C. R. (2006). Population size, change, and crime in U.S. cities. Journal of Quantitative Criminology, 22 (4), 341–367.

Sampson, R. J. (1986). Crime in cities: The effects of formal and informal social control. Crime and Justice, 8, 271–311.

Sampson, R. J. (1988). Local friendship ties and community attachment in mass society: A multilevel systemic model. American Sociological Review, 53 (5), 766.

Siegel, L. J. (2011). Criminology (11th ed.). Wadsworth Publishing.

Stults, B. J., & Hasbrouck, M. (2015). The effect of commuting on city-level crime rates. Journal of Quantitative Criminology, 31 (2), 331–350.

Takala, J.-P., & Aromaa, K. (2008). Victimology. In Encyclopedia of violence, peace, & conflict , pp. 2272–2288. Elsevier.

Thurner, S., Hanel, R., & Klimek, P. (2018). Introduction to the theory of complex systems . Oxford University Press.

Book   Google Scholar  

White, S., Yehle, T., Serrano, H., Oliveira, M., & Menezes, R. (2014). The spatial structure of crime in urban environments. In Social informatics , pp. 102–111. Springer.

Wirth, L. (1938). Urbanism as a Way of Life. American Journal of Sociology, 44 (1), 1–24.

Yang, V. C., Papachristos, A. V., & Abrams, D. M. (2019). Modeling the origin of urban-output scaling laws. Physical Review E, 100 (3), 32306.

Download references

No outside funding was used to support this work.

Author information

Authors and affiliations.

University of Exeter College of Engineering Mathematics and Physical Sciences, Exeter, UK

Marcos Oliveira

GESIS-Leibniz Institute for the Social Sciences, Cologne, Germany

You can also search for this author in PubMed   Google Scholar

Contributions

All authors read and approved the final manuscript.

Corresponding author

Correspondence to Marcos Oliveira .

Ethics declarations

Competing interests.

The authors declare that they have no competing interests.

Appendix 1: Results from the probabilistic scaling analysis

To test the plausibility of a nonlinear scaling, we compare each model against the linear alternative (i.e., \(\beta =1\) ) using the difference \(\Delta \mathrm{BIC}\) between the fits for each data set. We follow Leitão et al. ( 2016 ) and define three outcomes from this comparison. First, if \(\Delta \mathrm{BIC} < 0\) , we say that the model is linear ( \(\rightarrow\) ), since we can consider that the linear model explains the data better. Second, if \(0< \Delta \mathrm{BIC} < 6\) , we consider the analysis of \(\beta \ne 1\) inconclusive because we do not have enough evidence for the nonlinearity. Finally, if \(\Delta \mathrm{BIC} > 6\) , we have evidence in favor of the nonlinear scaling, which can be superlinear ( \(\nearrow\) ) or sublinear ( \(\searrow\) ). We also use \(\Delta \mathrm{BIC}\) to determine the model \(\mathrm{P}(Y|N)\) that describes the data better. In Tables  2 and Table  3 , we summarize the results in that a dark gray cell value indicates the best model based on \(\Delta \mathrm{BIC}\) , a light gray value indicates the best model given a \(\mathrm{P}(Y|N)\) model, and \(*\) indicates that the model is plausible ( p -value \(>0.05\) ).

Appendix 2: Data sources

See Table 4 .

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Oliveira, M. More crime in cities? On the scaling laws of crime and the inadequacy of per capita rankings—a cross-country study. Crime Sci 10 , 27 (2021). https://doi.org/10.1186/s40163-021-00155-8

Download citation

Received : 29 January 2021

Accepted : 08 September 2021

Published : 01 December 2021

DOI : https://doi.org/10.1186/s40163-021-00155-8

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Population size
• Urban scaling
• Complex systems

Crime Science

ISSN: 2193-7680

• General enquiries: [email protected]

Advertisement

Global Crime Patterns: An Analysis of Survey Data from 166 Countries Around the World, 2006–2019

• Original Paper
• Open access
• Published: 20 March 2021
• Volume 38 , pages 793–827, ( 2022 )

Cite this article

You have full access to this open access article

• Jan van Dijk   ORCID: orcid.org/0000-0002-3453-8568 1 ,
• Paul Nieuwbeerta 2 &
• Jacqueline Joudo Larsen 3  

32k Accesses

33 Citations

13 Altmetric

Explore all metrics

A Correction to this article was published on 27 July 2022

This article has been updated

This article explores the merits of commercially-based survey data on crime through cross-validation with established crime metrics.

Using unpublished data from 166 countries covering the period between 2006 and 2019, the article describes the geographical distribution across global regions and trends over time of three types of common crime, homicide, and organised crime. The article then explores possible determinants of the geographical distributions through regressing prevalence rates against indices of poverty, inequality, proportion of youth, presence of criminal opportunities (wealth and urbanisation), and governance/rule of law.

The results show that African and Latin American countries suffer from the highest levels of various types of crime across the board, followed by countries in Asia. European, North American and Australian countries experience intermediate or relatively low levels of most types of crime. Levels of common crime have dropped or stabilized globally except in Africa where they went up. Homicides have fallen almost universally. Trends in organised crime are diverging.

Conclusions

Dimensions of governance emerged as powerful determinants of levels of all types of crime. Important determinants of common crime besides governance were poverty, inequality, and proportion of youth. To some extent changes in these same characteristics of countries were found to be correlated with changes in levels of crime over the past fifteen years. The article concludes with a discussion of the study’s limitations and suggestions for further research.

Similar content being viewed by others

Crime Statistics in the European Union

The global crime drop and changes in the distribution of victimisation.

Crime drop in Ghana? Some insights from crime patterns and trends

Avoid common mistakes on your manuscript.

Introduction

In this article we explore the merits of the newly available, commercially produced crime data from the Gallup World Poll (GWP), a household survey carried out annually in over 160 countries since 2006, containing questions on experiences with theft as well as with violence (assaults and muggings) (Gallup World Poll 2020 ). Given that the GWP surveys are behind a pay-wall and thus not publicly available, GWP data on common crime have been analysed sparsely in the criminological literature. To our knowledge, so far only Corcoran and her colleagues have used GWP data to study the cross-national variation in experiences with violence (assaults and muggings) during the period 2006–2013 (Corcoran and Stark 2018 , 2020 ; Corcoran et al. 2018 ). In this article we will expand their analyses by also examining GWP data on experiences with another type of common crime, i.e. theft. Furthermore, we expand our analyses by analysing data over a longer period, i.e. the full period since the start of the GWP-survey (2006) and 2019.

We first describe the geographical distribution across global regions, sub-regions and countries in people’s experiences with theft and violence in 160 countries for the five-year periods 2006–2009, 2010–2014, 2015–2019 and for the entire 14-year period 2006–2019. Second, we describe the trends over time in countries’ levels of experienced thefts and violence, and examine whether these levels have increased, decreased or stayed stable over the 14-year period. Next, we will look at intercorrelations between prevalence rates of common crime (i.e. both theft and violence) and non-common crimes (homicide, and organised crime/corruption). Finally, we will use our dataset on the prevalence of these types of crime to examine whether and to what extent often used indicators of possible determinants of crime are correlated with levels and trends of both common and non-common crime across the world.

A Short History of Cross-National Studies on Crime

Studying cross-national variation in common crime has a long history. Throughout the nineteenth century criminologists have studied geographical variations in rates of recorded crime. A dominant theme in this early literature was the association of crime with poverty, urbanization and social disorganization. The traditional source of information on levels and trends of common crime were court statistics on convictions for criminal offences. In the twentieth century these court statistics were supplemented by statistics of crimes recorded by the police . Such statistics were later internationally collected by organisations like the United Nations and Interpol. However, comparing common crime recorded by police forces or courts cross-nationally is problematic, since countries, police forces and courts apply different legal definitions, work procedures and counting rules. Due also to varying readiness of citizens to report incidents of crime to the police and rates of detection, the proportions of crimes that remain hidden from the official records (the so-called dark numbers of crime) vary greatly across countries. Consequently, available information on crime recorded by the police or other criminal justice actors is known to be non-comparable across countries as a measure of crime.

To overcome the problems of internationally non-comparable court and police data on common crime, criminologists since the 1980s moved their international comparative research into two new directions: (a) analysing data on non-common crime, e.g. homicide, corruption and organised crime and/or (b) analysing data from national and international crime victimization surveys.

Cross-National Studies on Non-Common Crime

Using a purpose-built database focussing on homicide rates, Archer and Gartner ( 1984 ) gave new impetus to the field of cross-national criminological research. The general idea was that homicide is the type of crime that has the least cross-national differences in legal definitions and procedures, and that almost all incidents come to the attention of criminal justice systems and/or forensic doctors, in all countries. In the 1990s international homicide data became systematically available from the UNODC’s Global Study on Homicide and from the World Health Organization (WHO). These data today cover almost all countries in the world (UNODC 2019 ). When international homicide data became more readily available, a new wave of epidemiological studies using national data followed (Koeppel et al. 2015 ).

In addition, political scientists, economists and criminologists started to analyse other types of non-common crime. In particular, more recently, cross-national analyses have sought to understand determinants of global variation in measures of corruption and organised crime (see for review on corruption: Wysmulek 2019 ). This became possible, since several organisations (e.g. the World Economic Forum (WEF) and Transparency International (TI)) started to carry out surveys among business executives and/or households from many nations on experiences with corruption and organized crime (WEF 2019; TI 2019). So, for non-common crimes (homicide, corruption and organised crime) a growing variety of up-to-date, cross-nationally comparative data are available for a large number of countries around the world.

Crime Victimization Surveys

In order to circumvent the many uncertainties concerning crimes recorded by criminal justice actors as measures of crime, the USA and some European nations launched, in the 1970s, crime victimization surveys among households focussing on rates of victimization by offences as defined in the domestic criminal codes (Biderman and Reiss 1967 ). The results of these surveys provide an alternative, and in many respects more comprehensive, source of information about crime to those recorded by the police. However, national crime victimization surveys have different designs, data collection procedures and questionnaires in the different countries, thereby limiting the comparability of their results.

To overcome these problems, the International Crime Victim Survey (ICVS) was launched by criminologists with expertise in national crime surveys (Van Dijk et al. 1990 ; Lynch 2006 ). Since its initiation, surveys have been carried out once or more in altogether 85 countries (Van Kesteren et al. 2013 ). Over 350,000 citizens across the world have to date been interviewed with the same questionnaire, translated in at least thirty languages. For almost three decades these data have been the only source available on self-reported data on common crime covering multiple countries in several world regions. However, it must be recognized that the ICVS covers only about a third of all countries/territories in the world (and especially lacks data from Africa), most countries were only surveyed once or twice, and new surveys since 2010 have been few and far between.

Fortunately, the Gallup World Poll (GWP) with its much wider coverage and greater periodicity, contains questions on experiences with common crimes which can be harnessed for comparative international analysis.

Theoretical Perspectives on Cross-Country and Over-Time Variations in Crime

Throughout the nineteenth century statisticians and social scientists have studied geographical variations in rates of recorded crime. Pioneers such as Guerry (1802–1866), Quetelet (1796–1874), and Von Mayr (1841–1925) started comparative studies investigating and explaining differences in crime across European geographical areas. Later in the nineteenth century, Durkheim ( 1897 ) examined how socio-economic circumstances and changes impacted on European countries’ suicide and homicide levels. A dominant theme in this early ‘cartographic’ literature was the association of crime with, inter alia, poverty, urbanization and social disorganization (for an overview see Bonger 1916 ). Ecological research resurfaced in the beginning of the twentieth century, when scholars described the distribution of crime and delinquency, including homicide, across neighbourhoods and identified structural factors determining levels of crime (see e.g. Park and Burgess 1925 ; Shaw and McKay 1942 ).

Current empirical criminological studies explaining cross-national and over-time within country variation in crime rates show a broad range of different theoretical approaches. Three sets of factors emerge as the common ground of contemporary epidemiological studies seeking to explain variation in crime. The first set consists of motivational determinants, like poverty, unemployment, inequality and proportions of young people. A second set comprises opportunity-related factors like urbanization, community cohesion and wealth. And an emerging third set of determinants centres around the functioning of state and democratic institutions (governance).

Motivational Factors

Many scholars have theorized that countries’ crime levels are related to factors motivating persons to commit crime. Durkheim (2005 < 1897 >) already hypothesized that rapid social change creates normlessness or ‘anomie’ leading people to engage in deviant behaviour. In the same vein Merton’s strain/anomie theory argued that individuals facing economic hardship and blocked opportunities may experience feelings of injustice and resentment pressuring them to commit crime (Merton 1938 ).

Reviews of more recent empirical studies aiming to explain cross-national differences in crime confirm the importance of factors related to motivations to offend. In cross-national studies, poverty, economic inequality and, to a lesser extent, unemployment were found to be strongly correlated with homicide rates (LaFree 1999 ; Nivette 2011 ; Pare and Felson 2014 ; Lappi-Seppälä and Lehti 2014 ; Koeppel et al. 2015 ; UNODC 2019 ). Furthermore, analyses of ICVS data confirm moderately strong correlations between inequality and levels of victimization by various types of common crime (Nieuwbeerta 2002 ; Van Wilsem 2004 ; Van Dijk 2008 ). Structural correlations were also found between the percentage of young people in populations and levels of both common crime and homicides (Van Dijk 2008 ; LaFree and Tseloni 2006 ; McCall et al. 2013 ). In line with this finding, the existence of a ‘youth bulge’ has been linked to various forms of civil unrest, especially in the Global South (Urdal 2006 ).

Opportunity-Related Factors

A second set of determinants of crime prominent in current epidemiological studies are opportunity-related factors like urbanization, community cohesion and wealth. Several criminological theorists emphasised the role of criminal opportunities. For example, according to ‘routine activity theory’, crimes are more likely to occur at places where there are (1) more motivated offenders, (2) more suitable targets, and (3) fewer capable guardians (Cohen and Felson 1979 ). In this perspective, higher wealth increases the availability of easily ‘stealabe’ consumer goods and—especially in large urban settings—breeds social anonymity and weak social guardianship. This theoretical perspective also emphasised the role of the availability of alcohol, drugs and firearms to foster and facilitate crime. Criminal opportunity theory has also been used to explain the prevalence of various types of organised crime (Bullock et al. 2010 ).

Over the past decades, a large number of empirical studies have confirmed the epidemiological importance of factors related to criminal opportunities. Victimization surveys have shown that levels of common crime across the Western world went up in tandem with the availability of suitable targets, such as (poorly secured) motor vehicles (Van Dijk 2008 ; Farrell et al. 2014 ), and that a higher degree of urbanization and lower social cohesion are related to higher levels of common crime (Lee 2000 ; Van Wilsem 2004 ; Van Dijk 2008 ; Corcoran and Stark 2020 ).

A distinct third set of factors emerging in the literature explaining variation in countries’ levels of crime centres around the functioning of state and democratic institutions. Historically, the role of a (good functioning) state to prevent criminal and violent behavior was argued by the seventeenth century philosopher Thomas Hobbes. Furthermore, the importance of the role of the (emergence of) a nation-state for reducing crime is theorised in historical studies on homicide (Eisner 2014 ). In addition, political scientists and economists—especially from the New Institutionalist School—have increasingly pointed to the nefarious linkages between bad governance, corruption, violence and underdevelopment (Acemoglu and Robinson 2012 ; Wenmann and Muggah 2010 ; World Bank 2011 ; UNDP 2013 ; Acemoglu et al. 2017 ). From this perspective, poorer nations are caught in traps of weak state institutions, underdevelopment and high levels of crime and corruption (Kaufmann and Kraay 2002 ; Kaufmann et al. 2009 ).

The importance of the role of countries’ levels of ‘governance’ for their levels of violence and crime is also shown in several empirical studies of current crime levels. Van Dijk demonstrated how dysfunctional governance, organised crime and underdevelopment are intercorrelated at the country level (Van Dijk 2007 , 2008 ). In cross-national studies on homicide, various indicators of ‘poor governance’, rule of law or legitimacy emerged as key independent determinants of homicide rates (LaFree and Tseloni 2006 ; Nivette and Eisner 2013 ; Chu and Tusalem 2013 ; Lappi-Seppälä and Lehti 2014 ; Karstedt 2015 ; Stamatel 2016 ; Huebert and Brown 2019 ). Furthermore, bad governance was found to be one of the main dimensions of the Vulnerability Index for human trafficking/modern slavery grounded in GWP data (Joudo Larsen and Durgana 2017 ).

Data, Measures, and Indices

Gallup world poll data.

Two out of the broad range of Gallup World Poll’s many items relate to experiences with theft and violent crime respectively. We use these data from all Gallup World Poll surveys held in the 14-year period 2006–2019. The surveys are conducted in over 160 countries worldwide, making up more than 98% of the world’s adult population. The target population is the entire civilian, non-institutionalized adult (aged 15 + or 18 + years) population. The coverage area is the entire country including rural areas, and the sampling frame represents the entire country. The survey is translated into the primary languages of the countries and is then given to approximately 1000 individuals. The survey is conducted annually in most countries, but in some it is conducted once every two or three years—and in a few countries only every five or six years. The survey is conducted over the telephone in countries where at least 80% of the population has telephones by means of either random-digit dialling or a nationally representative list of phone numbers. Face-to-face interviewing is used in the other countries. More detailed information regarding the GWP’s methodology can be found at the website of the Gallup organisation ( www.gallup.com ).

To make the data optimally nationally representative and cross-nationally comparable, we apply the data weights as provided by the GWP. These weights account for oversamples, household size, national demographics (i.e. gender, age, and—where available—education or socioeconomic status), nonresponse, unequal selection probability, and design effects. In addition, we exclude all respondents under 18 years of age to ensure consistent age thresholds across countries. Footnote 1

Gallup Sample of Countries

Our sample of countries comprises all countries for which data on victimization of theft and assault are available in the Gallup World Poll over the years 2006–2019. This set of countries includes a total of 166 countries out of the 190 countries recognised by the United Nations during that period. Missing countries are all relatively small with the exception of North Korea. According to the geographical classification of the United Nations, the sample encompasses 40 countries from Europe, 2 from North-America, 26 from Latin America and the Caribbean, 49 from Africa, 47 from Asia and 2 from Oceania. Especially notable is the inclusion of 49 countries from Africa, a region conspicuously underrepresented in other datasets on crime, comprising of 17 countries from Eastern Africa, 7 from Middle Africa, 6 from Northern Africa, 5 from Southern Africa, and 14 from Western Africa (see Table 1 ).

Since in many years over the 14-year-period 2006–2019 questions on common crime are included in the GWP-questionnaire and Gallup interviews approximately 1000 residents per country annually, we have data on crime from 1780 so-called ‘country/years’. Since most countries are, as said, surveyed in multiple years, our measures for common crime are based on almost 1.8 million interviews, i.e. on average 10.6 thousand interviews per country (see Table 1 ).

Common Crime Measures

In our study we focus on two questions in the GWP-questionnaire pertaining to the actual experience of victimization by common crimes, i.e., whether a victim of theft (“Within the past 12 months, have you had money or property stolen from you or another household member?”) and whether a victim of violence, i.e., assault or mugging (“Within the past 12 months have you been assaulted or mugged?”). We constructed three measures on victimization of common crime based on these two questions. First, we used the yearly prevalence rate of persons who indicated they were a victim of theft—and calculated the average theft rate (in %) in each year in the 166 countries from which these data were available (N = 1780 country/years). Second, we did a similar exercise using the data on victimization of violent crime to calculate the violence ( assault/mugging) rate (in %) in the 165 countries and all years/years for which these data are available (N = 1315). Third, we used both GWP-questions together and calculated the rate of persons who were victim of theft and/or violence in all years and all 165 countries from which both questions are available (N = 1302). Footnote 2

Next, we cross-validated the resulting national prevalence rates from Gallup World Poll with estimates of similar types of crime from the last rounds of the International Crime Victim Survey (ICVS) (Van Dijk 2008 ). These last rounds of the ICVS, conducted between 1996 and 2010, cover many Western countries, i.e. Europe, North America, Australia, and New Zealand, several countries in Asia and Latin America and—in a more limited number—Africa. The comparison could be made for the 69 countries participating in both types of survey. Considering the many differences between the ICVS and GWP methodologies, including in formulation of questions on crime experiences, and in periods of data gathering (1996–2010 and 2006–2019 respectively) near perfect correlations between the ICVS and GWP victimization prevalence rates were not to be expected. Correlations were found to be moderately strong. For example, the Pearson correlations between the GWP and ICVS-based rates for various types of theft are 0.63 (N = 69) and for assault/muggings 0.67 (N = 68) respectively. Furthermore, the Pearson correlations between our combined GWP-based measure for theft and/or violence and the ICVS’s overall crime rate, covering all ten different types of crime used, is 0.62 (N = 69). The latter correlation suggests that the Gallup World Poll’s two broadly defined and partly overlapping items on theft and assault/muggings capture the volume of common crime in a country surprisingly well.

To monitor progress towards achieving crime-related Sustainable Development Goals the United Nations recommends the use of homicide statistics besides survey-based data on violence (UNODC 2019 ). The WHO collects data on homicide from public health sources through a dataset on the causes of mortality. The UNODC collects data primarily from criminal justice sources, although for some countries the UNODC relies on public health data. Where both criminal justice and public health sources exist, the two sources often provide similar results (Andersson and Kazemian 2018 ). In our sample the Pearson correlation between the average UNODC and WHO measures for the period 2006–2017 was 0.81 (N = 97 countries). We choose to use the more comprehensive UNODC dataset (UNODC 2019 ), available for 135 of the GWP-countries over the period 2006–2017 (see also Table 2 ).

Organised Crime

Organised crime is defined in the UN Convention on Transnational Organised Crime as the commission of ‘serious crimes by a structured group of three or more persons for material gain’. This definition has been specified by listing secondary characteristics such as high-level corruption, money-laundering, infiltration in the legal economy, and instrumental violence (Fijnaut and Paoli 2004 ). An item in the survey of the World Economic Forum (WEF) asks business executives, “In your country, to what extent does organized crime (mafia-oriented racketeering, extortion) impose costs on businesses?’ (Schwab 2015 ). This item captures infiltration into the legal economy through the perspectives of business executives. Answers to this question are strongly correlated to measures of corruption and of money-laundering and moderately to a proxy measure of mob-related violence consisting of country rates of unsolved homicides (Van Dijk 2007 ). For Africa, a new comprehensive index of organised crime harnessing experts’ assessments of the presence of various sorts of criminal markets and criminal actors in their country has been designed. This index was found to be strongly correlated to the WEF item on ‘costs on businesses’ (ENACT 2019 ).

To measure the extent of organised crime and related corruption, we combined measures from three widely available sources. The first measure is the item in the WEF survey just mentioned. The second measure is the Corruption Perceptions Index (CPI) of Transparency International (Transparency International 2015 ) which aggregates information from business people and country experts on the level of corruption. The third measure is an item in the GWP, which asks private respondents “Is corruption in government widespread?” The three measures are highly intercorrelated. Footnote 3 To construct an Organised Crime Index , the three measures were averaged, then standardised (with a mean of 0 and standard deviation of 1), and normalised Footnote 4 —so that they each range from 00 to 100 (with higher scores meaning more organised crime and corruption). We constructed this index by averaging the three normalised scores. Scores on this index could be calculated for 163 countries.

Determinants of Crime

In our review of studies on the possible determinants of crime (see above) we distinguish three main theoretical perspectives, respectively those focussing on motivations to offend, on opportunities of crime, and on the efficacy of state institutions (governance). To construct measures of the key determinants falling under these three headings, we selected indicators for which reliable data wasavailable for more than 150 of the 166 GWP countries and—to be able to study correlates of trends—for each of these countries for (almost) every year in the period 2006–2019. Footnote 5 The indicators encompassed three separate indicators of offender-related determinants (Poverty, Inequality, and % Youth), two indicators of opportunity-related determinants (Wealth and Urbanization) and one for Governance, which includes, inter alia, indicators of governance and the rule of law (See Table 2 for details).

Analytic Strategy

The first aim of our analyses is to describe the geographical distribution across global regions, sub-regions and countries in people’s experiences with crime in countries for the period 2006–2019. To do this, we use data for the entire 14-year period, but in order to take possible changes over time into account, also for the periods 2006–2009, 2010–2014, and 2015–2019 separately (see Table 3 ). Note that when calculating crime figures for (sub-) regions we first average on the data in each country, and then calculate the average over the countries in a (sub-)region. Footnote 6

The second aim is to describe developments in common crime between 2006 and 2019 in each of the 139 countries in the world for which we have sufficient Gallup World Poll (GWP) data available to adequately examine trends. Footnote 7 Since not for all countries survey data for all years are available, and we are interested in overall long terms trends, and not in yearly or short term fluctuations, we decided to use predicted values from OLS-linear regression models that were estimated for each country individually based on real values from available years. In these models, the dependent variable is the prevalence of crime (in %) in that country, and the independent variable is the calendar year. These models thus generate predicted levels of crime assuming a linear trend in each country. Footnote 8 We calculated the differences between the predicted levels of crime in the last year (i.e. 2019) and the first year (i.e. 2006) in each country as measure for the size of ‘linear change’. In other words: the trends are presented as “percentage point” i.e. the simple numerical difference between the predicted percentage in 2019 minus the predicted percentage in 2006 (see Tables 4 and 5 ). When presenting ‘linear change’ per (sub-)region (see also Tables 4 and 5 ), the measures of the countries in these (sub-)regions are averaged. Footnote 9

The third aim of the analyses in this paper is to explore whether and to what extent the patterns of cross-national and over-time variation in relevant country characteristics are related to levels and trends in countries’ levels of common and non-common crime. In these analyses we take three steps. As a first step we examine bi-variate correlations between (changes in) country characteristics and levels of and trends in common and non-common crime (see Table 7 ). Second, since many of the chosen possible determinants are highly correlated to each other, we have next run multi-variate ordinary least square (OLS) regression models in which the independent effects of (trends in) country characteristics are examined while controlling for the effects of (trends in) the other characteristics. OLS-regression models are run to explore effects of country characteristics on countries levels of crime (see Table 8 ), and separate models are run to explore effects of trends in country characteristics on trends in countries’ levels of crime (see Table 9 ).

As a third step of our analyses of the relationships between determinants of crime and (trends in) levels of crime, multi-level models were conducted wherein cross-sectional and trend analyses are done simultaneously. This strategy has the advantage that the full multi-country and multi-year dataset can be used in one model (for Theft there are a total of 1780 country/year observations, for Violence 1315 and for Theft and/or Violence 1302). These models have the additional advantage that when estimating the coefficients, separate residual components can be specified at the country- and year-level, and adjustments can be made for the correlation of the error components of the two levels. Since the results of the multi-level models fully correspond with the results of the separate cross-sectional and over-time OLS-regression analyses—and the latter analyses are more easily understandable—we focus on the OLS-regression results in the main text of the paper and present results of the multi-level models in “ Appendix B ”.

Describing Worldwide Patterns and Trends of Crime

Regional and sub-regional variation in common crime.

In this section we look at the distribution of victimisation rates for two separate types of common crime (theft and violence (assault/muggings)) and for the combined victimisation rate of theft and/or violence over six world regions and twenty-one sub-regions, including four individual nations (Canada, USA, Australia, and New Zealand). Table 3 summarizes the results for the world regions and sub-regions. It does this for the entire 14-year period 2006–2019, and, to check for consistency, for the periods 2006–2009, 2010–2014, and 2015–2019 separately.

Levels of common crime , expressed in one-year victimization percentages for theft and assault/muggings—and their combination—appear to vary substantially and consistently across regions and sub-regions around the world. Furthermore, the geographical distribution of levels of crime is largely uniform across theft and assault/muggings. Footnote 10 Generally, African and Latin American, including Caribbean countries appear to suffer most from these types of common crime across our measures. Africa ranks first for theft and violence. Especially Sub-Saharan Africa shows very high levels of both types of crime. Levels in North Africa are close to the global mean. Latin America ranks second for both theft and violence. Sub-regional variation is limited here. Asia, Europe, North America and Oceania (Australia and New Zealand) experience medium to high levels of both types of common crime with New Zealand experiencing the highest level of theft.

Two results stand out in the light of prior comparative research. First, the very high level of common crime across Sub-Saharan Africa. Prior research using ICVS data already indicated that populations in Sub- Saharan Africa experienced relatively high common crime rates—but that finding related to a much more limited number of countries and to an older period (mainly 1996–2000). Moreover, prior studies of GWP data (Corcoran and Stark 2018 , 2020 ) showed that levels of violent crime (i.e. of assault/muggings) were the highest in Sub-Saharan countries, but did not present data on levels of property crime. On average around a quarter of the population in Sub-Saharan Africa reported having their money or property stolen from them or another household member in the course of last year, and more than ten percent reported that they had been assaulted and/or mugged.

Second, the very high victimisation rates of property crimes in Latin America and the Caribbean stand out too. Prior research typically highlighted the high violence rates in Latin America (Corcoran and Stark 2018 , 2020 ). The current result shows that Latin American countries in an international perspective also experience very high victimisation rates for theft. A fifth of the populations in these countries reported having their money or property stolen from them or another household member in the past year. This too had already emerged from analyses of the ICVS datasets, but only for a smaller number of countries and for a less recent period (Van Dijk 2008 ).

Global Trends in Common Crime

Next, we look at the trends in victimisation rates of theft and violence (assault/muggings), and of the combined victimisation rate of theft and/or violence over six world regions and twenty-one sub-regions. Table 4 summarizes the results of the trend analyses, and presents average linear trend estimates (in percentage points) for all countries per world-region and sub-region. Footnote 11 As explained in the analytical strategy section above, these trend estimates represent the numerical difference between the predicted percentage in 2019 minus the predicted percentage in 2006. For example, the value of -4.1 of the trend estimate for Oceania for common crime (theft and/or violence) represents a decrease in estimated victimisation rates from 19.6 percent in the year 2006 to 15.5 percent in the year 2019—and thus a change of 19.6–15.5% = − 4.1 percent points (%p.). Footnote 12

The trend analysis of common crime around the world over the past fourteen years—as summarized in Table 4 —confirms prior studies showing that levels of common crime have, possibly due to improved security, been declining in Europe, Australia and New Zealand since the turn of the century (Van Dijk et al. 2012 ; Farrell et al. 2014 ). Our data show that since 2006 common crime has declined across Asia, except in Southern Asia, and across Latin America as well. Somewhat surprisingly, the GWP-data did not show a continuation of the much discussed ‘crime decline’ in the USA since 1995 (Zimring 2006 ). Estimated levels of theft and violence in the USA are roughly the same in 2019 as they were in 2006. The crime drop in the USA has set in some years earlier than elsewhere (Van Dijk 2008 ), and seems to have bottomed out sooner too. Footnote 13

Africa emerges as the only world region where crime has gone up since 2006 (with on average 11.6 percentage points). In 27 of the 35 African countries included in the trend analysis, levels of crime increased substantially over the period 2006–2019, i.e. more than five percentage points. Rises are most pronounced in Sub-Saharan countries.

To further examine the geographical distribution of trends in common crime we looked at the correlations between countries’ trends for theft, violent crime (assault/muggings) and common crime. The correlation between countries trends in national theft rates and rates of violence appeared to be moderate (r. = 47; n = 136). Footnote 14 When a country experiences an increase in levels of property crime it is somewhat more likely to experience an increase in levels of violent crime as well, and vice versa, but this is far from assured.

By way of summing up, Fig.  1 presents an overview of our findings on the distribution of both levels and trends in common crime across world regions, sub-regions and countries. It reveals at a glance the uniquely diverging position of Africa as the world region wherein most countries have experienced both the highest average prevalence during 2006–2019 and the most pronounced increases during this period. Latin America stands out as the world region with the second highest average prevalence but with a declining trend in most countries. In Asia prevalence is highest in South Asia. This is also the only Asian sub-region where the rate has in most countries gone up. Elsewhere in the world prevalence was relatively low during 2006–2019 and has declined or remained stable.

Prevalence and trends in common crime, 2006–2019 (Per country and region)

Regional and National Variation and Trends in Non-Common Crime

As discussed, criminologists in the 1980s switched their attention to the collection and analysis of international data on non-common crime, e.g. homicide, and corruption/organised crime. Now that data on common crime have also become more widely available, it is interesting to see to what extent levels and trends in common crimes are correlated to levels and trends in these non-common crimes. First, we will present the results on variations in homicides and organised crime.

Table 5 shows homicide rates calculated per 100,000 inhabitants based on the UNODC World Reports on Homicide from 2006 to 2017 for 142 countries. The world average rate per 100,000 is 8. Latin America is in a league of its own with an elevated rate of 23, with Central America leading with an average rate of 32. Africa comes in second place with a much lower regional rate of 8. This average is lifted upward by the exceptionally high rate of Southern Africa (24). In most other African regions rates of homicide lie far below the global mean. Rates of other world regions vary within a narrow range of 1 to 3. In Asia Central Asia stands out with a relatively high rate of 5. The lowest rates are found in Western Europe and in Oceania (Australia/New Zealand). Within the Western world, the USA shows a conspicuously high rate of 5.

Trend analysis of homicides (see also Table 5 ) could be adequately done for 121 countries worldwide. It shows a uniform modest decline across all world regions. Footnote 15 The only noticeable exception is Central America where the homicide rates went up by 1,5 points (i.e. per 100.000 inhabitants) (see also Alvazzi del Frate and Mugellini 2012 ).

Table 5 also shows the regional average scores on our composite Organised Crime/Corruption Index, which, as explained, includes business executives’ perceptions of mafia-type practices and two items on corruption in government, between 2007 and 2017. The findings display that the highest rates for exposure to organised crime/corruption are, once again, found in Africa and Latin America (both 68 on the 0–100 scale used), followed by Asia. Lowest rates are found in Western and Northern Europe and in Oceania. Within Europe, the levels are comparatively high in Eastern and Southern Europe (scores of 64 and 60 respectively). The level is relatively high in the USA too (49), especially when compared to that of Canada (27).

The trend data on organised crime show a small global increase with considerable regional variation. According to the index organised crime has gone up in North America (plus 6 percentage points), Australia (plus 9) and Latin America (plus 3). In Europe organised crime went down somewhat in Eastern Europe but up everywhere else. Trends were also divergent in Asia. Finally, organised crime went on average down in Africa with Northern Africa, where the level went up with 11 percentage points, being a clear-cut exception.

Intercorrelations of Common and Non-Common Crime

We have earlier reported on the strong cross-sectional associations between national rates for three types of common crime (r = 0.85 or more). Our analyses showed that rates of homicide and organised crime are moderately correlated with each other (r = 0.41; N = 136). Some degree of correlation was to be expected since, as discussed, instrumental violence is seen as one of the defining characteristics of organised crime.

To examine the geographical association of common and non-common crime, we calculated the Pearson correlations between the country rates for theft, violence (assault/muggings), and their combination (theft and/or violence), and our measures for homicide and organised crime (See Table 6 ).

As can be seen in Table 6 rates of common crime show weak to moderate correlations with rates of non-common crime. Correlation coefficients between our measure for common crime (theft and/or violence) and homicide and organised crime are 0.34 and 0.45 respectively. So, if certain countries experience high levels of theft and/or violence, they are only somewhat more likely to be exposed to high levels of homicide and/or organised crime too. Although the correlations are only moderately strong, it seems worth noting that countries in Africa and Latin-America rank highest for all types of common and non-common crime alike.

Next, we looked at the correlations between trends in common and non-common crime. Different from the correlations between levels of common crime, those between estimated linear change measures of common crime and non-common crime are non-existent or very weak. There is, for example, only a very low correlation between trends in common crime and those in homicide rates (r. = 0.18). Our trend data on common and non-common crime thus suggest that common and non-common types of crime show considerable divergence in their movements over time.

By and large, our epidemiological crime data show that countries’ levels of different types of common crime show roughly similar geographical distributions and trajectories over time but that this does not, or to a much lesser degree, hold for types of non-common crime. Homicides and organised crime show geographical patterns and trajectories distinctly different from those of common crime. This finding suggests that determinants of crime at the macro level may be different for common than for non-common types of crime. Whether this supposition is valid, we will explore in the next section.

Testing Explanations for Worldwide Patterns and Trends of Crime

Determinants of cross-national variation in crime.

As a first step in our explorative analysis of possible determinants of various types of crime, we examine bi-variate correlations between country characteristics (averaged over the period 2006–2019) and levels of common and non-common crime (also averaged over 2006–2019) (see Table 7 ). As can be seen in this table, all six determinants are significantly correlated with most or all five measures of crime with homicide as exception (with rates being uncorrelated to poverty and urbanization).

Two sets of indicators are correlated with crime in the expected direction. First, the motivational determinants, Poverty, Inequality, and Youth present an unambiguous picture. In line with offender-related hypotheses, poorer, more unequal, and younger populations experience higher levels of all five types of crime, except homicide. Second, our composite index of governance appears to be inversely related to all five types of crime, most strongly so with organised crime (r = − 0.73). However, contrary to expectation, the correlations between the two opportunity-related determinants and all five types of crime are reversed, suggesting that levels of crime are significantly lower in more urbanized, affluent societies.

Since many of the chosen independents including poverty and wealth are highly correlated to each other, a multi-variate analysis is obviously called for to examine the independent relationships with crime.

As a next step in our explorative analysis of possible determinants of various types of crime, we have run multi-variate OLS-regression models in which the independent effects of country characteristics are examined while controlling for the effects of the others. Such analyses are not unproblematic, since several of our country characteristics are so strongly inter-correlated that problems of multi-collinearity arise when running ordinary regression models. In particular, our index of governance is very strongly inter-correlated with our set of independents representing offender-related factors (Poverty, Inequality and Proportion), and, though to a lesser degree, with opportunity-related factors (Wealth and Urbanisation) as well. As a remedy, we have introduced dummy variables in our regression models representing three groups of countries: failed states (with governance scores < 1.5, N = 26), frail states (governance 1.5–8.5, N = 96), and effective states (governance > 8.5, N = 36) (as reference group) respectively—and centred the values of the other independent variables around their means of each of these groups separately. In this way we have analysed the correlations of crime with the variance in the other independents among countries with similar degrees of governance (a fixed effects model). Footnote 16

Doing this not only overcomes the ‘technical’ multi-collinearity problem. It also follows the theoretical ideas emerging from the New Institutionalist School in economics about the close linkage between bad governance and underdevelopment (Acemoglu and Robinson 2012 ). Good governance of states is regarded as the principal driver of sustainable development—e.g. less poverty and inequality. As can be seen in “ Appendix A ”, our dataset supports this theoretical notion. The results of the regression analyses are presented in Table 8 .

The results, first of all, show clear and consistent inverse relationships between the countries’ level of (good) governance, and the measures of all five types of crime. The estimated parameters for each of the two country groups represent the average crime rates of the ‘failed states’ and ‘frail states’ compared to the ‘effective states’. The results show that failed and frail states suffer from significantly more crime problems overall than more effective states. The importance of good governance is most pronounced for organised crime. These results fully confirm the institutionalist perspective: bad governance is linked not only to underdevelopment but to various forms of insecurity as well.

The second main finding is that the motivational, offender-focussed factors of poverty, inequality, and proportion of youth are important predictors of common crime, even after controlling for the effects of different levels of governance. Apparently, classical criminological notions about poverty-related, motivational causes of common crime have at the global scale lost none of their pertinence.

Third, in the multi-variate analysis the urbanization indicator is positively related to common crime, as predicted by criminal opportunity theory: anonymous urban environments breed more crimes by ‘opportunistic’ offenders (Felson and Cohen 1980 ). However, contrary to expectations, wealth is not positively related to theft or violent crime. The absence of a significant relationship between wealth and crime is probably due to the dual and opposing impact of wealth on common crime. While affluence brings a larger supply of suitable targets, its criminogenic impact, predicted by opportunity theory, may be offset by lower numbers of motivated offenders. In addition, above average investments in security measures in more affluent societies may have reduced their vulnerability for opportunistic property crimes (Van Dijk 2008 ; Farrell et al. 2014 ).

Finally, for non-common crime the results are less clear in several respects. A country’s percentage of youth in the population is a predictor of homicide, as suggested by the literature. And a nation’s level of inequality is associated with organised crime. However, poverty, wealth and urbanization are not associated with either homicide or organized crime. The latter findings hint at the complex, still poorly understood interrelationships between governance, organised crime and wealth (ENACT 2019 ).

Determinants of National Trends in Crime

The availability of trend data on the five types of crime for the period 2006–2019 allows an explorative test of the relevance of the determinants of crime for explaining changes over time in levels of common and non-common crime. Possibilities for such test are of course conditional on the availability of trend data of the independents as well. For this paper we use the time series data available on all six indicators included in the cross-national analyses (see Table 2 ).

Using these data, we generated predicted values for each indicator from OLS-linear regression models that were estimated for each country individually based on real values from available years (similar to what we did to obtain trend estimates for crime). Next, we calculated the differences between the predicted scores in the last year (i.e. 2019) and the first year (i.e. 2006) in each country as measure for the size of ‘linear change’. Subsequently, we examine the extent to which these linear trends in country characteristics are related to linear trends in crime. Again, as in the cross-sectional analyses, bivariate analyses (not shown here) were followed by OLS-regression analyses (see Table 9 ). Footnote 17

The results show that global variation in trends in crime cannot be explained as adequately by the determinants included in the analyses as cross-national variation. The explained variance of trends in crime is much smaller (i.e. less than 31% for violence and homicide, and around 50% for the other types of crime) than of cross-sectional variation (47% for homicide and 72% or more for the other crime types). Furthermore, several of the estimated multivariate correlates are statistically insignificant.

In line with the findings of the cross-sectional analyses presented above, positive correlations were found between changes in percentage youth and trends in theft, violence and in these crimes combined. In countries with increasing proportions of (marginalised) young various types of crime have gone up.

Contrary to our cross-sectional finding that poverty is associated with more common crime, negative correlations were found between changes in poverty and changes in common crimes, especially thefts. This could either mean that countries with increasing poverty are experiencing less common crime, or that countries with decreasing poverty experience more crime. Since changes in our measure of wealth are not significantly correlated with changes in levels of common crime, the finding that less poverty goes together with more thefts calls for further scrutiny.

Finally, trends in governance-related factors appear to be associated with our measure of organised crime/corruption but not with other types of crime. Trends in our index of organised crime were found to be inversely correlated with changes in the quality of Governance/Rule of Law and Wealth (GDP). In line with the institutionalist perspective, the increases in organised crime/corruption in some parts of the world seem to have gone together with deteriorations in the functioning of state institutions, less economic growth, and more poverty.

Discussion and Limitations

In recent years polling companies have started to supply survey-based datasets on levels of various types of common and non-common crime across the world. Cross-validation with results from in-depth criminological studies such as the ICVS has shown encouraging results in this study. Country prevalence of common criminality can apparently be reasonably well estimated with just two catch-all items on theft and assault/muggings in national sample surveys. Credible data on (recorded) homicides and perceptions of organised crime and corruption have also become available.

The datasets used on prevalence of thefts, assaults/muggings, common crime overall, (recorded) homicides, and organised crime/corruption, covering 166 countries worldwide, shows huge variation across regions and sub-regions. By far the highest levels of common crime are experienced by populations in the Global South, most notably in Africa—especially sub-Saharan Africa- and Latin America. Levels of homicide are highest in Latin America, particularly in Central America and the Caribbean, followed by Southern Africa. Organised crime shows, once again, the highest concentrations across Africa and Latin America, and in Asia. Within the Western world, Eastern and Southern Europe and the USA stand out with comparatively high scores on organised crime/corruption.

By and large, the geographical distribution reveals the existence of a deep North–South Security Divide in the beginning of the twenty-first century. Over the past fifteen years this gap has widened by the continued rise in levels of common crime in Africa and South Asia and of homicides in Central America while almost everywhere else these forms of crime have dropped. Trends in organised crime, including grand corruption, show a somewhat different picture with decreases in sub-Saharan Africa, and increases in Latin and North America, Australia and parts of Europe.

Differences in the geographical distribution of different types of crime and divergent trends in these types of crime between 2006 and 2019 shed doubt on general theories about the macro-causes of crime. This is confirmed by our explorative analyses of relationships between determinants of crime and levels of various types of crime. Traditional ideas about ‘root causes of crime’ such as poverty and inequality emerged strongly in regression analyses of levels of common crime but less so in those of homicides and organised crime. Apart from these common causes of crime, levels of common crime seem to be co-determined by the availability of easy targets of theft and reduced social control in urban environments. Urbanization, however, is totally unrelated to levels of homicides or organised crime.

Our explorative analyses furthermore point at the pervasive impact of governance-related factors on crime levels: prevalence of all types of crime is significantly higher to the extent that state and democratic institutions are weaker. This appears to be true not only, as was to be expected from previous studies, for homicide (La Free and Tseloni 2006 ; Chu and Tusalem 2013 ; Karstedt 2015 ) and organised crime (Van Dijk 2007 ; ENACT 2019 ), but, to some extent, for common crime as well.

A subsequent analysis of the correlates of trends over time of common crime confirmed the role of governance-related factors in explaining variation across countries in the movement of various types of crime. Besides, changes in common crime over the past fifteen years were strongly related to ongoing urbanization, increased proportions of young and growing inequalities. The exceptional, and pronounced increases in common crime in Sub-Saharan countries, discussed above, are likely to have been driven by expanding ‘youth bulges’ across the region.

The findings concerning governance conform to, and elaborate on the New Institutionalist School in comparative economics. The institutional capacity of countries seems to be the principal driver not just of sustainable development but of internal security as well. Between economic stagnation and organised crime appear to exist nefarious, mutually reinforcing relationships (Acemoglu et al. 2017 ). Linkages between governance, development and common crime, partly mediated by governance-related factors as (extreme) poverty, inequality, and high proportions of (marginalised) youth, are close as well. Many poorer nations seem in the grip of vicious circles of poorly functioning state institutions, underdevelopment, expanding youth bulges and accompanying high levels of all sorts of crime (Van Dijk 2008 ; Wenmann and Muggah 2010 ; World Bank 2011 ; UNDP 2013 ). Our results underline the importance of reducing levels of crime and violence as integral part of the United Nation’s 2030 Agenda for Sustainable Development.

Limitations

For our analyses we have used readily and widely available international measures of crime and of the main possible determinants of crime mentioned in criminological literature. This reliance on available data has introduced important limitations to our analysis. The results show that the chosen core determinants explain a considerable part of cross-sectional variance in prevalence of common crime and organised crime but less so in that of homicides. Their explanatory power for medium term trends over time proved to be limited for all types of crime. Unexplained cross-sectional and over time variance may be related to important criminogenic factors omitted in our analysis, both general and crime-specific ones. Examples of the latter are gender inequality, alcohol abuse and firearm possession as drivers of violent crime and homicides. Special cultural factors have remained altogether unexplored in our analyses. Subsequent studies should widen the choice of correlates and seek to specify relevant aspects of broadly operationalised variables as inequality, youth or ‘bad governance’.

Future international surveys or other studies will hopefully provide more refined measures of common crime, homicides, organised crime and other types of crime. These should then, ideally, be regressed against tailor-made operationalisations of possible determinants informed by current criminological theory. Such focussed cross-sectional and longitudinal analysis would bring research into the macro causes of crime to the next level. Even such improved correlational analysis would not, however, allow drawing conclusions on causality. Other types of research are required for causal inference. Correlational analysis should be complemented by studies focussing on crime trends in countries affected by major and sudden upheavals due to external forces like war, natural disasters or pandemics. A pertinent example would be an analysis of crime trends in countries where conflicts have disrupted the normal functioning of state institutions and/or the economy.

Change history

27 july 2022.

A Correction to this paper has been published: https://doi.org/10.1007/s10940-022-09553-w

We also did sensitivity analyses by excluding countries in which the GWP-documentation indicated that less than 85% of the population was represented (e.g. due areas where the safety of interviewing staff is threatened, scarcely populated islands in some countries, and areas that interviewers can reach only by foot, animal, or small boat). The results of these sensitivity analyses are very similar to the analyses using all countries. We therefore do not show the results of the sensitivity analyses in this paper. Furthermore, we also test for the robustness of the results by excluding countries with fewer than 3 years of GWP data. Again, these results were very similar to the analyses on all countries (not shown in this paper).

We recognise that the measure for common crime combines a household rate for theft and a personal rate for assault/mugging. In reports on the ICVS rates for personal and household crimes are similarly combined into an overall prevalence rate for victimization experiences as individual persons and/or household member (burglaries and car thefts). In this respect the GWP-based common crime rate and the ICVS over all prevalence rate are fairly similar.

The correlations between the measures are: WEF-TI .70 (n = 139), WEF-Gallup .51 (N = 139), TI-Gallup: .54 (N = 138). The reliability index of the scale is: .81 (N = 1.496).

Standardised: with mean of 0 and standard deviation of 1, and Normalised: Normalised Value = 1 + (Value – minimum value)*(100–1)/(maximum value – minimum value).

In early stages of our explorative, cross-sectional analyses we looked at a larger selection of possible determinants including unemployment, social disorganisation (operationalized as rapid changes in wealth), alcohol consumption and possession of firearms. Unemployment and social disorganisation appeared to be unrelated to our main measures of crime. Alcohol consumption was found to be significantly related to levels of violence and homicides in multivariate analyses and so was firearm possession to homicides. Our initial operationalisation of Governance was a scale of nine indicators taken from the World Bank Governance Index, the Fragile State Index and the WEF surveys (police performance). In our final analyses we have included only one of these indicators, the FSI P1 indicator of State Legitimacy. Results concerning the role of governance of earlier multi-variate analyses using the composite scale were identical to the ones presented here.

When doing this, we gave each country the same weight—regardless of number of inhabitants of a country or the number of respondents in Gallup World Poll in each country.

We included 139 countries for which data is available from at least one year in the period 2006–2010 and one year in the period 2015–2019.

The models are as follows: Crime ij  = γ i D i  + β i Year ij  + µ ij , with Crime ij —i.e. the level of crime in each country i and year j—being the dependent variable, D i dummies for all the countries, and Year ij the Years (from 2006 thru 2019, and β i the countries’ trend parameters. We centred the Year ij -variable around the value in the first year of the observation period, i.e. 2006.

The Pearson correlations between the countries’ average prevalence rates for Theft and Violence in the period 2006–2019 is 0.85, for Theft and ‘Theft and/or Violence’ 0.97 and for Violence and ‘Theft and/or Violence’ 0.95.

Note that the trends reported in Table 4 do not necessarily match with the figures reported in Table 3 . For example, Table 3 seems to suggest a stable overall crime rate in Asia between 2006/10 and 2015/19, while table 4 reports a decline by 1% point (across the 14-year period). These differences result from the different ways in which the analyses were done: Table 3 presents average levels of crime victimization rates in the counties per (sub-region) for the periods, 2006/10, 2011–2014, and 2015/19, whereas Table 4 present average linear trend estimates (in percentage points) for all countries per world-region and sub-region (see Sect.  4.4 . Analytical strategy). Moreover, the trend analyses presented in Table 4 are based on a more restricted number of countries, i.e., only countries for which data is available for at least one year in the period 2006/10 and one in the period 2015/19. Nevertheless, the overall pictures of Tables 3 and 4 are very similar.

Readers might wonder why no statistical significance levels are presented for the trend parameters. This is since the presented average linear change estimates per (sub-) region result from averaging estimates of country level OLS-regression analyses—and not from regression analyses per (sub-)region (see Sect.  4.4 . Analytical Strategy). Moreover, we preferred to focus on whether the trends in crimes are substantive (in percentage points) rather than on statistical significance (see also Fig.  1 ).

Rates of violent crime in the USA have been more or less stable since 2003 according to the NCVS (Morgan and Truman/BJS, 2020 ). Rates for property crimes are not comparable with those of GWP.

Correlations between trends in theft and violence and trends in the composite rate for common crime were – logically – very high (r. = 90 and .80 respectively).

Similar to the trend analyses for common crimes, we used predicted values for homicide and organised crime from OLS-linear regression models that were estimated for each country individually based on real values from available years. In these models, the dependent variable is the prevalence of homicide or organised crime in that country, and the independent variable is the calendar year. These models thus generate predicted levels of these type of crimes assuming a linear trend in each country. We calculated the differences between the predicted levels of crime in the last year (i.e. 2017) and the first year (i.e. 2006 for homicide and 2007 for organised crime) in each country as measure for the size of ‘linear change’.

For all regression models, both for cross-national variation and variation in trends, the variance inflation factors (VIFs) are under 3, suggesting that multicollinearity is not a problem. We have also examined the data for potential outliers (e.g. using residuals) and rerun all models excluding the possibly influential outliers from the analyses (the pertaining (number of) countries excluded differ across types of crime – but are not more than 9). These extra analyses (results not shown) yielded very similar or identical results.

Since the correlations between the linear trends in the determinants are low or at most moderate, and there is no problem of multi-collinearity, we did not include dummies for failed and fragile states in the regression models.

Acemoglu D, Robinson JA (2012) Why nations fail: the origins of power, prosperity, and poverty. Crown, New York

Google Scholar  

Acemoglu D, De Feo G, De Luca G (2017) Weak states: causes and consequences of the Sicilian Mafia, MIT Department of Economics Working Paper No. 17-10

Alvazzi del Frate A, Mugellini G (2012) The crime drop in ‘non-western’countries: a review of homicide data. In: Dijk S et al (eds) The international crime drop: new directions in research. Palgrave MacMillan, London, pp 134–155

Chapter   Google Scholar  

Andersson C, Kazemian L (2018) Reliability and validity of cross-national homicide data: a comparison of UN and WHO data. Int J Comp Appl Criminal Justice 42:287–302

Article   Google Scholar  

Archer D, Gartner R (1984) Violence and crime in cross-national perspective. Yale University, New Haven

Biderman AD, Reiss AJ Jr (1967) On exploring the “Dark Figure” of crime. Annals 374:1–15

Bonger W (1916) Crime and economic conditions. Little Brown, Boston

Bullock K, Clarke RVG, Tilley N (2010) Situational prevention of organised crime. Willan, Cullompton

Chu DC, Tusalem RF (2013) The role of the state on cross-national homicide rates. Int Crim Justice Rev 23:252–279

Cohen LE, Felson M (1979) Social change and crime rate trends: a routine activity approach. Am Sociol Rev 44:588–608

Corcoran KE, Pettinicchio D, Robbins B (2018) A doubled-edged sword: the countervailing effects of religion on cross national violent crime. Soc Sci Q 99(1):378–389

Corcoran KE, Stark R (2018) Culture, region and cross-national violent crime. Sociol Forum 33(2):310–333

Corcoran KE, Stark R (2020) Regional, structural, and demographic predictors of violent victimization: a cross-national, multilevel analysis of 112 countries. Int Rev Victimol 26(2):1–19

Dijk JJM (2007) Mafia markers: assessing organized crime and its impact upon societies. Trends Organ Crime 10(4):39–56

Durkheim E (1897 (2005)) Suicide: a study in sociology, translated by J. A. Spaulding and G. Simpson. Routledge, London

Eisner M (2014) From swords to words: does macro-level change in self-control predict long-term variation in levels of homicide? Crime Justice 43(1):65–134

ENACT (2019) Organised Crime Index Africa. www.ocindex.net

Fijnaut C, Paoli L (2004) (eds) Organised crime in Europe. Concepts, patterns and control policies in the European Union and beyond. Springer Press, Dordrecht

Farrell G, Tilley N, Tseloni A (2014) Why the crime drop? Crime Justice 43(1):421–490

Felson M, Cohen LE (1980) Human ecology and crime: a routine activity approach. Hum Ecol 8:389–405

Fund For Peace (2017) Fragile State Index and CAST methodology guide. https://fragilestatesindex.org/2017/05/13/fragile-states-index-and-cast-framework-methodology/

Gallup World Poll (2020) Methodology consulted on March 16, 2020. www.fao.org/fileadmin/templates/ess/voh/Gallup_world_poll_methodology.pdfGallup . (2008)

Huebert ET, Brown DS (2019) Due process and homicide: a cross-national analysis. Polit Res Quart 72:190–204

Joudo Larsen J, Durgana DP (2017) Measuring vulnerability and estimating prevalence of modern slavery. Chance 30(3):21–29

Karstedt S (2015) Does democracy matter? Comparative perspectives on violence and democratic institutions. Eur J Criminol 12:457–481

Kaufmann D, Kraay A (2002) Growth without governance. World Bank Policy research working paper no. 2928

Kaufmann D, Kraay A, Mastruzzi M (2009) Governance matters 2009: learning from over a decade of the worldwide governance indicators, online: Brookings.edu

Koeppel MDH, Rhineberger-Dunn GM, Back KY (2015) Cross-national homicide: a review of the current literature. Int J Comp Appl Criminal Justice 39(1):47–85

LaFree G (1999) Summary and review of cross-national comparative studies of homicide. In: DwayneSmith M, Zahn MA (eds) Homicide: a sourcebook of social research. SAGE, Thousand Oaks, pp 125–145

LaFree G, Tseloni T (2006) Democracy and crime: a multilevel analysis of homicide trends in forty-four countries, 1950–2000. Ann Am Acad Polit Soc Sci 605(1):25–49

Lappi-Seppälä T, Lehti M (2014) Cross-comparative perspectives on global homicide trends. In: Tonry M (ed) Why crime rates fall and why they don’t. Uni of Chicago Press, Chicago

Lee M (2000) Community cohesion and violent predatory victimization: a theoretical extension and cross-national test of opportunity theory. Soc Forces 79(2):683–706

Lynch J (2006) Problems and promise of victimization surveys for cross-national research. Crime Justice 34(1):229–287

McCall PL, Land KC, Brooks Dollar C, Parker KF (2013) The age structure-crime rate relationship: solving a long-standing puzzle. J Quant Criminol 29:167–190

Merton RK (1938) Social structure and anomie. Am Sociol Rev 3:672–682

Morgan RE, Truman JL (2020) Criminal victimization, 2019. Bureau of Justice Statistics. ( https://www.bjs.gov/index.cfm?ty=pbdetail&iid=7046 )

Nieuwbeerta P (ed) (2002) Crime victimization in comparative perspective. Results from the international crime victims survey, 1989–2000. Boom Juridische Uitgevers, Den Haag

Nivette A (2011) Cross-national predictors of crime: a meta-analysis. Homicide Stud 15(2):103–131

Nivette AE, Eisner M (2013) Do legitimate polities have fewer homicides? A cross-national analysis. Homicide Stud 17(1):3–26

Pare PP, Felson R (2014) Income inequality, poverty and crime across nations. Br J Sociol 65:434–458

Park RE, Burgess EW (1925) (Eds) The city. University of Chicago Press, Chicago

Schwab K (2015) The global competitiveness report 2015–2016. Consulted on March 15, 2020. http://www3.weforum.org/docs/gcr/2015-2016/Global_Competitiveness_Report_2015-2016.pdf

Shaw CR, McKay HD (1942) Juvenile delinquency and urban areas. University of Chicago Press, Chicago

Stamatel JP (2016) Democratic cultural values as predictors of cross-national homicide variation in Europe. Homicide Stud 20(3):239–256

TI/Transparancey Intenational (2020) Corruption perception index (CPI), 2020

Transparency International (2015) Corruption perceptions index 2015: data and methodology. Consulted on March 15, 2020. https://www.transparency.org/cpi2015

UNODC (2019) Global study on homicide, 2019. UNODC, Vienna

United Nations Development Program (2013) Citizen security with a human face, evidence and proposals for Latin America; regional human development report 2013–2014. UNDP, New York

United Nations (2020) Population prospects (2020). https://population.un.org/wpp/

Urdal H (2006) A clash of generations? Youth bulges and political violence. Int Stud Quart 50(3):607–629

van Wilsem J (2004) Criminal victimization in cross-national perspective: an analysis of rates of theft, violence and vandalism across 27 countries. Eur J Criminol 1(1):89–109

van Dijk JJM (2008) The world of crime: breaking the silence on problems of security, justice and development across the world. Sage Publication Inc, Thousand Oaks

van Dijk JJM (2015) The case for survey-based comparative measures of crime. Eur J Criminol 12(4):437–456

van Dijk JJM, Mayhew P, Killias M (1990) Experiences of crime across the world; key findings from the 1998 international crime survey. Kluwer Law and Taxation, Deventer

van Dijk JJM, Tseloni A, Farrell G (eds) (2012) The international crime drop: new directions in research. Palgrave MacMillan

van Kesteren JN, van Dijk JJM, Mayhew P (2013) The international crime victims surveys: a retrospective. Int Rev Victimol 20(1):49–69

WEF (World Economic Forum) (2020) Executive opinion survey, 2020. World Economic Forum, Geneva

Wenmann A, Muggah R (2010) More violence, less development. Geneva Declaration Secretariat, Geneva. http://www.genevadeclaration.org/fileadmin/docs/MDG_Process/MoreViolenceLessDevelo pment.pdf

World Bank (2011) World development report 2011: conflict, security and development. World Bank, Washington, DC

World Bank (2020) World development indicators database. https://databank.worldbank.org/source/world-development-indicators

Wysmulek I (2019) Using public opinion surveys to evaluate corruption in Europe: trends in the corruption items of 21 international survey projects, 1989–2017. Qual Quant 53:2589–2610

Zimring FE (2006) The great American crime decline. Oxford University Press, Oxford

Book   Google Scholar  

Download references

Acknowledgements

The authors express their gratitude to the Minderoo Foundation and Gallup Inc. for allowing access to data from Gallup World Poll.

Author information

Authors and affiliations.

Netherlands Study Center for Crime and Law Enforcement, Amsterdam, The Netherlands

Jan van Dijk

Department of Criminology, Leiden University, Leiden, The Netherlands

Paul Nieuwbeerta

Walk Free, An Initiative of the Minderoo Foundation, Perth, Australia

Jacqueline Joudo Larsen

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Jan van Dijk .

Ethics declarations

Conflict of interest.

The authors declare that they have no conflict of interest.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The original online version of this article was revised: The incorrect data from the Tables 1, 3, 4, 7, 8, 9 and Appendix B has been corrected.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Supplementary file1 (XLSX 35 kb)

See Table 10 .

See Table 11 .

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

van Dijk, J., Nieuwbeerta, P. & Joudo Larsen, J. Global Crime Patterns: An Analysis of Survey Data from 166 Countries Around the World, 2006–2019. J Quant Criminol 38 , 793–827 (2022). https://doi.org/10.1007/s10940-021-09501-0

Download citation

Accepted : 19 February 2021

Published : 20 March 2021

Issue Date : December 2022

DOI : https://doi.org/10.1007/s10940-021-09501-0

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• International
• Geographical distribution
• Time series
• Determinants
• Common crime
• Organised crime
• Find a journal
• Publish with us
• Track your research

The Official Journal of the Pan-Pacific Association of Input-Output Studies (PAPAIOS)

• Open access
• Published: 05 June 2020

Dynamic linkages between poverty, inequality, crime, and social expenditures in a panel of 16 countries: two-step GMM estimates

• Muhammad Khalid Anser 1 ,
• Zahid Yousaf 2 ,
• Abdelmohsen A. Nassani 3 ,
• Saad M. Alotaibi 3 ,
• Ahmad Kabbani 4 &
• Khalid Zaman 5  

Journal of Economic Structures volume  9 , Article number:  43 ( 2020 ) Cite this article

164k Accesses

58 Citations

254 Altmetric

Metrics details

The study examines the relationship between growth–inequality–poverty (GIP) triangle and crime rate under the premises of inverted U-shaped Kuznets curve and pro-poor growth scenario in a panel of 16 diversified countries, over a period of 1990–2014. The study employed panel Generalized Method of Moments (GMM) estimator for robust inferences. The results show that there is (i) no/flat relationship between per capita income and crime rate; (ii) U-shaped relationship between poverty headcount and per capita income and (iii) inverted U-shaped relationship between income inequality and economic growth in a panel of selected countries. Income inequality and unemployment rate increases crime rate while trade openness supports to decrease crime rate. Crime rate substantially increases income inequality while health expenditures decrease poverty headcount ratio. Per capita income is influenced by high poverty incidence, whereas health expenditures and trade factor both amplify per capita income across countries. The results of pro-poor growth analysis show that though the crime rate decreases in the years 2000–2004 and 2010–2014, while the growth phase was anti-poor due to unequal distribution of income. Pro-poor education and health trickle down to the lower income strata group for the years 2010–2014, as education and health reforms considerably reduce crime rate during the time period.

1 Introduction

The study evaluated different United Nation sustainable development goals (SDGs), i.e., goals 1 and 2 (poverty reduction and hunger), goals 3 and 4 (promotion of health and education), goal 10 (reduced inequalities), and goal 16 (reduction of violence, peace and justice) to access pro-poor growth and crime reduction in a panel of 16 heterogeneous countries. The discussion of crime rate in pro-poor growth (PPG) agenda remains absent in the economic development literature, though Bourguignon ( 2000 ) stressed to reduce crime and violence by judicious income distribution; however, a very limited literature is available to emphasize the need of social safety nets for vulnerable peoples that should be included in the pro-growth policy agenda for broad-based economic growth. Kelly ( 2000 ) investigated the relationship between income inequality (INC_INEQ) and urban crime, and found that INC_INEQ is the strong predictor to influence violent crime rather than property crime, while poverty (POV) and economic growth (EG) significantly affect on property crime rather than violent crime. The policies should be developed for equitable income and sound EG for reducing POV and crime across the globe. Drèze and Khera ( 2000 ) examined the inter-district variations of intentional homicides rate (IHR) in India for the period of 1981 and found that there is no significant relationship between urbanization/poverty and murder rates, while literacy rate has a strong impact to reduce criminal violence in India. The results further indicate the lower murder rate in those districts where female to male ratio is comparatively high. The study emphasized the need to reduce crime, violence and homicides by significant growth policies for sustained EG in India. Neumayer ( 2003 ) investigated the long-run relationship between political governance, economic policies and IHR using the panel of 117 selected countries for the period of 1980–1997 and concluded that IHR can be reduce by good economic and political policies. The results specified that higher income level, good civic sense, sound EG, and higher level of democracy all are connected with the lower homicides rate in a panel of countries. The study emphasized the need to improve governance indicators in order to lowering the IHR across the globe. Jacobs and Richardson ( 2008 ) examined the interrelationship between INC_INEQ and IHR in a panel of 14 developed democracies nation and found that intentional homicides is the mounting concerns in those nations where the inequitable income distribution exists, while results further provoke the presence of young males associated with the higher murder rates in a region. The policies should be formulated caution with care while devising for judicious income distribution with demographic variables in the pro-growth agenda. Sachsida et al. ( 2010 ) found inertial effect on criminality and confirmed the positive relationship between INC_INEQ, urbanization and IHR. The study emphasized the importance of public security spending to reduce IHR in Brazil. Pridemore ( 2011 ) re-assessed the relationship between POV, INC_INEQ and IHR in a cross-national panel of US states and found POV-homicides’ linkages rather than inequality-homicides’ association. The study argued that there is substantially desire to re-assess the inequality-homicides’ linkages as it might be the misspecification of the model. Ulriksen ( 2012 ) examined the relationship between PPG, POV reduction and social security policies in the context of Botswana and found that broad-based social security policies have a significant impact to reduce POV, thus there is a strong need to include social security protections in the pro-poor growth (PPG) agenda for lowering the POV rates across the globe. Ouimet ( 2012 ) investigated the impact of socio-economic factors on IHR in a panel of 165 countries for the period 2010 and found that GIP triangle are strongly connected with the IHR for all countries, while for sub-samples, the results only support the inequality-homicides association rather than POV and EG induced IHR. The results highlighted the importance of GIP triangle to reduce IHR in a panel of selected countries.

Liu et al. ( 2013 ) investigated the relationship between national scale indicators of socio-economic and demographic factors and crime rates in 32 Mexican states and found that EG, wages and unemployment negatively affect crime rates, while increase federal police force that is helpful to reduce crime rates; however, on the other way around, higher public security expenditures are linked with the higher crime rates in Mexican states. Chu and Tusalem ( 2013 ) investigated the role of state to reduce IHR in a panel of 183 nations and found that political instability increases IHR, while anocracies is the strong predictor to influence IHR in a panel of countries. The study concluded that IHR increases in those countries where there is high level of political instability and death penalty, while the amalgamation of democratic and autocratic features lead to increased IHR. The policies should be drawn to strengthen political governance across the globe. Adeleye ( 2014 ) evaluated the different determinants of INC_INEQ in a large panel of 137 countries using the time series data from 2000 to 2012 and found that per capita income (PCI), secondary education, rule of law index and unemployment rate are the strong predictors for INC_INEQ and IHR, while INC_INEQ considerably affected IHR rate in a region. Dalberis ( 2015 ) investigated the relationship between INC_INEQ, POV and crime rates in Latin American countries and found that INC_INEQ has no significant association with the crime rate in Colombia, Brazil, Uruguay and Salvador, while poverty is the strong predictor to influence crime in Brazil, Uruguay and Salvador. The results highlighted the need for pro-poorness of growth reforms that would be helpful to lowering the crime rates in Latin American countries. Harris and Vermaak ( 2015 ) considered the relationship between expenditures’ inequality and IHRe across 52 districts of South Africa and found that while keeping other district features constant, inequality does appear as a strong dominant player to induce IHR. The rational income distribution along with broad-based EG may play a vital role to reduce IHR in South Africa. Stamatel ( 2016 ) investigated the relationship between democratic cultural values and IHR in a panel of 33 democratic countries for the period 2010 and found that democratic cultural values have a positive and negative impact of IHR in the presence of strong democratic institutions and practices. Ahmed et al. ( 2016 ) identified the different predictors of economic and natural resources in the context of Iran using the time series data from 1965–2011 and found that labor productivity, exports, capital stock and natural resources are the main predictors of EG, which altogether are important for sustained long-term growth of the country. Enamorado et al. ( 2016 ) interlinked crime rates with higher INC_INEQ using a 20-year dataset of more than 2000 Mexican municipalities and confirmed the causal relationships between the two stated factors. The results confined that drug-related crime rates largely increase up to 36% if there is one-point increment in the INC_INEQ during the specified time period. The study concludes with the fact that drug-related violent crime rates are more severe due to high proliferation of large dispersion in the labor market in terms of negative job opportunities in illegal sector. Thus, the sound policies are imperative to seize drug trafficking organizations by force for pro-equality growth. Ling et al. ( 2017 ) analyzed the role of trade openness in Malaysian life expectancy using the data from 1960 to 2014. The results show that continued EG and trade openness substantially increase life expectancy during the study time period. Further, the results established the feedback relationship between income and life expectancy in a country. The study concludes that life expectancy may increase through imported healthcare goods, which improves the quality of life of the people, thus trade liberalization policies are imperative for healthy and wealthy wellbeing.

Zaman ( 2018 ) extensively surveyed the large weighted sample of intellectuals about crime–poverty nexus and explored the number of socio-economic factors that concerned with high crime rate and POV incidence in Pakistan, including INC_INEQ, injustice, unemployment, low spending on education and health, price hikes, etc. There is a high need to increase social spending on education and health infrastructure in order to combat POV and crime rates in a given country. Imran et al. ( 2018 ) considered a time series data of US for a period of 1965–2016 and concluded that incidence of POV increases the intensity of property crime in a given country, while other controlling factors including country’s PCI and unemployment rate are not significantly associated with property crime in a country. The study concludes that property crime should be restricted by strong legislative and regulatory measures, judicious income distribution, and increasing minimum wage rate, which altogether would be helpful for the poor to reap economic benefits from PPG reforms in a country. Zaman et al. ( 2019 ) evaluated the role of education in crime reduction in a panel of 21 countries for a period of 1990–2015 and found a parabola relationship between PCI and crime rates in the presence of quality education and equitable justice across countries. The study further confirmed few other causal conceptions among the variables for making sound policy implications in the context of criminal justice. Piatkowska ( 2020 ) examined the social cost of POV in terms of increasing suicides rates, crime rates, and total violent rates in the United States and across 15 European nations during the period of 1993–2000. The results show that suicides–crime–violent rates are substantially increasing due to increase in relative POV and infant mortality rates across countries. The study argued that relative POV is the strong predictor to increase social cost of nation that needs efficient economic policies to reduce crime rates. Mukherjee ( 2019 ) discussed the role of social sustainability in achieving economic sustainability by reducing different forms of violent/crime rates through state intervention in the context of Indian economy by utilizing the data for a period of 2005–2016. The results further highlighted the need of socio-economic infrastructure development that would be helpful to provide safety nets to the poor in order to reduce crime rates in a country. Duque and McKnight ( 2019 ) presented the channel through which crime rates and legal system provide a pathway to increase INC_INEQ and POV across countries. The study further discussed and highlighted the socio-economic vulnerability that escalates through unequal distribution of income and high POV incidence, which need effective legal system to reduce crime rates. Khan et al. ( 2019a ) surveyed the Bolivian economy to assess pro-poor environmental reforms that could improve the quality of life of the poor through judicious income distribution and sustainable environmental reforms. The results conclude that services’ sector and healthcare infrastructure would be helpful to reduce POV rate and achieve PPG process at country wide. Zaman et al. ( 2020 ) surveyed the large panel of countries (i.e., 124 countries) for a period of 2010–2013 to analyze the role of INC_INEQ and EG on POV incidence across countries. The results generally favor the strong linkages among the three stated factors to support GIP triangle, which forms PPG process. The study emphasized the need to adopt some re-corrective measures in order to provide social safety nets and income distribution in order to make a growth process more pro-poor. Kousar et al. ( 2019 ) confined its finding in favor of POV reduction through managing international remittances’ receipts and financial development that would be helpful to improve the mechanism of income distribution in a country like Pakistan. The study concluded that international remittances may play a vital role to reduce POV via the mediation of financial development in a country.

The real problem is how to make EG more equitable, which is helpful to reduce POV and crime rates, and make a growth more pro-poor. The SDGs largely provoked the need to sustained economic activities, which helpful to make growth policies more poor friendly. The previous studies are widely discussed crime rates and POV reduction (see Zaman 2018 ; Khan et al. 2015 ; Heinemann and Verner 2006 ; etc.); however, a very few studies interlinked POV–crime nexus under PPG and Kuznets curve (KC) hypothesis (see Saasa 2018 ; Berens and Gelepithis 2018 , etc.). Based on the interconnections between crime, POV, and PPG, the study formulated the following research questions, i.e.,

Does crime rate negatively influenced GIP triangle, which sabotages the process of PPG?

The recent study of Khan et al. ( 2019b ) provoked the need of PPG policies to ensure sustainability agenda by including socio-economic and environmental factors in policy formulation, which gives favor to the poor as compared to the non-poor. In the similar lines, the social spending on education and healthcare infrastructure, and reforms needed to reduce labor market uncertainty in the form of lessen unemployment rate is considered the viable option for crime and POV reduction across countries (Khan et al. 2017 ). Thus, the study evaluated the question, i.e.,

To what extent social spending on education, health, and labor market are helpful to reduce crime rate, poverty, and income inequality across countries?

This question would be equally benefited to the developmental economists and policy makers to devise a healthy and wealthy policy by increasing spending on social infrastructure for pro-equality growth (Wang 2017 ). The last question is based upon non-linear formulation of crime–POV nexus where it is evaluated as a second-order coefficient to check the parabola relationship between them, i.e.,

Does crime and poverty exhibit a parabola relationship between them?

The question is all about the second-order condition, which confirmed one out of three conditions, i.e., either it is accepted an inverted U-shaped or U-shaped or flat relationship between them. The second-order condition assessed the probability to reduce crime rates and incidence of POV in policy formulation.

In the light of SDGs, the study explored the impact of GIP triangle and crime rates on pro-growth and PPG policies, which is imperative for sustainable development across countries. The study added social expenditures in PPG dynamics to promote healthy and wealthy economic activities, which improves quality of life of the poor and helpful to reduce crime incidence across countries. The study is first in nature, as authors’ knowledge, which included GIP triangle and crime rate in PPG framework, while controlling different socio-economic factors, including education and health expenditures, unemployment rate, and trade openness. Further, an empirical contribution of the study is to include second-order coefficient of PCI for evaluating crime- and inequality-induced KC, while the study proceed to analyze forecast relationship between the crime and POV incidence over a next 10-year time period. Finally, the study estimated PPG index while including crime rate as a main predictor factor in GIP triangle for robust policy inferences. Thus, these objectives are achieved by different statistical techniques for robust analysis.

2 Data source and methodological framework

The study used number of promising socio-economic variables to determine the dynamic relationship between PPG factors and crime rate under the framework of an inverted U-shaped KC in a panel of 16 diversified countries, using system GMM estimator for the period of 1990–2014. The study used the following variables, i.e., crime rate (proxy by intentional homicides rate per 100,000 population), GINI index measures income inequality, poverty headcount ratio at $1.90 a day (2011 PPP) (% of total population), national estimates of unemployment in % of total labor force, education expenditures as % of GDP, per capita health expenditure in current US$, per capita income in constant 2005 US$, and trade openness as % of GDP. The samples of countries are presented in Table  7 in Appendix for ready reference. The data for the study are obtained from World Development Indicators published by World Bank ( 2015 ).

These countries are selected because of the devastating crime rate during the study time period. The recorded figures for Argentina crime rates about to 245% increase between the period of 1991 and 2007, while 2002 is considered the highest committed crime data recorded when the POV and INC_INEQ reached at their peak levels (Bouzat 2010 ). Brazil economy is working out for reduction of crime by focusing on three-point agenda, i.e., reduction in income disparity, to increase spending on education via an increase in enrollment of school dropout children, and to improve labor market conditionings. These three policies design to deter the crime rates in a given country (World Bank 2013 ). The robbery complaints largely increase since last two decades in Chile, which is being planned by controlling two action strategies, i.e., plan cuadrante and country security plan. Both the plan designed to restructured police force to reduce robbery and violence in a country (Vergara 2012 ). The rural China is suffered by high INC_INEQ that leads to higher crime rate (South China Monitoring Report 2015 ) while POV and INC_INEQ lead to crime and violent factor in Colombia (Gordon 2016 ). The socio-economic factors including low provision of education, health, high POV, and food challenges lead to increase crime in Indonesia (Pane 2017 ), while generating employment opportunities and increasing wage rate in Malaysia may be beneficial to reduce crime–POV nexus in a given country (Mulok et al. 2017 ). Mexican economy is suffered with high rate of homicides that negatively affect labor market outcomes, while country inhibits by increasing strict laws to diminish violence (Kato Vidal 2015 ). The safety situation in Morocco is cumbersome, as one of the country reports shows that an increased rate in crime is about to increase up to 23% in 2016 (OSAC 2017 ). The number of other factors remains visible in selected sample of panel of countries, including rural POV and social exclusion that is considered the main factor of socio-economic crisis in Poland (European Commission 2008 ); POV, unemployment, and INC_INEQ chiefly attributed to crime rate in South Africa (Bhorat et al. 2017 ); politics, democracy, and INC_INEQ arise conflicts in Thailand (Hewison 2014 ); corruption and high unemployment are the major conflicts in Tunisia (Saleh 2011 ); and Uruguay economy needs policy actions to reduce POV by investment in children education, modernizing rural sector, and balancing the gender gap (Thamma 2017 ). Thus, these facts about crime and POV in different countries put a focus to study crime–POV nexus under PPG framework in this study for robust evaluation. Figure  2 in Appendix shows the plots of the studied variables at level.

The study used the following non-linear equations to determine the dynamic relationship between PPG factors and crime rate in a panel of countries, i.e.,

where GDPPC indicates per capita GDP, GDPPC 2 indicates square of per capita GDP, GINI indicates Gini coefficient—income inequality, EDUEXP indicates education expenditures, HEXP indicates health expenditures, POVHCR indicates poverty headcount ratio, TOP indicates trade openness, UNEMP indicates unemployment, and CRIME indicates crime rate.

Equations ( 1 ) to ( 3 ) assessed the possible inverted U-shaped relationships between crime rate and PCI, between POVHCR and PCI, and between GINI and PCI, while Eq. ( 4 ) reviewed the PPG reforms across countries. Arellano and Bond ( 1991 ) developed the differenced GMM estimator, whom argued that the GMM estimator eliminates country effects and controls the possible endogeneity of explanatory variables using the appropriate instrumental list that evaluated by Sargan–Hansen test. The process further involves two-step GMM iterations with the time updated weights and adopted the weighting matrix by White period. The tests for autocorrelations by AR(1) and AR(2) and the Sargan test by Sargan–Hansen of over-identifying restrictions are presented for statistical reliability of the given models. The differenced GMM is superior to the 2SLS and system GMM, i.e., 2SLS regression estimator is used when the known endogeneity exists between the variables, which are handled by including the list of instrumental variables at their first lagged. Thus, the possible endogeneity problem is resolved accordingly. The system GMM further be used instead of 2SLS as if there are more than one endogenous issues exist in the model, which is unable to resolve through 2SLS estimator. Finally, the differenced GMM estimator is used as its estimated AR(1) and AR(2) bound values that would be helpful to encounter the issues of serial correlation and endogeneity problem accordingly.

Using the GMM estimator, the study verified different possibilities of KC, i.e., if the signs and magnitudes of \(\beta_{1} > 0\) and \(\beta_{2} < 0\) , than we may confirm the crime-induced KC, poverty-induced KC, and inequality-induced KC. The inverted U-shaped relationship between crime rate and PCI verified ‘crime-induced KC’, between POVHCR and PCI verified ‘POV-induced KC’, and inverted U-shaped relationship between GINI and PCI verified ‘inequality-induced KC’. On the other way around, if \(\beta_{1} < 0\) and \(\beta_{2} > 0\) , then we consider the U-shaped KC between crime rate and PCI, between POV and PCI, and between GINI and PCI, respectively. There are three other situations we may observe with the sign and magnitude of \(\beta_{1}\) and \(\beta_{2}\) , i.e., (i) \(\beta_{1} < 0\) and \(\beta_{2} = 0\) , (ii) \(\beta_{1} > 0\) and \(\beta_{2} = 0\) , and (iii) \(\beta_{1} = 0\) and \(\beta_{2} = 0\) , referred the monotonically decreasing function, monotonically increasing function, and flat/no relationship with the crime-PCI, poverty-PCI, and inequality-PCI in a panel of cross-sectional countries. The study further employed social accounting matrix by impulse response function (IRF) and variance decomposition analysis (VDA) in an inter-temporal relationship between the studied variables for a next 10-year period starting from 2015 to 2024. As it name implies, VDA explains the proportional variance in one variable caused by the proportional variance by the other variables in a vector autoregressive (VAR) system, while IRF traces the dynamic responses of a variable to innovations in other variables in the system. Both the techniques use the moving average representation of the original VAR system. Figure  1 shows the theoretical framework of the study to clearly outline the possible relationship between the stated variables.

Source: authors’ extraction

Research framework of the study.

Figure  1 shows the possible relationship between POV and crime rates in mediation of inequality, unemployment, and EG across countries. It is likelihood that POV increases inequality that leads to decrease in EG. The low-income growth further leads to increased unemployment, which causes high crime rates. This nexus is still rotated through crime rates that increase POV incidence across countries. The PPG process still works under the stated factors that need judicious income distribution to reduce crime rates.

The study further proceeds to evaluate the PPG reforms in a panel of selected countries. Kakwani and Pernia ( 2000 ) proposed an index of PPG called ‘PPG index’, which is evaluated by the growth elasticity and inequality elasticity with respect to POV. The same methodology is adopted in this study to assess the PPG and/or pro-rich growth reforms to assess the changes in the crime rate in a panel of countries. PPG defined as a state in which where the growth trickles down to the poor as compared to the non-poor. Poverty is largely affected by two main factors, i.e., higher growth rate may reduce the POV rates, while higher INC_INEQ reduces the impact of EG to reduce POV; therefore, the PPG index included the following mathematical illustrations, i.e.,

The study further assessed the pro-poorness of social expenditures and evaluates its impact to observe changes in IHR. The study shows the following mathematical illustrations that is extended from the scholarly work of Zaman and Khilji ( 2014 ); Kakwani and Pernia ( 2000 ) and Kakwani and Son ( 2004 ) i.e.,

where \(\alpha =\) 0, 1 and 2 indicate POVHCR, poverty gap and squared poverty gap, respectively, ‘P’ indicates FGT poverty measures, and ‘SOCIALEXP’ indicates social expenditures. Differentiating \(\eta_{\alpha }\) in Eq. ( 9 ) with respect to social expenditures gives more elaborated form of GEP, i.e.,

The elasticity of entire class of poverty measures \(P_{\alpha }\) with respect to Gini index is given by

which will be always positive only when \(S{\text{OCIALEXPE}} > z\) .Equations ( 10 ) and ( 11 ) are combined together to form TPE for all FGT poverty measures, i.e.,

or \(\delta_{\alpha } = \eta_{\alpha } + \xi_{\alpha }\) . Finally, pro-poorness of social expenditures estimated based on the following equation, i.e.,

Kakwani and Son ( 2004 ) presented the following bench mark applications to assess the pro-poor and/or anti-poor policies, i.e., the following value judgments regarding the PPG index ( \(\varphi\) ) are as follows, i.e.,

\(\varphi\)  < 0, growth is pro-rich or anti-poor,

0 <  \(\varphi\) \(\le\) 0.33, the process of PPG is considerable low,

0.33 <  \(\theta\) \(\le\) 0.66, the process of PPG is moderate,

0.66 <  \(\varphi\)  < 1.0, the process of EG considered as pro-poor, and

\(\varphi \ge\) 1.0, the process of EG is highly pro-poor.

The study utilized the PPG model for ready reference in this study.

This section presented the descriptive statistics in Table  1 , correlation matrix in Table  2 , dynamic system GMM estimates in Table  3 , IRF estimates in Table  4 , VDA estimates in Table  5 , while finally Table  6 shows the estimates for PPG in a panel of selected countries. Table  1 shows that GDPPC has a minimum value of US$ 199.350 and the maximum value of US$ 11257.600, with a mean and standard deviation (STD) value of US$ 4340.777 and US$ 2490.554, respectively. GINI has a minimum value of 25% and the maximum value of 64.790%, having an STD value of 8.580% with an average value of 45.095%. The minimum value of EDUEXP is about 0.998% of GDP and the maximum value of 7.657% of GDP, with an average value of 4.051% of GDP. The average value of HEXP per capita is about US$ 321.249 and a maximum value of US$ 1431.154, with an STD value of US$ 292.802. The maximum value of POVHCR is about 69% at US$1.90 a day with an average value of 12.394% at US$1.90 a day. The minimum value of trade is 13.753% of GDP and the maximum value of 220.407% of GDP, with an average value of 62.391% of GDP. The mean value for UNEMP is about 8.890% of total labor force with STD value of 6.010%. Finally, the minimum value of crime rate is about 0.439 per 100,000 inhabitants and the maximum value of 71.786 per 100,000 inhabitants, with an average value of 11.664 per 100,000 peoples. This exercise would be helpful to understand the basic descriptions of the studied variables in a panel of countries.

Figure  3 in Appendix shows the plots of the studied variables and found the stationary movement in the variables at their first difference. Table  2 presents the estimates of correlation matrix and found that GINI (i.e., r  = 0.264), EDUEXP ( r  = 0.243), HEXP ( r  = 0.730), TOP ( r  = 0.061), UNEMP (0.152) and CRIME ( r  = 0.031) have a positive correlation with the GDPPC, while POVHCR ( r  = − 0.599) significantly decreases GDPPC.

The results further reveal that GINI is affected by EDUEXP, HEXP, UNEMP and CRIME, while it considerably decreases by trade liberalization policies. EDUEXP, HEXP, PCI, TOP and UNEMP significantly decrease POVHCR, while crime rate has a positive correlation with the POVHCR. Finally, GINI have a greater magnitude, i.e., r  = 0.671, to influence CRIME, followed by UNEMP ( r  = 0.417), EDUEXP ( r  = 0.188), and POVHCR ( r  = 0.164) while trade liberalization policies support to decrease crime rates in a panel of countries. The study now proceeds to estimate the two-step system GMM for analyzing the functional relationship between socio-economic factors and crime rate. The results are presented in Table  3 .

The results of panel GMM show that GINI and UNEMP both have a significant and direct relationship with the CRIME, while TOP have an indirect relationship with CRIME in a panel of countries. The results imply that GINI and UNEMP are the main factors that increase CRIME, while trade liberalization policies have a supportive role to decrease crime rates across countries. Thorbecke and Charumilind ( 2002 ) evaluated the impact of income inequality on health, education, political conflict, and crime, and surveyed the different casual mechanism in between income inequality and its socio-economic impact across the globe. The policies have devised while reaching the conclusive relationships between them. Kennedy et al. ( 1998 ) concluded that social capital and income inequality are the powerful predictors of intentional homicides rate and violent crime in the US states. Altindag ( 2012 ) explored the long-run relationship between unemployment and crime rates in a country-specific panel dataset of Europe and found that unemployment significantly increases crime rates, while unemployment has a power predictor of exchange rate movements and industrial accident across the Europe. Menezes et al. ( 2013 ) confirmed the positive association between income inequality and criminality, as rational income distribution tends to decrease neighborhood homicides rate while it implies an increase in the intentional homicides rate in the surrounding neighborhoods.

In a second regression panel, the results confirmed the U-shaped relationship between POVHCR and GDPPC, as at initial level of EG, POV significantly declines, while at the later stages, this result is evaporated, as EG subsequently increases POVHCR that shows pro-rich federal policies across countries. The HEXP, however, significantly decreases POVHCR during the study time period. Dercon et al. ( 2012 ) investigated the relationship between chronic POV and rural EG in Ethiopia and argued that chronic POV is associated with the lack of education, physical assets and remoteness, while EG in terms of provide better roads and extension services may trickle down to the poor in a same way that the non-chronically poor benefited. Solinger and Hu ( 2012 ) examined the relationship between health, wealth and POV in urban China and found that wealthier cities prefer to allocate their considerable portion of savings for social assistance funds, while poorer places save the city money and work outside in a hope that the peoples would be better able to support themselves. Fosu ( 2015 ) examined the relationship between GIP triangle in sub-Saharan African countries and found that as a whole, South African countries lag behind the BICR (Brazil, India, China and Russia) group of countries; however, many of them in sub-Saharan African countries have outperformed India. The results further specified that PCI is the main predictor to reduce POV in sub-Saharan African countries; however, rational income distribution is a crucial challenge to reduce POV reduction through substantial growth reforms in a region. Kalichman et al. ( 2015 ) concluded that food poverty is associated with the multifaceted problems of health-related outcomes across the globe.

In a third regression panel, the results confirm an inverted U-shaped relationship between GDPPC and GINI that verified an inequality-induced KC in a panel of countries. The results imply that at initial level of economic development, GINI first increases and then decreases with the increased GDPPC across countries. CRIME, however, it is associated with the higher GINI during the studied time period. Kuznets ( 1955 ), Ahluwalia ( 1976 ), Deininger and Squire ( 1998 ), and others confirmed an inverted U-shaped relationship between INC_INEQ and PCI in different economic settings. Mo ( 2000 ) suggested different channelss to examine the possible impact of INC_INEQ on EG and found that ‘transfer channel’ exert the most important channel, while ‘human capital’ is the least important channel that negatively affects the rate of EG via INC_INEQ. Popa ( 2012 ) argued that health and education both are important predictors for EG, while POV and unemployment negatively correlated with the EG in Romania. Herzer and Vollmer ( 2012 ) confirmed the negative relationship between INC_INEQ and EG within the sample of developing countries, developed countries, democracies, non-democracies, and sample as a whole. In a similar line, Malinen ( 2012 ) confirmed the long-run equilibrium relationship between PCI and INC_INEQ and found that income inequality negatively affected the growth of developed countries.

The final regression shows that HEXP and TOP both significantly increase GDPPC, while POVHCR decreases the pace of EG, which merely be shown pro-rich federal policies in a panel of countries. Ranis et al. ( 2000 ) found that both the health and education expenditures lead to increased EG, while investment improves human development in a cross-country regression. Bloom et al. ( 2004 ) confirmed the positive connection between health and EG across the globe. Gyimah-Brempong and Wilson ( 2004 ) examined the possible effect of healthy human capital on PCI of sub-Saharan African and OECD countries and found the positive association between them in a panel of countries.

The statistical tests of the system GMM estimator confirmed the stability of the model by F-statistics, as empirically model is stable at 1% level of confidence interval. Sargan–Hansen test confirmed the instrumental validity at conventional levels for all cases estimated. Autocorrelations tests imply that except POVHCR model, the remaining three models including CRIME, GINI and GDPPC model confirmed the absence of first- and second-order serial correlation, and as a consequence, we verified our instruments are valid. As far as POVHCR model, we believed the results of Sargan–Hansen test of over identifying restrictions and AR(1) that is insignificant at 5% level, and confirmed the validity of instruments and absence of autocorrelation at first-order serial correlation. Table  4 shows the estimate of IRF for the next 10-year period starting from a year of 2015 to 2024.

The results show that the socio-economic factors have a mix result with the rate of crime, as POVHCR slightly increases with decreasing rate with the crime data, i.e., in the next coming years from 2016, 2018, 2019, and 2022, POVHCR exhibits a negative sign, while in the remaining years in between from 2015 to 2024, POVHCR increases crime rate. GINI will considerably increase crime rate from 2022 to 2024. UNEMP has a mixed result to either increase crime rate in one period while in the very next upcoming periods, it declines crime rate. Similar types of results been found with EDUEXP, HEXP and with the TOP; however, GDPPC will constantly increase the rate of crime in a panel of countries. In an inter-temporal relationship between POVHCR and other predictors, the results show that GDPPC would significantly decrease POVHCR for the next 10-year period; however, UNEMP, HEXP, and crime rate would considerably increase POVHCR. EDUEXP and TOP would support to reduce GINI for the next upcoming years, while remaining variables including crime rate, POV, UNEMP, HEXP, and GDPPC associated with an increased GINI across countries. The GDPPC will be influenced by crime rate, POVHCR, GINI, UNEMP, HEXP, and EDUEXP, while TOP would considerably to support GDPPC for the next 10-year time period. Figure  4 in Appendix shows the IRF estimates for the ready reference.

Table  5 shows the estimates of VDA and found that POVHCR will exert the largest share to influence crime rates, followed by GDPPC, TOP, HEXP, EDUEXP, GINI, and UNEMP. POVHCR would be affected by crime rate (i.e., 4.450%), UNEMP (1.751%), GDPPC (1.120%), GINI (1.043%), HEXP (0.639%), and EDUEXP (0.512%), and TOP (0.299%), respectively.

The results further reveal that GINI will affected by POVHCR, as it is explained by 7.680% variations to influence GINI for the next 10-year period. UNEMP, EDUEXP, and crime rate will subsequently influenced GDPPC about to 1.107%, 0.965%, and 0.312% respectively. The largest variance to explain UNEMP will be TOP, while the lowest variance to influence UNEMP will be GINI for the next 10-year period. Finally, GDPPC would largely influenced by HEXP, followed by UNEMP, CRIME, POVHCR, EDUEXP, TOP, and GINI for the period of 2015 to 2024. Figure  5 in Appendix shows the plots of the VDA for ready reference.

Finally, Table  6 presents the changes in crime rate by five different growth phases, i.e., phase 1: 1990–1994, phase 2: 1995–1999, phase 3: 2000–2004, phase 4: 2005–2009, and phase 5: 2010–2014. The results show that in the years 1990–1994, 1% increase in EG and INC_INEQ decrease POVHCR by − 0.023% and − 0.630%, which reduces TPE by − 0.629 percentage points. The PPG index surpassed the bench mark value of unity and confirmed the trickledown effect that facilitates the poor as compared to the non-poor. However, there is an overwhelming increase in the crime rate beside that the pro-poorness of EG, which indicate the need for substantial safety nets’ protection to the poor that escape out from this acute activities (Wang et al. 2017 ). In a second phase from 1995 to 1999, although EG decreases POVHCR by − 0.187; however, GINI has a greater share to increase POVHCR by 0.517% that ultimately increases TPE by 0.330%. This increase in the TPE turns to decrease PPG as 1.764, which shows anti-poor/pro-rich federal policies and low reforms for the poor that accompanied with the higher rates of crime in a panel of countries. The rest of the growth phases from 2000 to 2014 show anti-poor growth accompanied with the higher INC_INEQ and lower EG; however, crime rate decreases in the year 2000–2004 and 2010–2014 besides that the growth process is anti-poor across countries. The policies should be formulated in a way to aligned crime rate with the PPG reforms across countries (Vellala et al. 2018 ).

The results of PPE index confirmed an anti-poor growth from 1990 to 2004, while at the subsequent years from 2005 to 2014, education growth rate subsequently benefited the poor as compared to the non-poor, i.e., PPE index exceeds the bench mark value of unity. Crime rate is increasing from 1990 to 1999, and from 2005 to 2009, while it decreases the crime rate for the years 2000–2004 and 2010–2014. The good sign of recovery has been visible for the years 2010–2014 where the PPE growth supports to decrease crime rate in a panel of selected countries. Finally, the PPH index confirmed two PPG phases, i.e., from 1990 to 1994, and 2010 to 2014 in which crime rate increases for the former years and decreases in the later years. The remaining health phases from 1995 to 2009 show anti-poor health index, while crime rate is still increasing during the years from 1995 to 1999 and 2005 to 2009, and decreasing for the period 2000–2004. The results emphasized the need to integrate PPG index with the crime rate, as PPG reforms are helpful to reduce humans’ costs by increasing EG and social expenditures, and providing judicious income distribution to escape out from POV and vulnerability across the globe (Musavengane et al. 2019 ).

From the overall results, we come to the conclusion that social spending on education and health is imperative to reduce crime incidence, while it further translated a positive impact on POV and inequality reduction across countries (Hinton 2016 ). EG is a vital factor to reduce POV; however, it is not a sufficient condition under higher INC_INEQ (Dudzevičiūtė and Prakapienė 2018 ). INC_INEQ and unemployment rate both are negatively correlated with crime rates; however, it may be reduced by judicious income distribution and increases social spending across countries (Costantini et al. 2018 ). Trade liberalization policies reduce incidence of crime rates and improve country’s PCI, which enforce the need to capitalize domestic exports by expanding local industries. Thus, the United Nations SDGs would be achieved by its implication in the countries perspectives (Dix-Carneiro et al. 2018 ). The study achieved the research objectives by its theoretical and empirical contribution, which seems challenge for the developmental experts to devise policies toward more pro-growth and PPG.

4 Conclusions and policy recommendations

This study investigated the dynamic relationship between socio-economic factors and crime rate to assess PPG reforms for reducing crime rate in a panel of 16 diversified countries, using a time series data from 1990–2014. The study used PCI and square PCI in relation with crime rate, POVHCR, and GINI to evaluate crime-induced KC, poverty-induced KC and inequality-induced KC, while PPG index assesses the federal growth reforms regarding healthcare provision, education and wealth to escape out from POV and violence. The results show that GINI and UNEMP are the main predictors that have a devastating impact to increase crime rate. Trade liberalization policies are helpful to reduce crime rate and increase PCI. Healthcare expenditures decrease POVHCR and amplify EG. The EG is affected by POVHCR, which requires strong policy framework to devise PPG approach in a panel of selected countries. The study failed to establish crime-induced KC and poverty-induced KC, while the study confirmed an inequality-induced KC. The results of IRF reveal that PCI would considerably increase crime rate, while crime rate influenced GINI and PCI for the next 10-year period. The estimates of VDA show that POVHCR explained the greater share to influence crime rates, while reverse is true in case of POVHCR. The study divided the studied time period into five growth phases 1990–1994, 1995–1999, 2000–2004, 2005–2009, and 2010–2014 to assess PPG, PPH, and PPE reforms and observe the changes in crime rates. The results show that there is an only period from 1990 to 1994 that shows PPG, while crime rate is still increasing in that period; however, in the years 2000–2004, and 2010–2014, crime rate decreases without favoring the growth to the poor. PPE and PPH assessment confirmed the reduction in the crime rates for the years 2010–2014. The overall results confirmed the strong correlation between socio-economic factors and crime rates to purse the pro-poorness of government policies across countries. The overall results emphasized the need of strong policy framework to aligned PPG policies with the reduction in crime rate across the globe. The study proposed the following policy recommendations, i.e.,

Education, health and wealth are the strong predictors of reducing crime rates and achieving PPG, thus it should be aligned with inclusive trade policies to reduce human cost in terms of decreasing chronic poverty and violence/crime.

The policies should be formulated to strengthen the pro-poorness of social expenditures that would be helpful to reduce an overwhelming impact of crime rate in a panel of countries.

GIP triangle is mostly viewed as a pro-poor package to reduce the vicious cycle of poverty; however, there is a strong need to include some other social factors including unemployment, violence, crime, etc., which is mostly charged due to increase in poverty and unequal distribution of income across the globe. The policies should devise to observe the positive change in lessen the crime rate by PPG reforms in a panel of selected countries.

The significant implication of the Kuznets’ work should be extended to the some other unexplored factors especially for crime rate that would be traced out by the pro-poor agenda and pro-growth reforms.

There is a need to align the positivity of judicious income distribution with the broad-based economic growth that would be helpful to reduce poverty and crime rate across countries.

The result although not supported the ‘parabola’ relationship between income and crime rates; however, it confirmed the U-shaped relationship between income and poverty. The economic implication is that income is not the sole contributor to increase crime rates while poverty exacerbates violent crimes across countries. There is a high need to develop a mechanism through which poverty incidence can be reduced, which would ultimately lead to decreased crime rates. The improvement in the labor market structure, judicious income distribution, and providing social safety nets are the desirable strategies to reduce crime rates and poverty incidence across countries, and

The results supported parabola relationship between economic growth and inequality, which gives a clear indication to improve income distribution channel for reducing poverty and crime rates at global scale.

These seven policies would give strong alignment to improve social infrastructure for managing crime through equitable justice and PPG process.

Availability of data and materials

The data are freely available on World Development Indicator, published by World Bank on given URL ID: https://datacatalog.worldbank.org/dataset/world-development-indicators .

Adeleye NB (2014). The determinants of income inequality and the relationship to crime. Unpublished dissertation, University of Sussex, UK. https://www.researchgate.net/profile/Ngozi_Adeleye2/publication/276410308_The_Determinants_of_Income_Inequality_and_the_Relationship_to_Crime/links/5558c9f808aeaaff3bf98a45.pdf . Accessed 6 Jan 2016

Ahluwalia MS (1976) Income distribution and development: some stylized facts. Am Econ Rev 66(2):128–135

Google Scholar  

Ahmed K, Mahalik MK, Shahbaz M (2016) Dynamics between economic growth, labor, capital and natural resource abundance in Iran: an application of the combined cointegration approach. Res Policy 49:213–221

Article   Google Scholar  

Altindag DT (2012) Crime and unemployment: evidence from Europe. Int Rev Law Econ 32(1):145–157

Arellano M, Bond SR (1991) Some tests of specification of panel data: monte Carlo evidence and an application to employment equations. Rev Econ Stud 58:277–297

Berens S, Gelepithis M (2018) Welfare state structure, inequality, and public attitudes towards progressive taxation. Socio Econ Rev. https://doi.org/10.1093/ser/mwx063

Bhorat H, Thornton A, Van der Zee K (2017). Socio-economic determinants of crime in South Africa: an empirical assessment. DPRU Working Paper 201704. DPRU, University of Cape Town, Rondebosch

Bloom DE, Canning D, Sevilla J (2004) The effect of health on economic growth: a production function approach. World Dev 32(1):1–13

Bourguignon F (2000). Crime, violence and inequitable development. Annual World Bank Conference on development economics 1999, pp. 199–220

Bouzat G (2010) Inequality, crime, and security in Argentina. SELA (Seminario en Latinoamérica de Teoría Constitucional y Política) Papers. Paper 91. http://digitalcommons.law.yale.edu/yls_sela/91 . Accessed 6 October 2018

Chu DC, Tusalem RF (2013) The role of the state on cross-national homicide rates. Int Crim Justice Rev 23(3):252–279

Costantini M, Meco I, Paradiso A (2018) Do inequality, unemployment and deterrence affect crime over the long run? Reg Stud 52(4):558–571

Dalberis R (2015). Extreme levels of poverty and inequality may lead to equally high levels of social conflict and crime. Unpublished dissertation, CUNY Academic Works, New York. http://academicworks.cuny.edu/cc_etds_theses/346/ . Accessed 6 Jan 2016

Deininger K, Squire L (1998) New ways of looking at old issues: inequality and growth. J Dev Econ 57(2):259–287

Dercon S, Hoddinott J, Woldehanna T (2012) Growth and chronic poverty: evidence from rural communities in Ethiopia. J Dev Stud 48(2):238–253

Dix-Carneiro R, Soares RR, Ulyssea G (2018) Economic shocks and crime: evidence from the Brazilian trade liberalization. Am Econ J Appl Econ 10(4):158–195

Dreze J, Khera R (2000) Crime, gender, and society in India: insights from homicide data. Popul Dev Rev 26(2):335–352

Dudzevičiūtė G, Prakapienė D (2018) Investigation of the economic growth, poverty and inequality inter-linkages in the European Union countries. J Secur Sust Issues 7:839–854

Duque M, McKnight A (2019) Understanding the relationship between inequalities and poverty: mechanisms associated with crime, the legal system and punitive sanctions. LIP Paper, 6, CASE/215, Centre for Analysis of Social Exclusion, London School of Economics, London, UK. http://sticerd.lse.ac.uk/dps/case/cp/casepaper215.pdf . Accessed 8 March 2020

Enamorado T, López-Calva LF, Rodríguez-Castelán C, Winkler H (2016) Income inequality and violent crime: evidence from Mexico’s drug war. J Dev Econ 120:128–143

European Commission (2008 ). Poverty and social exclusion in rural areas—final report Annex I—Country Studies-Poland. European commission, Brussels

Fosu AK (2015) Growth, inequality and poverty in Sub-Saharan Africa: recent progress in a global context. Oxford Dev Stud 43(1):44–59

Gordon E (2016) Poverty, crime, and conflicts: socio-economic inequalities and prospects for peace in Colombia. Centre for security governance. http://secgovcentre.org/2016/10/poverty-crime-and-conflict-socio-economic-inequalities-and-the-prospects-for-peace-in-colombia/ . Accessed 6 October 2018

Gyimah-Brempong K, Wilson M (2004) Health human capital and economic growth in Sub-Saharan African and OECD countries. Q Rev Econ Finance 44(2):296–320

Harris G, Vermaak C (2015) Economic inequality as a source of interpersonal violence: evidence from Sub-Saharan Africa and South Africa. South Afr J Econ Manag Sci 18(1):45–57

Heinemann, Verner D (2006) Crime and violence in development: a literature review of Latin America and the Caribbean. The World Bank, Policy Research Working Papers, Washington D.C

Herzer D, Vollmer S (2012) Inequality and growth: evidence from panel cointegration. J Econ Inequality 10(4):489–503

Hewison K (2014) Considerations on inequality and politics in Thailand. Democratization 21(5):846–866

Hinton E (2016) From the war on poverty to the war on crime: the making of mass incarceration in America. Harvard University Press, Cambridge

Book   Google Scholar  

Imran M, Hosen M, Chowdhury MAF (2018) Does poverty lead to crime? Evidence from the United States of America. Int J Soc Econ 45(10):1424–1438

Jacobs D, Richardson AM (2008) Economic inequality and homicide in the developed nations from 1975 to 1995. Homicide Stud 12(1):28–45

Kakwani N, Pernia EM (2000) What is pro-poor growth? Asian Dev Rev 18(1):1–16

Kakwani N, Son HH (2004) Pro-poor growth: concepts and measurement with country case studies. Pakistan Dev Rev 42(4 Part I):417–444

Kalichman SC, Hernandez D, Kegler C, Cherry C, Kalichman MO, Grebler T (2015) Dimensions of poverty and health outcomes among people living with HIV infection: limited resources and competing needs. J Community Health 40(4):702–708

Kato Vidal EL (2015) Violence in Mexico: an economic rationale of crime and its impacts. EconoQuantum 12(2):93–108

Kelly M (2000) Inequality and crime. Rev Econ Stat 82(4):530–539

Kennedy BP, Kawachi I, Prothrow-Stith D, Lochner K, Gupta V (1998) Social capital, income inequality, and firearm violent crime. Soc Sci Med 47(1):7–17

Khan N, Ahmed J, Nawaz M, Zaman K (2015) The socio-economic determinants of crime in Pakistan: new evidence on an old debate. Arab Econ Business J 10(2):73–81

Khan HUR, Khan A, Zaman K, Nabi AA, Hishan SS, Islam T (2017) Gender discrimination in education, health, and labour market: a voice for equality. Qual Quant 51(5):2245–2266

Khan HUR, Zaman K, Yousaf SU, Shoukry AM, Gani S, Sharkawy MA (2019a) Socio-economic and environmental factors influenced pro-poor growth process: new development triangle. Environ Sci Pollut Res 26(28):29157–29172

Khan HUR, Nassani AA, Aldakhil AM, Abro MMQ, Islam T, Zaman K (2019b) Pro-poor growth and sustainable development framework: evidence from two step GMM estimator. J Cleaner Prod 206:767–784

Kousar R, Rais SI, Mansoor A, Zaman K, Shah STH, Ejaz S (2019) The impact of foreign remittances and financial development on poverty and income inequality in Pakistan: evidence from ARDL-bounds testing approach. J Asian Finance Econ Business 6(1):71–81

Kuznets S (1955) Economic growth and income inequality. Am Econ Rev 45(1):1–28

Ling CH, Ahmed K, Muhamad R, Shahbaz M, Loganathan N (2017) Testing the social cost of rapid economic development in Malaysia: the effect of trade on life expectancy. Soc Indic Res 130(3):1005–1023

Liu Y, Fullerton TM, Ashby NJ (2013) Assessing the impacts of labor market and deterrence variables on crime rates in Mexico. Contemp Econ Policy 31(4):669–690

Malinen T (2012) Estimating the long-run relationship between income inequality and economic development. Empirical Econ 42(1):209–233

Menezes T, Silveira-Neto R, Monteiro C, Ratton JL (2013) Spatial correlation between homicide rates and inequality: evidence from urban neighborhoods. Econ Lett 120(1):97–99

Mo PH (2000) Income inequality and economic growth. Kyklos 53(3):293–315

Mukherjee S (2019) Crime and Social Deprivation across States in India–insights from a Panel Data Discourse on Social Sustainability’. The Impact of Global Terrorism on Economic and Political Development. Emerald Publishing Limited, pp. 249–265

Mulok D, Kogid M, Lily J, Asid R (2017) The relationship between crime and economic growth in Malaysia: re-examine using bound test approach. Malays J Business Econ (MJBE) 3(1):15–26

Musavengane R, Siakwah P, Leonard L (2019) “Does the poor matter” in pro-poor driven sub-Saharan African cities? towards progressive and inclusive pro-poor tourism. Int J Tourism Cities 5(3):392–411

Neumayer E (2003) Good policy can lower violent crime: evidence from a cross-national panel of homicide rates, 1980–97. J Peace Res 40(6):619–640

OSAC (2017) Morocco 2017 crime & safety report: rabat. Overseas Security Advisory Council, Washington DC

Ouimet M (2012) A world of homicides the effect of economic development, income inequality, and excess infant mortality on the homicide rate for 165 countries in 2010. Homicide Stud 16(3):238–258

Pane H (2017) The Social Problems of National Poverty and Criminality in Indonesia. Int J Soc Sci Human Invention 4(8):3834–3836

Piatkowska SJ (2020) Poverty, inequality, and suicide rates: a Cross-National Assessment of the Durkheim Theory and the Stream Analogy of Lethal Violence. Soc Q. 19:1–26

Popa AM (2012) The impact of social factors on economic growth: empirical evidence for Romania and European Union countries. Rom J Fiscal Policy (RJFP) 3(2):1–16

Pridemore WA (2011) Poverty matters: a reassessment of the inequality–homicide relationship in cross-national studies. Br J Criminol 51(5):739–772

Ranis G, Stewart F, Ramirez A (2000) Economic growth and human development. World Dev 28(2):197–219

Saasa OS (2018) Poverty profile in sub-Saharan Africa: the challenge of addressing an elusive problem. Contested terrains and constructed categories. Routledge, Abingdon, pp 105–116

Sachsida A, de Mendonça MJC, Loureiro PR, Gutierrez MBS (2010) Inequality and criminality revisited: further evidence from Brazil. Empirical Econ 39(1):93–109

Saleh B (2011) Tunisia’s economic medicine, poverty and unemployment. https://www.pambazuka.org/governance/tunisias-economic-medicine-poverty-and-unemployment . Accessed 6 October 2018

Solinger DJ, Hu Y (2012) Welfare, wealth and poverty in urban China: the Dibao and its differential disbursement. China Q 211:741–764

South China Monitoring Report (2015). Big trouble in rural China: data reveals greater the wealth gap the higher the crime rate, and Hong Kong is feeling the effects. Law and crime section. https://www.scmp.com/news/hong-kong/law-crime/article/1876931/big-trouble-rural-china-data-reveals-greater-wealth-gap . Accessed 6 October 2018

Stamatel JP (2016) Democratic cultural values as predictors of cross-national homicide variation in Europe. Homicide Stud 20(3):239–256

Thamman B (2017) Causes of poverty in Uruguay. The Borgen project. https://borgenproject.org/causes-of-poverty-in-uruguay/ . Accessed 6 October 2018

Thorbecke E, Charumilind C (2002) Economic inequality and its socioeconomic impact. World Dev 30(9):1477–1495

Ulriksen MS (2012) Questioning the pro-poor agenda: examining the links between social protection and poverty. Dev Policy Rev 30(3):261–281

Vellala PS, Madala MK, Chattopadhyay U (2018) Econometric analysis of growth inclusiveness in India: evidence from cross-sectional data. Advances in finance & applied economics. Springer, Singapore, pp 19–38

Vergara R (2012) Crime prevention programs: evidence from CHILE. Dev Econ 50(1):1–24

Wang C (2017) Which dimension of income distribution drives crime? Evidence from the People’s Republic of China. Asian Development Bank, Working paper no: 704, Tokyo, Japan. https://www.adb.org/publications/which-dimension-income-distribution-drives-crime-prc . Accessed 6 October 2018

Wang H, Yao H, Kifer D, Graif C, Li Z (2017) Non-stationary model for crime rate inference using modern urban data. IEEE Trans Big Data 5(2):180–194

World Bank (2013). Brazil Fights Crime while Bringing Development to the Favelas. http://www.worldbank.org/en/news/feature/2013/03/21/brazil-crime-violence-favela . Accessed 6 October 2018

World Bank (2015) World Development Indicators. World Bank, Washington DC

Zaman K (2018) Crime-poverty nexus: an intellectual survey. Forensic Res Criminol Int J 6(5):327–329

Zaman K, Khilji BA (2014) A note on pro-poor social expenditures. Qual Quant 48(4):2121–2154

Zaman K, Usman B, Sheikh SM, Khan A, Kosnin ABM, Rosman ASB, Hishan SS (2019) Managing crime through quality education: a model of justice. Sci Justice 59(6):597–605

Zaman K, Al-Ghazali BM, Khan A, Rosman ASB, Sriyanto S, Hishan SS, Bakar ZA (2020) Pooled mean group estimation for growth, inequality, and poverty triangle: evidence from 124 Countries. J Poverty 24(3):222–240

Download references

Acknowledgements

The authors are thankful for King Saud university research project number (RSP-2019/87) for funding the study. The authors are indebted to the editor and reviewers for constructive comments that have helped to improve the quality of the manuscript.

Researchers Supporting Project number (RSP-2019/87), King Saud University, Riyadh, Saudi Arabia.

Author information

Authors and affiliations.

Department of Public Administration, Xi’an University of Architecture and Technology, Xi’an, China

Muhammad Khalid Anser

Higher Education Department Khyber Pakhtunkhwa, Government College of Management Sciences, Abbottabad, Pakistan

Zahid Yousaf

Department of Management, College of Business Administration, King Saud University, P.O. Box 71115, Riyadh, 11587, Saudi Arabia

Abdelmohsen A. Nassani & Saad M. Alotaibi

Department of Management, Aleppo University, Aleppo, Syria

Ahmad Kabbani

Department of Economics, University of Wah, Quaid Avenue, Wah Cantt, Pakistan

Khalid Zaman

You can also search for this author in PubMed   Google Scholar

Contributions

The collection, processing and analysis of the data were carried out by the authors. All authors read and approved the final manuscript

Corresponding author

Correspondence to Khalid Zaman .

Ethics declarations

Competing interests.

The authors declare that they have no conflict of interest.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

See Table  7 , Figs.  2 , 3 , 4 and 5 .

Source: World Bank ( 2015 )

Data trend at level.

Source: World Bank ( 2015 ). ‘D’ indicates first difference

Data trend at first differenced

Source: authors’ estimation. Note: ‘D’ shows first difference, while ‘LOG’ represents natural logarithm

Plots of IRF.

VDA Estimates.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Anser, M.K., Yousaf, Z., Nassani, A.A. et al. Dynamic linkages between poverty, inequality, crime, and social expenditures in a panel of 16 countries: two-step GMM estimates. Economic Structures 9 , 43 (2020). https://doi.org/10.1186/s40008-020-00220-6

Download citation

Received : 03 February 2020

Revised : 09 March 2020

Accepted : 29 May 2020

Published : 05 June 2020

DOI : https://doi.org/10.1186/s40008-020-00220-6

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Pro-poor growth
• Income inequality
• Quality education
• Healthcare expenditures
• System panel GMM

Jel classification