This article was downloaded by: [46.198.255.

23]
On: 28 January 2014, At: 08:12
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Global Crime
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/fglc20

Is peniaphobia an incentive to crime?

a

Siti Nur Zahara Hamzah & Evan Lau

Department of Economics , Universiti Malaysia Sarawak , Kota

Samarahan, Malaysia
Published online: 19 Sep 2011.

To cite this article: Siti Nur Zahara Hamzah & Evan Lau (2011) Is peniaphobia an incentive to
crime?, Global Crime, 12:4, 312-326, DOI: 10.1080/17440572.2011.616058
To link to this article: http://dx.doi.org/10.1080/17440572.2011.616058

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
Content) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions

Global Crime
Vol. 12, No. 4, November 2011, 312326

Is peniaphobia an incentive to crime?

Siti Nur Zahara Hamzah and Evan Lau*

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Department of Economics, Universiti Malaysia Sarawak, Kota Samarahan, Malaysia

This article analysed the relationship between crime categories and unemployment rates
using a set of panel data for 14 states in Malaysia with data spanning from 1982 to 2008.
It is well documented that crime and unemployment are negatively related in Malaysia;
the same is the case for both violent and property crime. Increases in unemployment
rates cause the consumption expenditure to decrease, especially among households,
hence, causing potential earnings from illegitimate activities to drop and discouraging
a person from committing a crime. However, the significant properties of the t-statistics
indicate that it is important to consider the labour market conditions in employing
appropriate policies in fighting crime. That being said, unemployment can indirectly
explain hunger, poverty, decreasing standards of life and economic downturn.
Keywords: unemployment; crime; panel data analysis; Malaysia

Introduction
The crimeunemployment nexus is one of the most antediluvian issues in the social science
literature. However, consensus on the causal linkages between unemployment and crime
has yet to be achieved.1 Explaining crime in an economic model allows for the assumption
of rational choice theory that outweighs costs (preparation, conviction and apprehension)
and benefits (profits) before engaging in illegitimate activities.2 Basically, the economics

*Corresponding author. Email: lphevan@feb.unimas.my; ygwiex@yahoo.com

1. R. Freeman, Crime and the Labour Market, in Crime and Public Policy, ed. J.Q. Wilson (San
Francisco, CA: Institute for Contemporary Studies, 1983), 89106; S.K. Long and A.D. Witte,
Current Economic Trends: Implications for Crime and Criminal Justice, in Crime and Criminal
Justice in a Declining Economy, ed. K.N. Wright (Cambridge, MA: Oelgeschlager, Gunn, and Hain,
1981), 69143; and R. Freeman and W.M. Rodgers, Area Economic Conditions and the Labor
Market Outcomes of Young Men in the 1990s Expansion (working paper no. 7073, National Bureau
of Economic Research, 1999).
2. B. Fleisher, The Effects of Unemployment on Juvenile Delinquency, Journal of Political
Economy 71 (1963): 54355; B. Fleisher, The Effects of Income on Delinquency, American
Economic Review 56 (1966): 11837; G.S. Becker, Crime and Punishment: An Economic
Approach, Journal of Political Economy 76 (1968): 1169217; I. Ehrlich, Participation in
Illegitimate Activities: A Theoretical and Empirical Investigation, Journal of Political Economy
38 (1973): 52165; I. Ehrlich, On the Usefulness of Controlling Individuals: An Economic
Analysis of Rehabilitation, Incapacitation, and Deterrence, American Economic Review 71 (1981):
30722; I. Ehrich, Crime, Punishment, and the Market for Offenses, Journal of Economic
Perspectives 10 (1996): 4368; I. Ehrich, Crime, Punishment, and the Market for Offenses,
Journal of Economic Perspectives 10 (1996): 4368; and S.D. Levitt, Using Electoral Cycles in
Police Hiring to Estimate the Effect of Police on Crime, American Economic Review 87 (1997):
27090.
ISSN 1744-0572 print/ISSN 1744-0580 online
2011 Taylor & Francis
http://dx.doi.org/10.1080/17440572.2011.616058
http://www.tandfonline.com

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Global Crime

313

of crime aims to determine the socio-economic factors that explain criminal decisions and
behaviours, differing from the previous theoretical and sociological-based approaches to
explain crime.
Labour market conditions (unemployment rates) have received substantial attention
from economists, illustrating that high unemployment rates are related to a significant
number of crimes.3 Peniaphobia,4 as a result of unemployment, is said to motivate future
criminals to choose illegitimate activities as the only source to continue life.5
The arguments and debates on the unemployment and crime nexus began with Becker6
and Ehrlich.7 Since then, the crimeunemployment notion has developed into a positive
motivation effect8 and a negative opportunity effect9 of unemployment to crime. A positive relationship between crime and unemployment was posited by Becker10 and Brenner,11
who explain the basic idea of the positive relationship among the two. An individual
is forced to neglect his/her standard of living when he/she becomes unemployed, and
therefore, is likely to indulge in criminal activities to survive.12
Elsewhere in the literature, Cantor and Land13 justified that when people are
unemployed, the expenditure on property and luxury goods are reduced and people
have more protection to their property, thus reducing the incidence of crime as
a negative relationship of crimeunemployment.14 In addition, some studies on

3. D. Lee and S.J. Holoviak, Unemployment and Crime: An Empirical Investigation, Applied
Economics Letters 13 (2006): 80510.
4. Peniaphobia is an abnormal fear of poverty, or becoming broke. It is derived from the
Greek word penia that means poverty. Events, such as economic disasters, where people lose their
jobs or homes can also cause this phobia. A person suffering from this fear may think that their
lack of savings could cause them to lose everything they have worked to obtain. A severe case of
peniaphobia can become stressful and cause anxiety throughout the entire household.
5. H.M. Brenner, Crime in Society, in Economic Crises and Crime, ed. L. Savits and N. Johnston
(New York: Wiley, 1978a), 55572 and H.M. Brenner, Impact of Economic Indicators on Crime
Indices, in Unemployment and Crime. Hearings Before the Subcommittee on Crime of the Committee
on the Judiciary, US House of Representatives. Serial No. 47, (1978b), 2054.
6. Becker, Crime and Punishment.
7. Ehrlich, Participation in Illegitimate Activities.
8. B. Reilly and R. Witt, Crime and Unemployment in Scotland: An Econometric Analysis Using
Regional Data, Scottish Journal of Political Economy 39 (1992): 21328; K.L. Papps and R.
Winkelmann, Unemployment and Crime: New Evidence for an Old Question, New Zealand
Economic Papers 34 (2000): 5372; S. Raphael and R. Winter-Ebmer, Identifying the Effects of
Unemployment on Crime, Journal of Law and Economics 44 (2001): 25983; and K. Edmark,
Unemployment and Crime: Is There a Connection?, Scandinavian Journal of Economics 107
(2005): 35373.
9. D. Cantor and K.C. Land, Unemployment in Crime Rates in the Post World War II United
States: A Theoretical and Empirical Analysis, American Sociological Review 50 (1985): 31732;
C. Britt, Crime and Unemployment among Youths in the United States, 19581990, American
Journal of Economics and Sociology 53 (1994): 99109; and M. Melick, The Relationship Between
Unemployment and Crime, The Park Place Economist 6 (2004): 306.
10. Becker, Crime and Punishment.
11. Brenner, Crime in Society; and H.M. Brenner, Impact of Economic Indicators on Crime
Indices.
12. Ibid.
13. Cantor and Land, Unemployment in Crime Rates in the Post World War II United States.
14. D.R. Osborn, A. Trickett, and R. Elder, Area Characteristics and Regional Variates as
Determinants of Area Property Crime Levels, Journal of Quantitative Criminology 8 (1992):
26585; and R.C. Allen, Socioeconomic Conditions and Property Crime: A Comprehensive Review

Downloaded by [46.198.255.23] at 08:12 28 January 2014

314

S.N.Z. Hamzah and E. Lau

crimeunemployment found a mixture of results for the disaggregated data used in their
studies.15
Crimeunemployment exhibitions will display prolific information to viewers regarding the methodological advancement that took place in the subject matter. Economists
believe that variety in the methodological approaches employed for empirical analysis is the
reason for ambiguous results in the crimeunemployment literature.16 Correlation analysis
is among the first techniques used to analyse the crimeunemployment relationship among
economists.17
Not long after, economists argued that the correlation analyses employed in many studies were suffering from information constraints since they do not imply causation.18 As
a result, cointegration and causality19 started to gain in popularity. These methods gain
popularity after Granger and Newbold20 and Nelson and Plosser21 pointed out that most
time series data are non-stationary and can provide misleading or spurious regression.
Cointegration and error correction suggest that when two variables are non-stationary,
there exists a co-movement amongst them which could be exploited to test for the existence of equilibrium relationships within a fully dynamic specification framework. Error
corrections can then be utilised to identify the speed with which our model returns back to
equilibrium following an exogenous shock in a short-run dynamic. These methodologies

and Test of the Professional Literature, American Journal of Economics and Sociology 55 (1996):
293308.
15. Ya-Hwei Yang, Economic Crime and Business Cycles in Taiwan, Journal of the Asia Pacific
Economy 2 (1997): 388405; A.H. Baharom and A.S. Habibullah, Is Crime Cointegrated with
Income and Unemployment? A Panel Data Analysis on Selected European Countries (MPRA paper
no. 11927, 2008); and P.K. Narayan and R. Smyth, Crimes Rates, Male Youth Unemployment and
Real Income in Australia: Evidence from Granger Causality Tests, Applied Economics 36 (2004):
207995.
16. N. Dritsakis and A. Gkanas, The Effect of Socioeconomic Determinants on Crime Rates: An
Empirical Research in the Case of Greece with Cointegration Analysis, International Journal of
Economic Sciences and Applied Research 2 (2009): 5164.
17. S. Box, Recession, Crime and Punishment (London: MacMillan, 1987); and T.G. Chiricos, Rates
of Crime and Unemployment: An Analysis of Aggregate Research Evidence, Social Problems 34
(1987): 187212.
18. Narayan and Smyth, Crimes Rates, Male Youth Unemployment and Real Income in Australia.
19. G. Withers, Crime, Punishment and Deterrence in Australia: An Empirical Investigation,
Economic Record 60 (1984): 17685; C. Hale and D. Sabbagh, Testing the Relationship Between
Unemployment and Crime: A Methodological Comment and Empirical Analysis Using Time Series
Data from England and Wales, Journal of Research in Crime and Delinquency 28 (1991): 40017;
A.M.M. Masih and R. Masih, Temporal Causality and the Dynamics of Different Categories of
Crime and Their Socioeconomic Determinants: Evidence from Australia, Applied Economics 28
(1996): 1093104; R. OBrien, Measuring the Convergence/Divergence of Serious Crime Arrest
Rates for Males and Females: 19601995, Journal of Quantitative Criminology 15 (1999) 97114;
J.M. Luiz, Temporal Association, the Dynamics of Crime and Their Economic Determinants: A
Time-Series Econometric Model of South Africa, Social Indicators Research 53 (2001): 3361;
G. Saridakis, Violent Crime in the United States of America: A Time-Series Analysis Between
19602000, European Journal of Law and Economics 18 (2004): 20321; and Narayan and Smyth,
Crimes Rates, Male Youth Unemployment and Real Income in Australia.
20. C.W.J. Granger and P. Newbold, Spurious Regressions in Econometrics, Journal of
Econometrics 2 (1974): 11120.
21. C.R. Nelson and C.I. Plosser, Trends and Random Walks in Macroeconomic Time Series: Some
Evidence and Implications, Journal of Monetary Economics 10 (1982): 13962.

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Global Crime

315

also generate mixed results. Some authors found cointegration in their studies,22 some
failed to identify cointegration,23 while others proved their case with a mix of both
cointegration and no cointegration24 in their effort to relate different crime types with
socio-economic factors.
Masih and Masih25 utilised annual data from 1963 to 1990 for Australia in their
analysis of the relationship between crime rates and socio-economic- and demographicdependent variables. They used the cointegration and vector error correction model to
run their analysis and found the presence of a cointegrating vector. To continue the analysis, they applied causality and variance decompositions and concluded that dwelling
commencements (proxy for wealth) are the cause for all crime categories followed by
the urbanisation rate, except for serious assault. Gillani et al.,26 in the case of Pakistan,
concludes that there is an existence of cointegration between crime and poverty and
unemployment and inflation. Using TodaYamamoto test, they also managed to establish
a unidirectional causality from unemployment, poverty and inflation to crime from the
19752007 data. These results are inconsistent with those of Hale and Sabbagh, OBrien,
Luiz and Saridakis,27 among others, who failed to establish a cointegration relationship
between crime and unemployment. Once again, a consensus has never been achieved.
Talk on panel data analysis then became endemic in crimeunemployment literature
as one of the best ways to analyse crimeunemployment-related data. Levitt28 discussed
various prolific ways to deal with unemployment and crime data analysis, such as crosssectional and panel data utilisation, to gauge for a better reliable outcome. Crime and

22. Masih and Masih, Temporal Causality and the Dynamics of Different Categories of Crime and
Their Socioeconomic Determinants; S.Y.M. Gillani, H. Ur-Rehman, and A.R. Gill, Unemployment,
Poverty, Inflation and Crime Nexus: Cointegration and Causality Analysis of Pakistan, Pakistan
Economic and Social Review 47 (2009): 7998; and Dritsakis and Gkanas, The Effect of
Socioeconomic Determinants on Crime Rates.
23. R. OBrien, Measuring the Convergence/Divergence of Serious Crime Arrest Rates for Males
and Females: 19601995; Luiz, Temporal Association, the Dynamics of Crime and Their Economic
Determinants; and Saridakis, Violent Crime in the United States of America: A Time-Series
Analysis Between 19602000.
24. A. Scorcu and R. Cellini, Economic Activity and Crime in the Long Run: An Empirical
Investigation on Aggregate Data from Italy, 195194, International Review of Law and Economics
18 (1998): 27992; R. Witt and A. Witte, Crime, Prison and Female Labour Supply, Journal of
Quantitative Criminology 16 (2000): 6985; D. Lee, Income Inequality and Crime: Co-Integration
Analysis and Causality Tests (working paper, Grove College of Business, Shippensburg University,
2002); Narayan and Smyth, Crimes Rates, Male Youth Unemployment and Real Income in

Australia; Lee and Holoviak, Unemployment and Crime; and Y. Kustepeli and G. Onel,
Different
Categories of Crime and Their Socio-Economic Determinants in Turkey: Evidence from Vector Error
Correction Model (Turkish Economic Association, International Conference on Economics, Ankara,
Turkey, September 1113, 2006).
25. Masih and Masih, Temporal Causality and the Dynamics of Different Categories of Crime and
Their Socioeconomic Determinants.
26. Gillani et al., Unemployment, Poverty, Inflation and Crime Nexus.
27. Hale and Sabbagh, Testing the Relationship Between Unemployment and Crime; OBrien,
Measuring the Convergence/Divergence of Serious Crime Arrest Rates for Males and
Females: 19601995; Luiz, Temporal Association, the Dynamics of Crime and Their Economic
Determinants; and Saridakis, Violent Crime in the United States of America: A Time-Series
Analysis Between 19602000.
28. S.D. Levitt, Alternative Strategies for Identifying the Link Between Unemployment and Crime,
Journal of Quantitative Criminology 17 (2001): 37790.

Downloaded by [46.198.255.23] at 08:12 28 January 2014

316

S.N.Z. Hamzah and E. Lau

unemployment are data types with a tremendous degree of variability at the local level,29
which is hard to locate with national-level time series data with limited degrees of freedom.
Panel data, on the other hand, consist of annual fixed effects and state fixed effects, which
allow these to control for any shocks during a year that affect the entire country or any
differences across places that are not easily quantified. Panel data also possess a large
degree of freedom, which can control for a wide range of time-varying factors that might
be linked to crime and unemployment. Hence, it reduces the outcome of possible spurious
coefficients when national-level time series data are utilised.30 According to Levitt,31 in
crimeunemployment analysis, one should focus on the specific predictions about a range
of possible behavioural conducts through which the crimeunemployment nexus operates,
rather than merely focusing on the general linkages between them.
Because of these interesting features, studies have been conducted using states, countries, metropolitan areas, municipalities and cities in the United States.32 Interestingly,
these studies provide consistent results with each other, even in the studies that replace the
measurement of the labour market conditions with different measures.33 The consistency
of these results seems promising compared with national-level time series data, which are
much more sensitive to the particular type of estimation. A panel sample of 705 countries
over 19 years established robust results with all economic variables (education, unemployment and income) being significant for both property crime and violent crime indices,
including all individual crime rates.34 The ordinary least squares (OLS) regression also
provides expected signs, with income and education being positive, while unemployment
is negative. In particular, a 3.05% increase in unemployment will cause a 7.1% increase in
property crime and a 3.8% increase in violent crime.
Baharom and Habibullah35 found that income and unemployment were both related to
aggregated and disaggregated crime rates for their study of England. Their analysis of data,
spanning from 1993 to 2001, shows that the random effects model is more appropriate in
explaining the relationship between crime, income and unemployment after conducting a
Hausman test.
Thus, a major contribution of this study is to determine the relationship between unemployment and a range of property and violent crime in Malaysia using state-level data
from 1982 to 2008. Since its initial development, the crimeunemployment nexus has been
analysed and debated widely using developed countries data, especially for the United

29. J.Q. Wilson, Thinking About Crime (New York: Basic Books, 1983); and J. Lynch, Crime in
International Perspective, in Crime, ed. J.Q. Wilson and J. Petersilia (San Francisco, CA: ICS Press,
1995).
30. Levitt, Alternative Strategies for Identifying the Link Between Unemployment and Crime.
31. Ibid.
32. D. Lee, An Empirical Investigation of the Economic Incentives for Criminal Behavior
(Unpublished manuscript, 1993); S.D. Levitt, The Effect of Prison Population Size on Crime
Rates: Evidence from Prison Overcrowding Litigation, Quarterly Journal of Economics 111 (1996):
31951; S.D. Levitt, Using Electoral Cycles in Police Hiring to Estimate the Effect of Police
on Crime, American Economic Review 87 (1997): 27090; and Raphael and Winter-Ebmer,
Identifying the Effects of Unemployment on Crime.
33. E.D. Gould, B. Weinberg, and D. Mustard, Crime Rates and Local Labor Market Opportunities
in the United States: 19791991 (working paper no. 98-11, 1998); and S. Machin and C. Meghir,
Crime and Economic Incentives (Unpublished manuscript, 2000).
34. E.D. Gould, B. Weinberg, and D. Mustard, Crime Rates and Local Labor Market Opportunities
in the United States: 19791997, Review of Economics and Statistics 84 (2002): 4561.
35. Baharom and Habibullah, Is Crime Cointegrated with Income and Unemployment?.

Global Crime

317

Downloaded by [46.198.255.23] at 08:12 28 January 2014

States and Australia.36 Studies on Malaysia have primarily employed national-level time
series data.37 By employing panel data in this article we aim to provide a better picture of
the relationship between crime and unemployment. This will also assist policymakers in
better understanding the relationship between crime and unemployment for future policy
development that can benefit the economy.
The nature of crime in Malaysia
Increasing by leaps and bounds is the best way to describe the Malaysian crime rates in
recent years. Mass media, printed and electronic alike, have been replete with crime stories and violence everyday that trigger fear and conflict amongst the public. The situation
worsens when police statistics on crime indicate that the crime rates in Malaysia are indeed
on an increasing rate. A report by the Malaysian Quality of Life testifies that the crime
rate in Malaysia has soared high to 6.2 points in 2002 from 3.8 points in the year 1990.
Figure 1 presents data on the crime rates in Malaysia from 1982 to 2008, indicating that
property crime has been the main contributor to the index crime in the country for the
period studied.38 The graph shows three peaks around the years 1986, 1998 and 2008,
illustrating that these peaks coincide with the major economic downturns that occurred in
those same years.
Figure 1 also clearly illustrates that crime in Malaysia has been increasing over the past
20 years; this increase has become alarming in recent years. Ninety percent of all crimes
reported every year include crimes related to property. Violent crimes are growing at a
rather gradual path and contribute a small portion to the index crime in Malaysia. Breaking
the figures down by state-level data provides us with more detail, given the localised nature
of the crime data.
Table 1 illustrates that Selangor and Johor are among the top states with the highest
crime levels in Malaysia. In recent years, Johor has been profoundly criticised as a city of
sins, due to the dramatic increase in the crime rates, ranging from petty theft to robbery
with assaults. All states have faced increasing crime rates between 2000 and 2008, on
average. Perlis illustrates the lowest crime cases reported in all years occurred between
2000 and 2008. This is likely because of the small geographical location of Perlis compared

36. L.E. Cohen, M. Felson, and K.C. Land, Property Crime in the United States: A Macrodynamic
Analysis, 19471977, With Ex Ante Forecasts For the Mid-1980s, American Journal of Sociology
86 (1980): 66192; G. Withers, Crime, Punishment and Deterrence in Australia; Cantor and Land,
Unemployment in Crime Rates in the Post World War II United States; Hale and Sabbagh, Testing
the Relationship Between Unemployment and Crime; Masih and Masih, Temporal Causality and
the Dynamics of Different Categories of Crime and Their Socioeconomic Determinants; B. Reilly
and R. Witt, Crime, Deterrence and Unemployment in England and Wales: An Empirical Analysis,
Bulletin of Economic Research 48 (1996): 13759; F. Carmichael and R. Ward, Male Unemployment
and Crime in England and Wales, Economics Letters 73 (2001): 11115; and Gould et al., Crime
Rates and Local Labor Market Opportunities in the United States: 19791997.
37. C.F. Tang and H.H. Lean, Will Inflation Increase Crime Rate? New Evidence from Bounds
and Modified Wald Tests, Global Crime 8 (2007): 31123; Baharom and Habibullah, Is Crime
Cointegrated with Income and Unemployment?; A.H. Baharom and M.S. Habibullah, Crime and
Income Inequality: The Case of Malaysia, Journal of Politics and Law 2 (2009): 5570; M.S.
Habibullah and A.H. Baharom, Crime and Economic Conditions in Malaysia, International Journal
of Social Economics 36 (2009): 107181; and C.F. Tang and H.H. Lean, New Evidence from the
Misery Index in the Crime Function, Economics Letters 102 (2009): 1125.
38. In Malaysia, index crime is the term used for the total crime (property crime plus violent crime)
recorded by the Royal Malaysian Police Department.

318

S.N.Z. Hamzah and E. Lau

with other states in Peninsular Malaysia. However, the natural population increase rate in
Perlis is higher compared with states like Pulau Pinang and Perak.
The overall natural population increase rate declined over the years of 20002008 for
all states. At the same time, in the years 20052008, each state recorded an increase in the
crime rates. Pahang had the highest crime rate, a 105.8% increase in the cases of crime
reported in the years from 2005 to 2008, while Johor had the lowest increase of 13.2%. On
average, there was a 43.95% increase in the overall crime rates in Malaysia from 2005 to
2008.

Crime statistics in Malaysia, 19822008

250,000

Number of crime

150,000
Index

100,000

Violent
50,000

2008

2006

2004

2002

2000

1998

1996

1994

1992

1990

1988

1986

1984

Property

1982

Downloaded by [46.198.255.23] at 08:12 28 January 2014

200,000

Year

Figure 1. Crime statistics in Malaysia, 19822008.

Table 1. Number of index crimes reported by states, 19902008.

2000

2005

2008

States

Crime

Population

Crime

Population

Crime

Population

Johor
Kedah
Kelantan
Melaka
Negeri Sembilan
Pahang
Perak
Perlis
Pulau Pinang
Sabah
Sarawak
Selangor
Terengganu
Kuala Lumpur

25,291
7168
2331
5408
4354
3439
15,690
496
11,015
5818
7191
47,562
3281
28,086

20.3
16.8
24.2
18.7
17.1
18.8
17.1
16.3
15.4
18.4
18.7
20.0
23.2
18.5

23,504
10,184
4425
4293
5342
3517
10,768
517
11,512
5805
8763
44,521
4524
19,690

14.0
13.2
16.8
13.0
12.4
13.6
10.2
11.1
10.3
14.2
14.8
16.5
16.7
12.4

26, 624
13,107
7749
5606
7668
7239
12,581
808
16,651
8498
12,929
57,752
5995
28,438

12.8
12.4
15.3
12.1
11.3
12.2
8.9
10.7
9.5
13.7
13.5
16.2
16.3
11.5

Notes: Crime refers to number of index crimes and is expressed as the annual incidence per 100,000
residents. Population refers to state natural population increase rate (%) obtained from Department of
Statistics.
Source: Data obtained from Royal Malaysian Police Department (RMPD).

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Global Crime

319

Sidhu,39 while predicting future crime levels and trends in Malaysia, took into consideration macroeconomic factors, such as economic strength, unemployment and consumer
spending level, to explain the increase of crime rates in the country. He projected that the
crime rates will rise at an increasing rate of 2.6% per annum amounting to 200,000 cases
of crime in 2015. To our surprise, the prediction was successfully achieved two years in
advance (2007), with 209,559 crime cases recorded in Malaysia.40
Links between unemployment and crime were recorded in several studies for both
local and international cases. Salleh,41 in his research study, found a significant correlation between unemployment and property-related crime in police districts. The effects of
unemployment and property crime were detected during the economic downturn in 1997.
Seventy-five percent of the offenders of crimes chose illegitimate activities due to unemployment and difficulties in finding a better job, while 43% did so to continue their drug
supplement.42 Muhammad43 further verified the findings with the monthly income statistics of the convicted prisoners. Table 2 illustrates that 95% of the convicted offenders
earnings were below the Malaysian real per capita income of RM 13,708 (equals RM 1142
per month). There was a strong correlation between the economic conditions and crime
rates, which was also demonstrated in Sidhu44 during the periods of 1986 and 1997, two
economic downturns which were accompanied by an influx of criminal activities.
As major statistical indices in the Criminal Justice system, increasing crime level portrays a negative impression towards the police force and the law and enforcement strategy.
All states in Malaysia are obligated to the same law status and securities enacted through
the parliament as Malaysia practises centralised governance. All 14 states adhere to the
Securities Commission Act of 1993 and the Securities Commission Act (Amendment) of
1997 and 2007 as the guide for law and enforcement strategies. The Royal Malaysian Police
Table 2. Convicted prisoners by monthly income, 2004.
Monthly income (RM)
Unemployed
Below 250
250500
500750
7501,000
10001500
15003000
Total

Total prisoners

Percentage (%)

8543
4027
11,880
14,778
7573
1696
746
49,243

17.3
8.2
24.1
30
15.5
3.4
1.5
100

Source: Statistics from Prisons Department, Malaysia.

39. A. Sidhu, Crime Levels and Trends in the Next Decade, Journal of the Kuala Lumpur Royal
Malaysia Police College 5 (2006): 113.
40. Ibid., analysed data from 1980 to 2004 in his research study.
41. M.A.R. Salleh, Pengangguran Merupakan Faktor Utama Menyebabkan Peningkatan Jenayah
Harta Benda di Negeri Pahang (Diploma Sains Kepolisian, project paper, Universiti Kebangsaan
Malaysia/RMP, 1987).
42. M.S. Muhamad, Penggaguran dan Jenayah: Satu Kajian Keatas Banduan Di Penjara Sungai
Buluh (Diploma Sains Siasatan, project paper, Universiti Malaya/RMP, 2002).
43. Ibid.
44. A. Sidhu, The Rise of Crime in Malaysia: An Academic and Statistical Analysis, Journal of the
Kuala Lumpur Royal Malaysia Police College 4 (2005): 128.

320

S.N.Z. Hamzah and E. Lau

Department (RMPD) is responsible for all recorded criminal case reporting, while convictions and cases are handed over to the court for punishment purposes. In the Malaysian
case, the police department and law constitution are a separate institution without any direct
linkages. The Yang di-Pertuan Agong (King) is the highest organisational head who has all
the power of punishment in Malaysia.

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Methodology and data description

Data description
Time series data from 1982 to 2008 for 14 cross sections of states in Malaysia are employed
in the analysis.45 Crime data46 are divided into disaggregated and aggregated crime categories, according to the RMPD characterisation. All crime data were provided by the
RMPD, while unemployment47 data for states were extracted from the Malaysia Statistics
Department. Following the research recommendations of Cherry,48 models for this study
were transformed into log-linear form to certify that the estimated coefficients are elastic.
Panel data analysis
The basic issues in the panel data analysis are the modelling of the error term in the
regression model. The starting point of the analysis is the following panel model:
crime = + ueit + it ,

i = 1, 2, . . . , N; t = 1, 2, . . . , T

(1)

where i denotes the states (cross section) and t denotes the time (time series). The crime
function refers to crime rate of states i at time t. is a scalar that captures the unobserved
state effects; ueit is the itth observation on unemployment; and is ue 1. The model
employs a common one-way error component for the disturbances, with it = i + t
where i denotes the unobservable individual-specific effect and t denotes the remainder
disturbances. Estimation using standard cross-sectional techniques assumes homogeneity
in the unit of observation (in our case, states) by restricting the to equal each other for all
is. This assumption, however, yields inconsistent and biased estimates when state effects
are not homogenous.49 The pooled ordinary least squares model also suffers from a complicated error process, such as heteroskedasticity across panel units and serial correlation
within panel units.50

45. Refer to the Appendix for clear definitions of crime data and states employed in this study.
46. Note that these figures are merely crimes reported to the RMPD; it is not an accurate reflection of
the total crime situation in Malaysia, since there are many crime cases that are unreported and others
that go unrecorded. Nevertheless, it can still become part of a real crime database and can be utilised
to reflect the degree of the crime situation in Malaysia.
47. In this study, unemployment is the only independent variable, since the state-level data for other
socio-economic variables are either unavailable or were only recorded for a very short time span.
This study also intends to explain the basic theory of crimeunemployment in the context of Malaysia
using state-level data. By this token, we resort to this rationalisation and adopt an appropriate panel
econometrics analysis to bring about the conclusion.
48. T.L. Cherry, Unobserved Heterogeneity Bias When Estimating the Economic Model of Crime,
Applied Economic Letters 6 (1999): 75357.
49. Cherry, Unobserved Heterogeneity Bias When Estimating the Economic Model of Crime.
50. Baltagi, Econometric Analysis of Panel Data, 3rd ed. (Chichester: Wiley & Sons, 2005).

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Global Crime

321

Thus, the appropriate effects model (random or fixed) is recommended for analysing
panel data, since it controls for heterogeneity. Hence, it provides consistent and efficient
estimates. The heterogeneity is restricted to the intercept terms of the relationship by
imposing restrictions on the overly restrictive OLS model, allowing only the constant to differ over i. The restrictions can be imposed to the slope coefficients that vary to be constant
over both units and time and allow for an intercept varying by unit or time.51 The random
effects model assumes i is uncorrelated with the regressor. The fixed effects model, on
the other hand, assumes i to be correlated with the regressor. The hot issues for now in
the panel literature are which model to use for a study. There are three competing formulations in determining how to model the error term between the pooled model and the effects
model. In this study, we employed the BreuschPagan Lagrangian multiplier (LM) test52
to distinguish between the pooled model, the random effects model and the Hausman test53
and to distinguish between the random effects model and the fixed effects model. Basically,
the competing argument for the two tests is which is the most appropriate effects model.
The BreuschPagan54 LM test uses the pooled OLS residuals to calculate the LM value.
The BreuschPagan hypothesis and model are as follows:
H0 : 2 = 0
HA : 2 > 0
where Yit = Xit + it and it = it + it .


N
i=1


T

2

it
t=1
NT
N T 2
2(T 1)
it
i1
t=1

When the calculated value exceeds the tabulated chi-squared value, reject H 0.
Consequently, the random effects model is more appropriate. In other words, there are
country-specific effects in the data. Denoting the variancecovariance matrix of fe and
re by Vfe and Vre , respectively, and letting k be the dimension of :
Hausman = (fe re ) [Vfe Vre ]1 (fe re )
The Hausman test investigates against the following hypothesis:

51. C.F. Baum, An Introduction to Modern Econometric Using STATA (College Station, TX: A Stata
Press Publication, 2006).
52. R. Witt, A. Clarke, and N. Fielding, Crime, Earnings Inequality and Unemployment in England
and Wales, Applied Economics Letters 5 (1998): 2657; M.J.C. Mendona, P.R.A. Loureiro, and
A. Sachsida, Criminalidade e Desigualdade Social no Brasil, Texto para Discusso 967 (2003):
120; and V. Daniele and U. Marani, Criminalit E Investimenti Esteri. Unanalisi Per Le Province
Italiane, Rivista Economica del Mezzogiorno 1 (2008): 189218.
53. Mendonca et al., Criminalidade e Desigualdade Social No Brasil; Daniele and Marani,
Criminalit E Investimenti Esteri. Unanalisi Per Le Province Italiane.
54. T.S. Breusch and A.R. Pagan, The Lagrange Multiplier Test and Its Applications to Model
Specification in Econometrics, Review of Economic Studies 47 (1980): 23953.

322

S.N.Z. Hamzah and E. Lau

H0 : cov(i , xit ) = 0
HA : cov(i , xit ) = 0

where Y it = Xit + it , and it = it + it .

H 0 will be rejected when the p-value is less than 0.05 and the fixed effects estimator will
be preferred.

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Results and discussion

Six broad categories of disaggregated crime, along with aggregated narrower crime categories, are the focus of this study. From the descriptive statistics in Table 3, it is obvious
that theft (10.46), housebreaking (9.42) and robbery (9.05) are among the criminal activities with a higher occurrence. Murder has the lowest crime incidence in Malaysia, with
a minimum of a zero occurrence from its mean in the years investigated. There is also a
clear considerable variation in the unemployment rates between states with the minimum
and maximum value varying between 0.70 and 12.30 (Table 3). The crime variables also
exhibit reasonable variations among the states and possess a high value of SDs.
The results in Table 4 are distributed into three columns: a simple OLS regression
(column 1), a random effects model (column 2) and a fixed effects model (column 3). The
results for all three models are consistent. From Table 4, it is evident that all coefficients in
the various crime categories are negative and statistically significant at the 5% significance
level in relation to the unemployment rates in all three models, except for the case of murder
under the pooled OLS estimation. As is confirmed throughout the literature, violent crime
does not possess any underlying trends; therefore, economic objectives must be the reason
for the insignificant results for murder. Only property crime possessed an underlying direct
pecuniary gain related to the economy.55
The analysis further determined the most appropriate model for this study by conducting the BreuschPagan LM test and the Hausman test. The results for the tests are provided
Table 3. Descriptive statistics.
Crime categories

Mean

Median

Maximum

Minimum

SD

Mdr
Rbr
Rpe
VCH
Hbrk
Tft
Ppty
Vlnt
Ttl
UE

2.93
5.47
4.20
5.21
7.17
8.02
8.40
6.36
8.53
4.09

5.47
5.43
4.14
5.34
7.22
8.14
8.44
6.34
8.55
3.60

5.22
9.05
8.78
7.48
9.42
10.46
10.76
9.37
10.98
12.30

0.00
1.61
1.39
2.20
4.53
5.28
5.74
3.09
5.88
0.70

1.18
1.62
1.09
0.97
0.95
1.11
1.05
1.25
1.06
1.99

Notes: All crime categories are expressed as the annual incidence per 100,000 residents. Our complete
panel consists of 378 observations for 14 states during the period 19822008.

55. A. Imrohoro
glu, A. Merlo, and P. Rupert, Understanding the Determinants of Crime, Journal
of Economics and Finance 30 (2006): 27084.

Global Crime
Table 4.

323

Estimation results of pooled, random effects and fixed effects models.

Variables
Mdr
Rbr
Rpe
VCH
Hbreak
Tft
Ppty
Vlnt
Ttl

OLS

RE

FE

0.027 (0.908)
0.241 (6.051)
0.156 (5.755)
0.130 (5.351)
0.113 (4.701)
0.145 (5.236)
0.133 (5.082)
0.185 (5.968)
0.137 (5.142)

0.037 (2.802)
0.079 (4.359)
0.178 (7.434)
0.109 (8.638)
0.069 (6.225)
0.091 (6.232)
0.083 (6.549)
0.100 (6.817)
0.083 (6.540)

0.037 (2.804)
0.077 (4.272)
0.179 (7.431)
0.109 (8.607)
0.069 (6.186)
0.090 (6.180)
0.082 (6.504)
0.098 (6.756)
0.082 (6.494)

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Note: Figures in the parentheses are the t-statistics.

Statistically significant at 5% significant level. RE, random effects model; FE, fixed effects model.

Table 5.

Results for BreuschPagan Lagrangian multiplier (LM) test and Hausman test.

Crime categories
Mdr
Rbr
Rpe
VCH
Hbrk
Tft
Ppty
Vlnt
Ttl

Unemployment

BreuschPagan
LM test

Hausman test

R2

Conclusion

0.0370 (0.0132)
0.0791 (0.0181)
0.1778 (0.0239)
0.1089 (0.0126)
0.0691 (0.0111)
0.0907 (0.0146)
0.0827 (0.0126)
0.0990 (0.0145)
0.0826 (0.2718)

3651.86
3254.90
858.93
3135.78
3428.91
3061.95
3315.21
3317.07
3350.69

0.01 (0.90)
2.75 (0.09)
0.19 (0.66)
0.11 (0.74)
0.48 (0.48)
0.61 (0.44)
0.56 (0.46)
1.18 (0.28)
0.62 (0.43)

0.02
0.05
0.13
0.17
0.09
0.09
0.10
0.11
0.10

RE
RE
RE
RE
RE
RE
RE
RE
RE

Notes: Figures in parentheses are the SEs for the unemployment and probability for the Hausman test. Conclusion
is the most appropriate model chosen after the BreuschPagan LM test and Hausman test were conducted. RE
stands for random effects model.
Statistically significant at 1% significant level.

in Table 5. The BreuschPagan LM test56 illustrates that we should reject the null hypothesis of the simple pooling specification; hence, the pooled OLS is biased and inconsistent.
This implies that we should use panel data estimators, rather than an OLS estimator.57
Including the results for the OLS estimation is instructive, which makes it clear why we
test other econometric specifications.
The Hausman test58 was then employed; results indicate that the null hypothesis of individual effect correlated with some explanatory variables can be successfully rejected. This
implies that estimating the model by using random effects generates consistent estimators,
rather than adopting a model with fixed effects.

56. Breusch and Pagan, The Lagrange Multiplier Test and Its Applications to Model Specification
in Econometrics.
57. J.M. Wooldridge, Cluster-Sample Methods in Applied Econometrics, American Economic
Review 93 (2003): 1338; Baltagi, Econometric Analysis of Panel Data.
58. J.A. Hausman, Specification Test in Econometrics, Econometrica 46 (1978): 125171; and J.A.
Hausman and W. Taylor, Panel Data and Unobservable Individual Effects, Econometrica 49 (1981):
137798.

Downloaded by [46.198.255.23] at 08:12 28 January 2014

324

S.N.Z. Hamzah and E. Lau

From the results, it is clear that all models depicted a consistent result. Unemployment
is negatively related to the various crime categories in Malaysia. The coefficients of unemployment are negative and statistically significant at the 1% significance level. The random
effects model was found to be the most appropriate model for all the crime categories
tested in this experiment. Baharom and Habibullah59 also found that the random effects
model was preferable in their study of European countries, except for the drug trafficking subcategories of crime. This is also the case, since the objective of the study is to
determine support for the population from a small sample, since crime rates are usually
underestimated because many crime activities go unreported. Baltagi60 explained the population in the random effects model from Haavelmos61 view that the population consists
not of individuals, in general, but of an infinity of decisions that each individual might
make.
The R-squared results indicate that the strength of the trend of the negative relationship
is low due to the small R-squared value. The magnitudes of the relationships are also small,
ranging from a 1% to a 17% effect of unemployment for every 1% change in the respective
crime categories. This is well documented in Freemans62 book, in which he pointed out
that Unemployment is related to crime, but if your prior was that the relationship was overwhelming, you were wrong. Joblessness is not the overwhelming determinant of crime that
many analysts and the public a priori expected it to be. The indications of the relationship
are still important to consider in policymaking, since the relationships still exist.
Conclusions
The results suggest that all crime categories are negatively related to unemployment rates
in Malaysia. This result is the same for both violent and property crimes. Following Cantor
and Land and Osborn et al.,63 it can be concluded that when unemployment increases,
the guardianship against property also increases; hence, reducing crime incidents. It is
also well documented that an increase in unemployment rates causes the consumption
expenditure to decrease, especially among households, hence causing potential earnings
from illegitimate activities to drop and discouraging a person from committing a crime.
Although Figure 1 illustrates a simple relationship between the crime and unemployment
rate as positive, this result does not concur with the outcome of our estimation. This may
be due to many other socio-economic factors that can also influence crime directly, or
indirectly, but not captured in the estimation. It is suggested that future research be conducted, including other socio-economic variables, such as education and wages when data
compilation improves in Malaysia.
The significant result illustrates that it is important to consider the labour market conditions in employing appropriate policies in fighting crime. The security problem was one
of the issues potential investors posed to the government when they planned multi-billion
ringgit development strategies in Malaysia. Like it or not, investors will never risk their

59. Baharom and Habibullah, Is Crime Cointegrated with Income and Unemployment?.
60. Baltagi, Econometric Analysis of Panel Data.
61. T. Haavelmo, The Probability Approach in Econometrics, Econometrica 12 (1994): 1118,
Supplement.
62. R. Freeman, The Economics of Crime, in Handbook of Labor Economics, ed. O. Ashenfelter
and D. Card (Amsterdam: Elsevier, 1999): 325971.
63. Cantor and Land, Unemployment in Crime Rates in the Post World War II United States; Osborn
et al., Area Characteristics and Regional Variates as Determinants of Area Property Crime Levels.

Global Crime

325

lives for a project if they do not feel safe. Increasing crime rates in Malaysia implies
that there is little security. Hence, foreign investment is pulling out. This will affect the
economy in the future. A negative relationship does not indicate that unemployment can
reduce crime, because unemployment is bad for the economy and one cannot sacrifice
the economy to reduce crime. Instead, it is suggested that the law and enforcement strategies are strengthened and public awareness on crime prevention increased to help decrease
crime rates, even when less guardianship is offered to properties. Since only unemployment
and various crime categories are used in this study, the results have to be interpreted with
caution. There are many other variables that can be included in future analysis to provide
more detailed results such as gross domestic product growth rate, income distribution and
race to name a few.

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Acknowledgements
The authors are grateful to two anonymous referees and the editor of this journal, Professor Carlo
Morselli for helpful comments and suggestions that have substantially improved the quality of
the present version. We are thankful for the comments and suggestions of the participants at the
International Borneo Business Conference (IBBC), Miri, Sarawak, 1315 December 2010. All
remaining flaws are the responsibility of the authors.

Notes on contributors
Siti Nur Zahara Hamzah is a master student in the Faculty of Economics and Business of the
Universiti Malaysia Sarawak.
Evan Lau is currently a senior lecturer at the Universiti Malaysia Sarawak.

Appendix
Disaggregated and aggregated crime categories.
Abbreviation
Disaggregated crime categories
Mdr
Rbr
Rpe
VCH
Hbreak
Tft
Aggregated crime categories
Ppty
Vlnt
Ttl

Description
Murder
Robbery
Rape
Voluntarily causing hurt
Housebreaking
Theft
Property crime
Violent crime
Total crime (index crime)

Notes: All crime categories are expressed as the annual incidence per 100,000 populations.
Disaggregated crimes are broader definition of crime data collected by types of felony.
Aggregated crime refers to crime data compiled according to relative categories of crime. All
data are provided by Royal Malaysian Police Department (RMPD).

326

S.N.Z. Hamzah and E. Lau

Crime definition.
Crime categories
Property crime

Violent crime

Downloaded by [46.198.255.23] at 08:12 28 January 2014

Index crime

Definition
Property crime includes those offences involving the loss of property
during which there is no use of violence by the perpetrators. There
are seven types of crimes in this category and they include
housebreaking and theft by day, housebreaking and theft by night,
theft of lorries and van, theft of motor cars, theft of motorcycles
and scooters, theft of bicycles and lastly other forms of theft
Violent crime which generally includes crimes of violence that are
sufficiently regular and significant in occurrence of which there are
eight in all. These include murder, attempted murder, gang robbery
with firearm, gang robbery without firearm, robbery with firearm,
robbery without firearm, rape and lastly voluntarily causing hurt
Crimes that are reported with sufficient regularity and with sufficient
significance to be meaningful as an index to the crime situation

Note: All the definitions are obtained from RMPD according to their measures of crime.

List of states employed in the analysis

Johor
Kedah
Kelantan
Melaka
Negeri Sembilan
Pahang
Perak
Perlis
Pulau Pinang
Sabah
Sarawak
Selangor
Terengganu
Kuala Lumpur
Notes: All states in Malaysia are included in this study.
Data for Sabah include Labuan Federal Territory.