ESUT
JSS

co iaS l  f So c l iea nJnrs ocue

co iaS l  f So c l iea nJnrs ocue

co iaS l  f So c l iea nJnrs ocue SE SS JUT

Volume 5, Issue 2, 2020

Enugu State University 
of Science & Technology
Journal of Social Sciences

& Humanities

F SO C IY ET NIS CER  E AV NI

DN  U T EE CT HAT N

S O

U L OG

GU

YNE

Journal of Social Sciences
Published by the Faculty of the Social Sciences

published by

Faculty of Social Sciences

Enugu State University of Science & Technology
www.esutjss.com

Editor-In-Chief
Prof. Oby Omeje

Managing Editor
Barnabas E. Nwankwo, Ph.D

Associate Editor
Nicholas Attamah, Ph.D

F SO C IY ET NIS CER  E AV NI

DN  U T EE CT HAT N

S O

U L OG

GU

YNE

F SO C IY ET NIS CER  E AV NI

DN  U T EE CT HAT N

S O

U L OG

GU

YNE

F SO C IY ET NIS CER  E AV NI

DN  U T EE CT HAT N

S O

U L OG

GU

YNE

Journal of Social SciencesJournal of Social SciencesJournal of Social Sciences
Published by the Faculty of Social Sciences

F SO C IY ET NIS CER  E AV NI

DN  U T EE CT HAT N
S O

U L OG
GU

YNE

F SO C IY ET NIS CER  E AV NI

DN  U T EE CT HAT N
S O

U L OG
GU

YNE

Journal of Social SciencesJournal of Social Sciences



 

 

 

1 

 

Effects of Ethnic Diversity and Poverty on Crime Rate in Nigeria 

 

Adenuga Fabian Adekoya 
Department of Economics,  

School of Arts and Social Sciences,  

Michael Otedola College of Primary Education, Noforija-Epe, Lagos, Nigeria. 

Email: adenugaadekoya@gmail.com 

& 
 

Nor Azam Abdul Razak 
Department of Economics,  

School of Economics, Finance and Banking,  

Universiti Utara Malaysia Sintok 06010, Malaysia 

 

  

Abstract 

Ethnic diversity in Nigeria and its effects on the poor is endemic. Among the various ethnic 

diversity, there is high rate of criminal activities and poverty. This study examine the effects 

of ethnic diversity and poverty on crime in Nigeria. Further, this study tested annual time 

series based on Nigeria data from 1970 to 2017. While noting endogeneity in the nature of 

the variable based on literature. The Autoregressive Distributed Lag Model Instrumental 

Variables used to estimate the model. Estimated result showed that the variables co-

integrated at the 1% level of significance. In addition, ethnic heterogeneity, poverty, and 

under five mortality significant to cause crime. Income growth and security expenditure have 

significant reduction effect on crime. Thus, this paper suggested to the policy makers to 

embark on income generation and good health care programmes for the poor. Also, adequate 

and up-to-date skills and apparatus be given to the security personnel to control crime. 

Keywords: Ethnic diversity, poverty, income growth, security expenditure, crime,  

 

 

Introduction 

Shaw and McKay (1942) contended that the society disorganized especially when there was 

breakdown of bonds, norms, and order. That disorderliness took place because the new poor 

migrants in urban areas are in majority in any ethnic population confronted with residential 

immobility and ethnic heterogeneity. As they are challenged with severe poverty over the 

minority population that is well to do. Further, the problem of economic deprivation made 

ethnic heterogeneity problem to surface on the pretext to satisfy their economic needs. By 

this, they jettison the norms and orders in the urban area/society- a situation that often leads 

to violence and criminal activities. In testing the assertion, many researchers have examined 

the relationship between ethnic heterogeneity and crime; while some studies found positive 

linear empirical supports (Ebbe, 1989; Osgood & Chambers, 2000; Altheimer, 2008; 

Sandoval, 2018), Trawick & Howsen, (2006) found negative linear empirical support as De 

mailto:adenugaadekoya@gmail.com


Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

Soysa and Noel (2018) found an inverted U-shape when testing the linear and non-linear 

relationship; their mixed findings pave room for further study to be conducted. Therefore, 

this study aim to examine the effects of ethnic diversity and poverty on crime rate in Nigeria. 

 

In filling the existing gap in the literature, this study analysed how ethnic diversity and 

poverty determines crime rate using a time series from 1970 to 2017 especially with the aim 

to answer a pertinent research question how does ethnic diversity and poverty affect crime 

rate in Nigeria? The answering to the question demanded attention because Brennan-Galvin 

(2002) discovered that violence usually erupted among the poor in the different ethnicity 

residing in cities like Kano, Lagos and other major cities in Nigeria. Ever since, researchers 

in Nigeria has not considered the effects of ethnic diversity and poverty on crime rate. 

Moreover, Ebbe (1989) concluded that high crime and delinquency rates was due to 

heterogeneity. The heterogeneity is made of segregation in residence in terms of low, 

medium and high-income residents’ areas but it limited to Lagos metropolis. Similarly, 

Badiora et al. (2014) showed that the poor among the inhabitants of Ile-Ife which have mix of 

different ethnic groups involved in violent crime (Badiora et al., 2014). 

 

The contributions of this study are numerous; firstly, it revealed the roles play by ethnic 

heterogeneity and poverty in Nigeria; secondly, it pointed the devastating effects of crime in 

the country which increase the cost of operation and national budget spending on security, 

discourage investment in the country; thirdly, it provided guide to the policy makers, 

investors and organisation to take the issues of deterrence and adequate investment in health 

as serious matter in the country; fourthly, it provided a viable document to the literature on 

crime studies in Nigeria; and lastly, this research served as insight to the policymakers on the 

need to reduce crime and poverty in the country. The remaining part of this paper is divided 

into five sections: Section two considered literature review, Section three discussed the 

empirical strategy, Section four presented the results, Section five focused on discussion and 

Section six is the conclusion of the study.  

 

Literature Review 

Conceptual Issues 

Crime rate in Nigeria 

What constituted immoral/illegal and criminal acts depends on the norms/values or standard 

set by any society (Meera & Jayakumar, 1995).It is an illegal act motivated by incentives for 



 

3 

 

self-enrichment in case of property crime and by hate or passion in the case of person’s crime 

(Ehrlich, 1973). It means any exhibited acts which is against the rules and regulation in a 

define territory either it may be injurious or not is referred to as crime. It is often measured  

as 𝐶𝑟𝑖𝑚𝑒 𝑟𝑎𝑡𝑒 =
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐶𝑟𝑖𝑚𝑒

𝑇𝑜𝑡𝑎𝑙 𝑃𝑜𝑝𝑙𝑢𝑎𝑡𝑖𝑜𝑛
× 100,000. 

 

Various illegal act existed in Nigeria, among them are murder, assault and burglary as 

presented in Table 1. For instance, the average murder rate from 2.16 per 100,000population 

in the period of 1980-1989 increased to 3.89 per 100,000 population in the period of 2000-

2009, but reduced to 1.64 per 1000,000 population in the period of 2010-2017. The growth 

rate of armed robbery reduced by 47.59% and 42.38% in the period of 1980-1989 and 2000-

2009 but increased by 53.08% in the period of 2010-2017.Considering the share of total 

crime cases among regions crime in Nigeria from 2016 to 2017 as presented in Figure 1; the 

South East has the highest 10.09% change in share of total crime while the least change in 

share of -6.21% occurred in North Central. But, the South West has the highest recorded of 

45.88% in 2016 and 45.31% in 2017 while the North West has 7.87% share in total crime 

cases in 2016.  

There are effects of crime rate in the country; for instance, business and foreign investment 

suffers losses worth billions of dollars (Brennan-Galvin, 2002). In 2008, 191.62 million 

metric tons valued at ₦14.594billion lost because of Niger Delta militants’ rebellious act 

(Nigeria Bureau of Statistics (NBS), 2011).Between 2006 and 2011, an average death per 

person of 1,655 recorded and the killings by armed robbers was 50% of 8,516 deaths (Nigeria 

Watch, 2011; Nwankwo & James, 2016). Besides, several numbers of police officers 

murdered by militants and armed bandits (Ojedokun, 2014). Thus, crime amount to wastage 

of resources and human potentials. Also, the criminality activity affected growth and public 

service provision, distorts the environment for engaging in business and functions as a tax on 

private investments in Nigeria (Okonjo-Iweala & Osafo-Kwaako, 2007). 

 



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

Table2.1: Crime Types in Nigeria 

Types of Crime Average Rate of Crime Growth Rate of Crime 

 1970-

1979 

1980-

1989 

1990-

1999 

2000-

2009 

2010-

2017 

1970-

1979 

1980-

1989 

1990-

1999 

2000-

2009 

2010-

2017 

Murder 1.99 2.16 1.50 3.89 1.64 38.69 -28.63 -8.79 1549.86 26.49 

Felonious 

Wounding 

10.39 16.81 15.13 11.48 5.28 193.16 -27.70 -18.34 -35.81 20.61 

Assault 78.45 61.82 44.89 20.18 10.92 35.87 4.73 -43.13 -68.13 58.13 

Armed Robbery 2.637 1.79 1.78 2.02 1.76 -13.60 -47.59 13.74 -42.38 53.08 

Burglary 25.61 32.40 18.83 8.16 3.37 51.00 94.90 -52.63 -24.20 -67.54 

Other Crimes 

against Persons 

6.035 17.92 20.17 9.91 15.62 -55.89 729.66 -48.51 -74.75 333.28 

Other Crimes 

against 

Property 

15.64 115.56 27.88 27.72 19.70 74.27 3281.77 -36.15 316.33 817.16 

Other Minor 

offences/crimes 

16.69 62.52 7.05 1.43 3.48 446.61 -36.83 -34.46 -97.49 4526.94 

Total Crime 

Rate 

306.44 370.37 238.41 107.58 67.38 26.053 20.29 -49.30 -56.04 4.49 

Note: above elaboration should not be cited without making reference to the author. 

Source: Author computation based on crime report from the Nigeria Bureau of Statistic. 

 

 
Figure 1: Share of total crime cases among regions crime in Nigeria 

Author: Author computation based on NBS crime statistics 

 

Ethnic diversity in Nigeria 

Diversity is significant to determine the stability and prosperity of a country. As ethnic 

diversity is a concept that show diverse attributes in a nation. These attributes include 

minority, race, class and group in a country. The group can include immigrants or settlers and 

other nationalities (Saqib, 2018). Ethnicity emphasis culture sharing, languages and religion. 

It can cause distrust in country thereby making each group hate one another (De Soysa & 

Noel, 2018). Such hatred make it difficult for the policy makers to manage the ethnic tension 

(Saqib, 2018). It is thus associated with violence. Ethnic diversity measured by ethnic 

-10

0

10

20

30

40

50

North West North Central North East South East South West South South

(%
)

2016 2017 % Change in Share of Total Cases



 

5 

 

fractionalization index. The index reflects the likelihood that two people chosen at random 

within a given country will be from different ethnic groups. 

 

Heterogeneity effects increases in Nigeria because most urban centres were not planned. 

Urban dwellers face acute shortage of housing, a situation that led to more slums created. A 

large number of people facing socioeconomic deprivation resides in slum areas. As slum 

areas accommodate the poor from the different ethnic population.  Heterogeneity became 

worse as residential classified into high, medium, and low-income in Nigeria (Ebbe, 1989; 

Opoko & Oluwatayo, 2014). And the poor in urban areas faced high level of job immobility. 

This reduces social bond due to new contacts made (Shopeju, 2007). Thus, poor social 

integration increased among migrants from rural to urban areas. This creates difficult 

situation for the security personnel. This is because available public facilities and resources 

overstretched and become inadequate. It becomes difficult to apprehend criminals given the 

opportunity to commit crime. 

 

Poverty in Nigeria  

The report of Nigeria Bureau of Statistics of various years indicated that high poverty rate 

exist in the country (see Table 2.3). Between 1980 and 2010, poverty rates increased from 

27.2% in 1980 to 46.3% in 1985 and by 1996, it increases to 65.6%; this later rose to 69% in 

2010. Further, the magnitudes of poverty in Nigeria showed that poverty in the urban areas 

intensified but not like the poverty in the rural areas. Also, a cursory look at Table 2.3 

revealed that the urban poverty rose from 43.2% in 2004 to 73.2% in 2010 while the rural 

poverty increased from 63.3% in 1980 to 73.2% in 2010. However, various measures to 

reduce poverty have been adopted in the country as highlighted in Obadan (2002) but these 

measures seems not adequate (Umukoro, 2013). This is because the available social 

protection in Nigeria were weak to move the large number of population out of poverty; that 

is, halving number of people living below US$ 1 a day and reduction of poverty to 21.40% 

proved difficulty (MDG Report, 2013). Thus, the prevalence of poverty in Nigeria may 

continue to encourage high poor economic development in the country. 

 

 



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

TABLE 2.3: Spread and Trend in Poverty Levels (%) in Nigeria 

Years 1980 1985 1992 1996 2004 2010 

Levels       

NATIONAL  27.2 46.3 42.7 65.6 54.4 69 

Urban  17.2 37.8 37.5 58.2 43.2 61.8 

Rural  28.3 51.4 46 69.3 63.3 73.2 

ZONE         

South South 13.2 45.7 40.8 58.2 35.1 63.8 

South East  12.9 30.4 41 53.5 26.7 67 

South West  13.4 38.6 43.1 60.9 43 49.8 

North Central  32.2 50.8 46 64.7 67 67.5 

North East  35.6 54.9 54 70.1 72.2 76.3 

North West  37.7 52.1 36.5 77.2 71.2 77.7 

Source: National Bureau of Statistics 2010. 

 

Empirical Review 

Hansmann and Quigley (1982) observed in a cross-country study of 58 and 40 that population 

heterogeneity in the cultural group due to interaction among different ethnicity has a 

significant positive effect on crime of homicide while that of religion and language has an 

individual impact in reducing crime of homicide using ordinary least square method. Income 

inequality as not too distant from poverty detected to increase the crime of homicide as 

income per capita has a reducing effect on homicide. Besides, when the population of young 

people is higher with a low level of per capita income would further exacerbate the rate of 

homicide. Sandoval (2018) concluded that demographic disparity especially among the 

young population and disagreement between armed gangs increases rate of homicide in 

Colombia.  

 

Cherry (1999) conducted a panel study in the United State by asserting that it is improper to 

leave unobserved heterogeneity in a crime model. The unobserved heterogeneity as minority 

population measures it estimated along with other socioeconomic factors and deterrence 

measures with the random/fixed-effect method. The result indicated that unobserved 

heterogeneity is significant to increase crime while unemployment increase all types of 

crime, income only has a reducing effect on murder. The deterrence measures vary in the 



 

7 

 

outcome, while the probability of arrest has a reducing impact on all types of crime, the 

length of sentence is not significant but the presence of Police in the crime environment is 

only significant and negative on murder and positive on larceny. The study results revealed 

that when unobserved heterogeneity is not considered, the deterrence measures are overstated 

in the model, therefore, studying the minority population along with the municipal population 

provided opportunity to see the individual effect of the deterrence measures on various types 

of crime. Zhang (2016) included the use of police and incarceration showed reducing the 

effect on crime when taking note of unobserved heterogeneity. Similarly, Cerro & Meloni 

(2000) found that Police arrest and conviction of criminals led to crime reduction in 

Argentina in the prevalence of high and inequality causing crime but, found that high income 

area attract criminal activities as income increases crime. 

 

Trawick and Howsen (2006) examined community heterogeneity on the crime rate in the 

United State using a Tobit analysis. Community heterogeneity measured with race, ethnicity 

and religion. The result showed that race/ethnic and religion are significant and negative on 

the crime rate. This means a homogeneous race community is not prone to criminal activities 

because when community share the same norms, cultures, beliefs and teaching such 

community would experience low criminal activities; this supports the sociological theory 

and thus, the community would require small resources to control crime. However, the selling 

of alcohol spirit in the community would promote criminal activities, which include larceny, 

burglary, autotheft,and robbery. The reducing effect of religion on crime by Trawick and 

Howsen (2006) provided support for Hansmann and Quigley (1982). 

 

Altheimer (2008) examined ethnic heterogeneity and interacted it with social supports on the 

crime of homicide in a cross-country study of 51. Using the estimation of weighted least 

square method; the study revealed that ethnic heterogeneity and economic inequality causes 

homicide in any form of social support introduced. However, when ethnic heterogeneity 

interacts with health care and human development index at a different time, the homicide rate 

reduces. Likewise, the general support and income per capita significant reduce the homicide 

rate. Similar results found in Ouimet (2012), Ouimet (2012) studied ethnic heterogeneity and 

socioeconomic factors on the crime of homicide in a cross-country study of 165, which 

include low, medium, and high income in multiple regression. The result showed that ethnic 

heterogeneity increases homicide, which further supported with the evidence of positive 

significant impact on homicide in medium income countries. Further, poverty measured with 



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

an excess of under-five mortality rate causes homicide and likewise in high-income countries; 

but income inequality causes homicide in medium countries and the overall countries. 

However, De Soysa and Noel (2018) examined the linked between ethnic diversity and 

violent crime in study of 140 countries from 1995 to 2013; the data analysed using a robust 

panel Ordinary Least Squares. Their result indicated that ethnic heterogeneity has an inverted 

U-shape relationship with violent crime. While they contended that existence of polarisation 

increased the risk of armed violence, they concluded that ethnic heterogeneity was not 

responsible for social dislocation happened in the society. Further, they found that GDP per 

capita reduces violent crime and income inequality did nor drive violent crime. 

 

The empirical finding in Hansmann and Quigley (1982), Trawick and Howsen (2006), Cherry 

(1999), Altheimer (2008) and Sandoval (2018) supported that ethnic heterogeneity causes 

crime which support the social disorganization theory. Also, studies with measures of poverty 

provide support to the social disorganization and rational choice theory (Hansmann & 

Quigley, 1982; Altheimer, 2008 and Ouimet, 2012). But, De Soysa and Noel (2018) 

demonstrated a contrary result, and thus concluded that ethnic heterogeneity was not 

responsible for violent crime. From the previous studies, it is noted that most of the studies 

are cross country studies and based on developed nations and not in developing country like 

Nigeria. Also, this study used time series and thus, differed from previous studies that used 

cross country data, cross-section and time series. 

 

Empirical Strategy 

Theoretical Framework 

This study used the crime supply function as proposed by Becker (1968), this model 

presented in equation 1 and it helps to determine the reasons for criminal activities. In the 

model, determinants are into two forms; while the first form caused crime, the other form 

sees crime reduction, which referred to, as deterrence. In equation 1,  𝐶𝑅𝑡  is the total number 

of crime which depends on 𝑃𝑅𝑡, 𝑃𝐴𝑡 and𝑈𝑡. The 𝑈𝑡 represents socioeconomic factors that 

motivates crime or increases the opportunity cost of crime in a society, which include 

poverty. As mentioned earlier, the second form are various deterrence policy such as 

prosecution and arrest (𝑃𝑅𝑡) and punishment through the imprisonment (𝑃𝐴𝑡). Similar 

studies that have employed this model includes Cherry (1999), Trawick and Howsen (2006).  



 

9 

 

𝐶𝑅𝑡 = 𝑓(𝑃𝑅𝑡, 𝑃𝐴𝑡  , 𝑈𝑡)(1)                                                                                                   (1) 

 

The model in equation 1extended to study racial and ethnic homogeneity along with other 

socioeconomic factors by Trawick and Howsen (2006) presented in equation 2, which did not 

include any of the deterrence policy. In equation 2, 𝐶𝑅𝑡 is crime rate per person in a 

community, 𝑅𝐸𝐺is the religious factor;𝐺𝐼𝑁𝐼𝑡represents the income inequality, indicates the 

ethnic homogeneity for community, and  𝑍𝑡includes other socio variables in the model.  

 

𝐶𝑅𝑡 = 𝑓(𝑅𝐸𝐺𝑡, 𝐺𝐼𝑁𝐼𝑡 , 𝐻𝐸𝑇𝑡 , 𝑍𝑡) (2) 

 

Following Trawick and Howsen (2006), we modified Equation 2which led to equation 3 by 

testing heterogeneity (𝐻𝐸𝑇𝑡), poverty (𝑃𝑂𝑉𝐶𝑡), and under five mortality (𝑀𝑅𝑈𝑡), the growth 

rate of income per capita (𝐺𝑅𝑃𝐶𝑡) and public expenditure on internal security (𝑃𝐸𝑆𝑡) on the 

crime rate (𝐶𝑅𝑡). This equation 3 includes public expenditure on internal security (𝑃𝐸𝑆𝑡) as a 

deterrence policy suggested in equation 1 by Becker (1968).  

 

𝐶𝑅𝑡 = 𝑓( 𝐻𝐸𝑇𝑡, 𝑃𝑂𝑉𝐶𝑡, 𝑀𝑅𝑈𝑡,   𝐺𝑅𝑃𝐶𝑡, 𝑃𝐸𝑆𝑡) (3) 

 

Where the total crime rate in the country is𝐶𝑅𝑡, 𝐻𝐸𝑇𝑡 is the ethnic heterogeneity. 

𝑃𝑂𝑉𝐶𝑡ispoverty rate and 𝑀𝑅𝑈𝑡is poverty of under-five mortality, 𝐺𝑅𝑃𝐶𝑡 is growth in 

income and 𝑃𝐸𝑆𝑡is the public expenditure on internal security. In this model, ethnic 

heterogeneity (𝐻𝐸𝑇𝑡), poverty rate (𝑃𝑂𝑉𝐶𝑡) and the under five mortality (𝑀𝑅𝑈𝑡) expected to 

increase crime rate with positive sign since this increases the chances to commit crime. 

Growth rate of income (𝐺𝑅𝑃𝐶𝑡) and the public security expenditure (𝑃𝐸𝑆𝑡) expected to be 

negative as an anti-dote to crime because it expects to reduce the chances to commit crime.  

 

Model Specification 

In line with the crime model expressed in equation 3, equation 4 thus specified. In this model, 

𝑙𝑛𝐶𝑅𝑡 represents log of the crime rate variable; 𝐻𝐸𝑇𝑡 represents the ethnic diversity index, 

𝑃𝑂𝑉𝐶𝑡(%) and 𝑀𝑅𝑈𝑡 represents the poverty rate and under five infants mortality 

respectively.𝐺𝑅𝑃𝐶𝑡 denotes growth rate of income (%) and 𝑙𝑛𝑃𝐸𝑆𝑡 stands for log of security 

expenditure. a 𝛽0 𝑖𝑠 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, 𝛽1, 𝛽2 ,𝛽3  , 𝛽4𝑎𝑛𝑑 𝛽5   are parameters and𝜇𝑡 is the white 

noise.  



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

𝑙𝑛𝐶𝑅t= 𝛽0 + 𝛽1𝐻𝐸𝑇𝑡 + 𝛽2𝑃𝑂𝑉𝐶𝑡  + 𝛽3𝑀𝑅𝑈𝑡 + 𝛽4𝐺𝑅𝑃𝐶𝑡 + 𝛽5𝑙𝑛𝑃𝐸𝑆𝑡 +  𝜇𝑡(4) 

 

Cointegration Test 

To test the model in equation 4, the preliminary is to conducted unit root tests due to the 

annual time series employed in the study, and sequel cointegration test. This study conducted 

unit root test usingAugmented Dickey-Fuller (ADF) and Phillip Perron tests. Based on these 

tests all the variables are integrated at the level I(0) and first difference I(1) (see Table 3 in 

section 5.0). Pesaran, Shin, and Smith (2001) proposed that a bound test approach be used to 

examine the exist of long run relationship specifically when the result of the unit root tests 

showed that variables are integrated at I(0) and I(1). Besides, the existence of high correlation 

among the variables in a model suggests high probability of endogeneity. The bound test 

minimised such problem of endogeneity in the model. This is because the bound test 

approach has the ability to generate sufficient lags for variables in the model and sufficiently 

provide for means to ascertain residual correlation. Once there is no serial correlation then 

endogeneity issue is treated. The dynamism based on the transformation of the variable at the 

period of one lag in the model using the optimal lag length. These variables in the model 

transformed based on the Akaike information criterion (AIC) due to the small sample size 

used in this study (see Adekoya & Abdul Razak, 2017). Liew (2004) found that AIC is most 

suitable to decide the optimal lag length for the variable in smaller sample size. Further, the 

Autoregressive Distributed Lag Model (ARDL) Instrumental Variable (this discussed in the 

next section) used because it is capable of providing the short-run and long run at the same 

time. 

 

Using the ARDL framework in equation 5, the F-statistic tested the joint significance of the 

coefficients at one period of lag and the result presented in Table 3. The null hypothesis of no 

co-integration shows that 𝐻0 : 𝛽1 =  𝛽2 = 𝛽3 = 𝛽4 = 𝛽5 = 𝛽6 = 0 (implies non-existence of 

cointegration) and the alternative is 𝐻𝐴 : 𝛽1 ≠ 𝛽2 ≠ 𝛽3 ≠ 𝛽4 ≠≠ 𝛽5 ≠ 𝛽6 ≠ 0 where at least 

one of the 𝛽1 𝑡𝑜 𝛽6 ≠ 0  (implies the existence of cointegration). 𝛾1 𝑡𝑜 𝛾6  signifythe short-

run dynamics of the ARDL model in equation 6.  

 



 

11 

 

∆𝑙𝑛𝐶𝑅𝑡 = 𝛽0𝐴 + 𝛽1𝑙𝑛𝐶𝑅𝑡−1 + 𝛽2𝐻𝐸𝑇𝑡−1 + 𝛽3𝑃𝑂𝑉𝐶𝑡−1 + 𝛽4𝑀𝑅𝑈𝑡−1 + 𝛽5𝐺𝑅𝑃𝐶𝑡−1

+ 𝛽6𝑙𝑛𝑃𝐸𝑆𝑡−1 + ∑ 𝛾1 ∆𝑙𝑛𝐶𝑅𝑡−1

𝑝

𝑖=1

+ ∑ 𝛾2 ∆𝐻𝐸𝑇𝑡−1

𝑝

𝑖=0

+ ∑ 𝛾3 ∆𝑃𝑂𝑉𝐶𝑡−1

𝑝

𝑖=0

 

+ ∑ 𝛾4 ∆𝑀𝑅𝑈𝑡−1

𝑝

𝑖=0

 + ∑ 𝛾5 ∆𝐺𝑅𝑃𝐶𝑡−1

𝑝

𝑖=0

 + ∑ 𝛾6 ∆𝑙𝑛𝑃𝐸𝑆𝑡−1

𝑝

𝑖=0

   +  𝜇𝑡            (5) 

 

∆𝑙𝑛𝐶𝑅𝑡 = 𝛽0𝐵 + ∑ 𝛾1 ∆𝑙𝑛𝐶𝑅𝑡−1

𝑝

𝑖=1

+ ∑ 𝛾2 ∆𝐻𝐸𝑇𝑡−1

𝑝

𝑖=0

+ ∑ 𝛾3 ∆𝑃𝑂𝑉𝐶𝑡−1

𝑝

𝑖=0

 + ∑ 𝛾4 ∆𝑀𝑅𝑈𝑡−1

𝑝

𝑖=0

 

+ ∑ 𝛾5 ∆𝐺𝑅𝑃𝐶𝑡−1

𝑝

𝑖=0

 + ∑ 𝛾6 ∆𝑙𝑛𝑃𝐸𝑆𝑡−1

𝑝

𝑖=0

   +  𝜇𝑡𝐵 

Endogeneity and Bewley Model  

Sequel to the bound test used to establish the cointegration among these variables, it is 

relevant to reaffirm that endogeneity problem is at stake among these variables. This is due to 

the significant high correlation value among some variables presented in Table 2, which 

include crime, poverty measurements, and security expenditure. Likewise, Mehlum, Miguel, 

and Torvik (2006) did not rule out that the causal effects of poverty on property crime are 

potentially subject to bias due to endogeneity and omitted variables. Also, Levitt (1997), 

Bourguignon (1999), Adekoya and Abdul Razak (2018)contended that expenditure on 

internal security have the issue of endogeneity with crime since high criminal activities may 

suggest that more security with increase insecurity expenditure. Therefore, model 5 (ARDL 

OLS) becomes sufficient provided there is no serial correlation but we are more concerned 

about the problem of simultaneity to avoid spurious regression. Therefore, to address 

potential bias caused by endogeneity and omitted variable bias, the ARDL Instrumental 

Variable (ARDL IV) was employed (although the estimated standard error in ARDL OLS is 

identical to ARDL IV, results are available on request). However, the test of homogeneity not 

conducted since there is the existence of significant high correlation among some of the 

variables. Notwithstanding, Stučka (2004) affirmed that once ARDL IV used for estimation, 

there is no need to test for endogeneity.   

 

Therefore, we dealt with the potential bias caused by endogeneity and omitted variable using 

ARDL instrument variable (IV) by rewritten the Bewley equation. The ARDL IV allows the 



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

estimated short and long run of ARDL using the lag of the dependent variable as the 

independent variable (Stučka, 2004). Therefore, this study overcomes the problem of whether 

the instrument should correlate with any of the variables. In addition, Inder (1993) concluded 

that when ECM (dynamic equation) reparametrized in Bewley equation endogeneity bias 

problems are insignificant and reasonably inconsequential. Equation 5 and 6 are 

reparametrized to generate equation 7 and 8 (see Pesaran & Shin, 1997). Thus, the estimated 

long run and short run model is expressed in equation 7 and 8 respectively based on ARDL 

selected model1, 1, 0, 0, 1, 0with the instruments as 1, 𝑙𝑛𝐶𝑅𝑡−1,𝐻𝐸𝑇𝑡−1 , 𝑃𝑂𝑉𝐶𝑡 , 𝑀𝑅𝑈𝑡 , 

𝐺𝑅𝑃𝐶𝑡−1 , 𝑙𝑛𝑃𝐸𝑆𝑡, ∆𝐻𝐸𝑇𝑡−1 , ∆𝐺𝑅𝑃𝐶𝑡. 

 

𝑙𝑛𝐶𝑅𝑡 =  
𝑎𝑐0

∅(1)
+ 𝜗1𝐻𝐸𝑇𝑡  + 𝜗2𝑃𝑂𝑉𝐶𝑡 + 𝜗3𝑀𝑅𝑈𝑡 + 𝜗4𝐺𝑅𝑃𝐶𝑡  + 𝜗5𝑙𝑛𝑃𝐸𝑆𝑡  

+
1

∅(1)
∑ 𝜗6 ∆𝐻𝐸𝑇𝑡

𝑚1−1

𝑖=0

 +
1

∅(1)
∑ 𝜗7 ∆𝐺𝑅𝑃𝐶𝑡

𝑚1−1

𝑖=0

+
1

∅(1)
∑ 𝜗8 ∆𝑙𝑛𝐶𝑅𝑡

𝑚2−1

𝑖=0

+
𝜂𝑡

∅(1)
                                                                                                               (7) 

 

 

∆𝑙𝑛𝐶𝑅𝑡 =
𝑎𝑑0

∅(1)
+ 𝜗9𝑃𝑂𝑉𝐶𝑡 + 𝜗10𝑀𝑅𝑈𝑡 + 𝜗11𝑙𝑛𝑃𝐸𝑆𝑡 + 𝜗12𝐻𝐸𝑇𝑡−1 + 𝜗13𝐺𝑅𝑃𝐶𝑡−1 

                         +
1

∅(1)
∑ 𝜗14 ∆𝐻𝐸𝑇𝑡

𝑚1−1

𝑖=0

+
1

∅(1)
∑ 𝜗15 ∆𝐺𝑅𝑃𝐶𝑡

𝑚1−1

𝑖=0

+
1

∅(1)
∑ 𝜗16 ∆𝑙𝑛𝐶𝑅𝑡−1

𝑚2−1

𝑖=0

   

+
𝜂𝑡

∅(1)
                                                                                                       (8)   

Data 

This study used annual time series from 1970 to 2017 to examine the effects of ethnic 

diversity and poverty on crime. Data on the total crime rate (𝐶𝑅𝑡) sourced from the Nigeria 

Bureau Statistics and the Nigeria Police Annual Reports of various publications. Crime rate 

define as total crime committed/total population multiply by 100,000. It is the interest of the 

researcher to disaggregate crime but these data are not available up to date. The poverty rate 

(𝑃𝑂𝑉𝐶𝑡) is the inverse of logarithm of per capita household consumption multiply by 100 and 

is based on data obtained from Economics and Research Department, United Nations. The 

index used because poverty time series is not available and thus, considered the disadvantage 



 

13 

 

index approach used in Mata and Bollman (2007). Other variables in the study are the under-

five mortality (𝑀𝑅𝑈𝑡) andincome growth (𝐺𝑅𝑃𝐶𝑡) which is the growth (%) in GDP per 

capita sourced from the World Development Indicator. Ethnic diversity (𝐻𝐸𝑇𝑡) proxy by 

ethnic violence index sourced from Marshall (2016), it is the magnitude score of episode(s) 

of ethnic violence involving the country in each year which is on the scale of 1 (lowest) to 10 

(highest) as non-occurrence of violence denoted with zero (0). This data used because ethnic 

violence is associated with ethnic diversity. As ethnic tensions based on hatred or 

discrimination often cause of civil unrest (Ouimet, 2012). In Nigeria, this lawlessness caused 

bloodletting ethnic squabble and secrecy in the entire political landscape (Dike, 2014). Public 

expenditure on internal security (𝑃𝐸𝑆𝑡) obtained from the Central Bank of Nigeria various 

annual reports as a percentage to total expenditure. Thus, crime and security expenditure data 

are in logarithm form. 

 

Results  

The descriptive statistics of variables employed in this study are presented in Table 1. It 

revealed that means of each variables lies between the minimum and maximum values; this 

means data are fit for further estimation. Moreover, the results in Table 2 shows high 

correlation exist among crime rate, poverty rate, under five mortality, and security 

expenditure rate. For instance, significant correlations of 89.85% exist between poverty rate 

and crime rate, 86.88% for poverty rate and security expenditure rate, and 94.97% in the case 

of security expenditure and crime rate. These high correlations suggested that these variables 

are endogenous and therefore the analysis required estimation technique, which can detect 

and minimize the problem of endogeneity such as the Autoregressive Distributed Lagged 

Model (ARDL) Two Stage Least Square. 

Table1 

Descriptive Statistics 

Variables  Observations  Median  Mean  Std. Dev.  Minimum  Maximum 

𝑙𝑛𝐶𝑅𝑡  48  5.5138  5.2235  0.6718  4.0904  6.1619 

𝐻𝐸𝑇𝑡  48  1.0000  1.3333  1.1361  0.0000  4.0000 

𝑃𝑂𝑉𝐶𝑡   48  8.3656  8.3289  0.1781  8.0117  8.5936 

𝑀𝑅𝑈𝑡  48  208.0500 190.8250   47.1498  100.2000  283.7000 

𝐺𝑅𝑃𝐶𝑡   48  1.6729  1.5310  7.6418 -15.4547  30.3565 

𝑙𝑛𝑃𝐸𝑆𝑡   48  22.2583  22.8628  2.7133  18.9513  27.2820 

 



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

Table 2 

Results of the Correlational Test 

Variables 𝑙𝑛𝐶𝑅𝑡 𝐻𝐸𝑇𝑡  𝑃𝑂𝑉𝐶𝑡  𝑀𝑅𝑈𝑡 𝐺𝑅𝑃𝐶𝑡  𝑙𝑛𝑃𝐸𝑆𝑡  

𝑙𝑛𝐶𝑅𝑡 1.000000      

𝐻𝐸𝑇𝑡  -0.156709 1.000000     

 (0.2875) -----      

𝑃𝑂𝑉𝐶𝑡  0.898596 -0.171890 1.000000    

 (0.0000) (0.2427) -----     

𝑀𝑅𝑈𝑡 0.873804 -0.296938 0.772282 1.000000   

 (0.0000) (0.0404) (0.0000) -----    

𝐺𝑅𝑃𝐶𝑡  -0.221820 0.086739 -0.227598 0.031927 1.000000  

 (0.1297) (0.5577) (0.1198) (0.8294) -----   

𝑙𝑛𝑃𝐸𝑆𝑡  -0.949730 0.232953 -0.868803 -0.903601 0.178475 1.000000 

 (0.0000) (0.1111) (0.0000) (0.0000) (0.2249) -----  

Note: the probability values are in parenthesis, whichindicates the significance level of the correlation value 

 

To ensure that the data employed is not having serial correlation, the Augmented Dickey-

Fuller (ADF) and Phillip Perron unit root tests are conducted. The result of the stationarity test 

as presented in Table3 showed that all the variables are integrated at the levelI (0) and first 

difference I(1). For instance,𝑙𝑛𝐶𝑅𝑡, 𝐻𝐸𝑇𝑡, 𝑃𝑂𝑉𝐶𝑡 and 𝑙𝑛𝑃𝐸𝑆𝑡 are stationary at I(1) as 𝑀𝑅𝑈𝑡 

and 𝐺𝑅𝑃𝐶𝑡 are stationary at I(0) in the two tests conducted with exception of 𝑀𝑅𝑈𝑡 that only 

stationary at I(0) under ADF test. Notwithstanding, this suggests the Bound Test approach in 

the study. It is possible to determine whether variables with different integration order would 

jointly move together using the Bound Test approach. 

 

Table 3 

Results of the Unit Root Test 

Note: the figures reported are t-ratio and showed the p-values of MacKinnon (1996) one-sided at various levels 

of significance. The asterisks (***) is at 1%. 

 

With the bound test approach to cointegration, the result showed that the null hypothesis of 

no cointegration is reject atthe1% level of significance. This is because the F-statistic of 

4.689 lies above the upper critical bound of 4.15 (see Table 4). That is, variables in the model 

are cointegrated. Moreover, the results of the long-run and short-run based on Bewley 

Variables Model Lag length Augmented Dickey-Fuller  

    Statistics Prob. value 

Phillip Perron                    

Statistics Prob. Value 

Decision 

𝑙𝑛𝐶𝑅𝑡 Intercept and trend 1 -7.057*** 0.000 -7.071*** 0.000 I(1) 

𝐻𝐸𝑇𝑡 Intercept and trend 1 -6.608*** 0.000 -6.951*** 0.000 I(1) 

𝑃𝑂𝑉𝐶𝑡 Intercept and trend 1 -9.879*** 0.000 -11.658*** 0.000 I(1) 

𝑀𝑅𝑈𝑡 Intercept and trend 0 -13.765*** 0.000 -2.283 0.4343 I(0) 

𝐺𝑅𝑃𝐶𝑡  Intercept and trend 0 -6.004*** 0.000 -6.005*** 0.000 I(0) 

𝑙𝑛𝑃𝐸𝑆𝑡  Intercept and trend 1 -7.309*** 0.000 -7.446*** 0.000 I(1) 



 

15 

 

equation using the ARDL instrument variable (IV) (see Pesaran & Shin, 1997) are presented 

in Table 4 along with the results of the diagnostic tests.  

 

The long run estimation revealed that ethnic heterogeneity causes crime at the 10% level of 

significance. This revealed that an increase in ethnic heterogeneity would increase the 

occurrence of crime in Nigeria. A similar result obtained in Altheimer (2008) and Ouimet 

(2012). In addition, poverty and under five mortality are both significant at the 5% level of 

significance. While an increase of 1% in the poverty rate leads to 0.92% in the crime rate, the 

under-five mortality   increases the crime rate by 0.004% point. Similar results obtained in 

the studies of Poveda (2012), Ouimet (2012), Rogers and Pridemore (2013). 

 

Moreover, economic growth is negative and significant in determining the crime rate at the 

10% level of significance. A rise of 1% in the growth rate of income as a measure of 

economic growth, the crime rate would reduce by 0.014%. Further, the result in Fajnzylber, 

Lederman and Loayza (2002), Deller and Deller (2012) are evidence that support the result. 

Nevertheless, Cerro and Meloni (2000) found non-significant impact of income growth on 

crime. The deterrence policy proxy by security expenditure is negative related to the crime 

rate at the 1% level of significance. That is, when security expenditure increases by 1%, the 

crime rate would reduce by 0.120%. This result is in line with Adekoyaand Abdul Razak 

(2018), Wu and Wu(2012). 

 

Related to the long-run results, the result of the short-run revealed that poverty and under five 

mortality are significant to determine crime at the 5% level of significance. This means an 

increase of 1% in poverty increases criminal activities by 0.468%. In addition, an increase in 

under five mortality   increases criminal activities by 0.002% point. Besides, the deterrence 

policy proxy with security expenditure is significant to reduce crime in the short-run at the 

1% level of significance. A rise by 1% in security expenditure would assist to increase the 

rate of crime detection with a high possibility to apprehend criminals thereby leading to a 

reduction of crimeby0.060%. 

 



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

Table 4 

Estimates of the ARDL Crime Model using Bewley’s regression form 

Dependent variable: Crime 

Critical bounds (F-test) 1% significance level Test of ARDL specification significance 

Lower  3.06 Adj-R-square=0.970 HQ=-1.146 

Upper  4.15 SB=-0.925 AIC*=-1.279 

ARDL specification 1, 1, 0, 0, 1, 0   

F-statistic 4.689   

Conclusion (sign level) Cointegrated (1%)   

K 5   

Long-run Estimates    

𝐶 -0.738 -0.181  

𝐻𝐸𝑇𝑡  0.061 1.793*  

𝑃𝑂𝑉𝐶𝑡  0.926 2.227**  

𝑀𝑅𝑈𝑡 0.004 2.284**  

𝐺𝑅𝑃𝐶𝑡 -0.014 -1.860*  

𝑙𝑛𝑃𝐸𝑆𝑡  -0.120 -2.885***  

∆𝐻𝐸𝑇𝑡  -0.122 -2.446**  

∆𝐺𝑅𝑃𝐶𝑡 0.009 1.808*  

∆𝑙𝑛𝐶𝑅𝑡 -0.978 -2.580**  

Short-run Estimates    

𝐶 -0.373 -0.180  

𝑙𝑛𝐶𝑅𝑡−1 -0.505 -5.216***  

𝐻𝐸𝑇𝑡  -0.030 -1.244  

𝑃𝑂𝑉𝐶𝑡  0.468 2.034**  

𝑀𝑅𝑈𝑡 0.002 2.021*  

   𝐺𝑅𝑃𝐶𝑡   -0.002 -0.824  

𝑙𝑛𝑃𝐸𝑆𝑡  -0.060 -2.725***  

𝐻𝐸𝑇𝑡−1 0.061 2.454**  

   𝐺𝑅𝑃𝐶𝑡−1   -0.004 -1.725*  

Diagnostics Tests 

𝑵𝒐𝒓𝒎𝒂𝒍𝒊𝒕𝒚 𝒕𝒆𝒔𝒕 (𝝌𝑵
𝟐 ) 𝐽𝐵 = 8.658 (0.013) 𝑲𝒖𝒓𝒕𝒐𝒔𝒊𝒔 𝒕𝒆𝒔𝒕 = 4.492  

𝑺𝒆𝒓𝒊𝒂𝒍 𝑪𝒐𝒓𝒓𝒆𝒍𝒂𝒕𝒊𝒐𝒏 𝒕𝒆𝒔𝒕 (𝝌𝑺𝑪
𝟐 ) 𝜒2 =   0.038 (0.845)   

𝑯𝒆𝒕𝒆𝒓𝒐𝒌𝒆𝒅𝒂𝒔𝒕𝒊𝒄𝒊𝒕𝒚 𝒕𝒆𝒔𝒕 (𝝌𝑯
𝟐 ) 𝜒2 = 9.048 (0.338)   

Note:the asterisk (*) showed that the estimated coefficients are significant at 1% (***); 5% (**) and 10% (*). 

Also, the null of the diagnostic tests 𝜒𝐻
2 , 𝜒𝑁

2 and 𝜒𝑆𝐶
2  arereject at the 5% level of significance. Variables are 

defined as crime is𝑙𝑛𝐶𝑅𝑡 as a dependent variable while ethnic heterogeneity (𝐻𝐸𝑇𝑡), poverty (𝑃𝑂𝑉𝐶𝑡), under 

five mortality  (𝑀𝑅𝑈𝑡), growth rate of income (𝐺𝑅𝑃𝐶𝑡) and security expenditure (𝑙𝑛𝑃𝐸𝑆𝑡) are independent 

variables. In the ARDL specification, the F-stat in the bounds test based on critical upper bounds. 

 

The diagnostic test indicated that long-run coefficients estimated are robust for policy 

implementation. These tests are normality; functionality; serial correlation; heteroscedasticity 

and the structural stability test. The normality test ( 𝜒𝑁
2  ) of Jarque-Bera showed that the 

model failed to pass the test at the 5% level of significance; but, the Kurtosis the model is 

4.492 which is in excess. Thus, the results of the Kurtosis were in excess which indicate that 

there is normal distribution following Saridakis (2011). In addition, this study found no traces 

of autocorrelation using the Chi-squared tool in LM test (𝜒𝑆𝐶
2 ) of Breusch-Godfrey Serial 



 

17 

 

Correlation at lag 1 at 5% level of significance. Also, the Chi-squared tool of Breusch-Pagan-

Godfrey Heteroskedasticity Test (𝜒𝐻
2 ) showed that the models passed this test at the 5% level 

of significance. 

 

Discussion 

The paper examined the effects of ethnic diversity and poverty on crime rate in the country. It 

is found that ethnic diversity and poverty causes crime; further, it confirmed that the poor 

among various ethnic diversities are frustrated because of the economic hardship, the stress 

induced on them encourage them to engage in crime. Similarly, Ebbe (1989) affirmed that 

children from the poor or slum area in Nigeria engage in crime. This is because economic 

deprivation is associated with a lack of opportunities, a situation that has caused much 

violence especially in major cities in the country (Brennan-Galvin, 2002). In most of the 

major cities, poor areas are characterised with high population density in heterogeneous form 

of ethnic diversity. This different group of population provide support for themselves even in 

the midst of one of them engaging in crime related matter. It therefore became it becomes 

uneasy for security agents to detect criminal among the high number of the poor population 

in the country. The possibility of this hinged on the fact the Nigeria Police are confronted 

with challenges that affect their efficiency; these are one, low intelligence capacity to gather 

facts relating to actual crime committed (Ojedokun, 2014) and two, there is a lack of modern 

equipment to combat crime in Nigeria (Otu, 2012).  The state of the economy in the country 

is fair to discourage crime but a high economy featured with high rate of poverty and under 

five mortality   required efforts to invigorate it. This is because when an economy improve 

three thing are emerged; one, it creates a better opportunity to enhance income; two, there is 

high possibility to reduce poverty and under five mortality   in the country due to improved 

welfare and lastly, the poor are demotivated to engage in criminal activities. Further, sanction 

and detection prevents the criminal to commit crime because of the fear of imprisonment as 

criminal weight the cost of apprehension over the gain in committing a crime. Therefore, 

increasing the Police presence and other security agencies serves to demotivate criminal from 

engaging in crime in the country. 

 

Conclusion 

Ethnic diversity which are heterogeneous in nature causes crime. Similarly, poverty and 



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

under five mortality increases crime in the country. But, income growth and security 

expenditure as a deterrence measure causes reduction in crime. Thus, the results obtained in 

this study informed that there is need to reduce crime and at same reduce poverty in the 

country. To do so, this study suggested the following policies. 

Adequate measures to stimulate economic growth must be pursed. Growth policy on key 

sectors that could generate viable income-employment must be the focused for the policy 

makers. This shall provide job and enhanced the income for the poor, which is capable to 

move them out of poverty.  

Promoting economic growth vide investment in health sector affords reduction in crime 

especially among the poor. This is necessary for two reasons; one, it subsidies the money 

spent by the poor on health bills, thereby complementing the income earn by the poor, and 

secondly, to aid the poor to access to good health care facilities thereby reducing under five 

mortality.  

Investment in security must be pursed. This is to provide adequate and improve security 

personnel with modern skills and equipment. This is important to detect and combat crime; to 

come up with up to date crime statistics; to enhance the image of the country; and also, to 

boost economic growth indirectly as less criminal activities in the country would stimulate 

more investment.  

This study has not been without challenges in the area of data. Data on crime are not up to 

date in the country and available ones are not easy to obtain from the Nigeria Police. This is 

also common in other countries (Wu & Wu, 2012). Similarly, time series data on poverty is 

not available but this is overcome by using a proxy for it as earlier discussed. 

Notwithstanding, the estimates in this study are robust for policy implementation. Future 

studies are required to test this link using panel data in the country. 

 

References 

Adekoya, A. F., & Abdul Razak, N. A. (2018).“Unemployment and Violence: ARDL 

 Endogeneity Approach”. EnsayosRevista de Economía, 37 (2), 155-176. 

Altheimer, I. (2008). “Social support, ethnic heterogeneity, and homicide:Across-national 

 approach”. Journal of Criminal Justice, 36, 103–114. 

Becker, G. S. (1968). “Crime and punishment: An economic approach”. Journal of Political 

 Economy, 66 (2), 169-121. 

Bourguignon, F. (1999).“Crime as a social cost of poverty and inequality: a review 

 focusing on developing countries”. DesarrolloySociedad, 44, 61-99. 

Cerro, A. M., &Meloni O. (2000). “Determinants of the Crime Rate in Argentina during the 

 ‘90s”. Estudios de Economía, 27 (2), 297-311. 

Cherry, T. L. (1999).“Unobserved heterogeneity bias when estimating the economic model 

 of crime”. Applied Economics Letters, 6, 753–757. 

Deller, S.,& Deller, M. (2012). “Spatial Heterogeneity, Social Capital, and Rural Larceny 

 and Burglary”.Rural Sociology,77(2), 225–253. 

De Soysa, I., & Noel, C. (2018). Does ethnic diversity increase violent crime? A global 



 

19 

 

analysis  of homicide rates, 1995–2013. European Journal of Criminology, 1–24. 

Ebbe, O, N. I. (1989). “Crime and Delinquency in Metropolitan Lagos: A Study of "Crime 

 and Delinquency Area" Theory”. Social Forces, 67 (3), 751-765. 

Ehrlich, I., 1973. Participation in illegitimate activities: a theoretical and empirical 

 investigation.Journal of Political Economy, 81, 521–567. 

Fajnzylber, P., Lederman, D., &Loayza, N. (2002).“Inequality and Violent Crime”.Journal

 of Law and Economics, 45 (1), 1-39. 

Hansmann, H. B., & Quigley, J.M. (1982).“Population Heterogeneity and the Sociogenesis

 of Homicide.”Social Forces, 61(1), 206-244. 

Inder, B. (1993). “Estimating long-run relationships in economics: a comparison of 

 differentapproaches”.Journal of Econometrics, 57, 53–68. 

Inyang, J. D., & Abraham, U. E. (2013).“The Social Problem of Kidnapping and its 

 Implications on the Socio-EconomicDevelopment of Nigeria: A Study of Uyo

 Metropolis”. Mediterranean Journal of Social Sciences, 4(6), 531-544. 

Levitt, S. (1997). “Using electoral cycles in police hiring to estimate the effect of police on 

 crime”.American Economic Review, 87, 270–290. 

Marshall, M. G. (2016). Major Episodes of Political Violence (MEPV) and Conflict Regions, 

 1946-2015. Center for Systemic Peace <www.systemicpeace.org>Codebook: Major 

 Episodes of Political Violence (MEPV2015) Page 1 of 17. 

Mata, F. & Bollman, R. (2007). “An Analysis of Socio-Economic Strains and Population 

 Gains: Urban and Rural Communities of Canada 1981-2001”. Canadian Studies in 

 Population, 34 (2), 191-215. 

Mehlum, H., Miguel, E., &Torvik, R. (2006).“Poverty and crime in 19th century Germany”.

 Journal of Urban Economics, 59, 370–388. 

Merton, R. K. (1938). “Social structure and anomie”.American Sociological Review, 3 (5), 

 672-682. 

Ojedokun, U.A. (2014). Contributing factors to Police homicide in Nigeria. Police Journal: 

 Theory, Practice and Principles, 87, 41–48. 

Otu, N. (2012). Balanced policing model in the Republic of Nigeria. The Police Journal, 85, 

 235-253. 

Ouimet, M. (2012).“A World of Homicides: The Effect of Economic Development, Income

 Inequality, and Excess Infant Mortality on the Homicide Rate for 165Countries in 

 2010”.  Homicide Studies, 16, 238-258.   

Pesaran, M. H.,& Shin, Y. (1997). An Autoregressive Distributed Lag Modelling Approach 

 to Cointegration Analysis. Revised paper presented at the Symposium at the 

 Centennial of  Ragnar Frisch, The Norwegian Academy of  Science and Letters, Oslo, 

 March 3-5, 1995.   

Pesaran, M. H., Shin, Y., & Smith, R. J. (2001).“Bounds testing approaches to the 

 analysis of level relationships”.Journal of Applied Econometrics, 16, 289–326. 

Poveda, A. C. (2012). Violence and economic development in Colombian cities: A dynamic 

 panel data analysis. Journal of International Development, 24, 809–827. 

Rogers, M. L., &Pridemore, W. A. (2013).The effect of poverty and social protection on 

 national homicide rates: Direct and moderating effects”. Social Science Research, 42, 

 584–595. 

Sandoval, L. E. (2018). “Socio-economics characteristics and spatial persistence of homicides 

 in Colombia, 2000-2010”.Estudios de Economía, 45 (1), 51-77. 

Saqib Amin, (2018) "The endless nexus between ethnic diversity, social exclusion and 

 institutionalquality of Pakistan".International Journal of Sociology and Social Policy, 

 https://doi.org/10.1108/IJSSP-06-2018-0108. 

Shaw, C., & McKay, H. (1942).Juvenile Delinquency and Urban Areas.University of 



Adenuga Fabian Adekoya & Nor Azam Abdul Razak 
 

 

 Chicago Press: Chicago. 

Soares, R. R. (2004). “Development, crime and punishment: accounting forthe international 

 differences in crime rates”.Journal of Development Economics, 73, 155– 184. 

Stučka, T. (2004).“The Effects of Exchange Rate Change on the Trade Balance in Croatia”.

 IMF Working Paper, WP 04/65. 

Trawick, M. W., &Howsen, R. M. (2006).“Crime and community heterogeneity: race, 

 ethnicity,and religion”. Applied Economics Letters, 13, 341–345. 

Wu, D., & Wu, Z. (2012).“Crime, inequality,and unemployment in England and Wales”.

 Applied Economics, 44 (29), 3765–3775. 

Zhang, T. (2016).“Addressing Unobserved Heterogeneity in the Relationship between Crime 

 and Consumer Confidence”.Journal of Quantitative Criminology, 32, 47–59 
 


	Page 1