Economics of Cybercrime: The Role of Broadband

and Socioeconomic Status

Jiyong Park
College of Business, KAIST
85 Hoegiro Dongdaemoon-gu, Seoul, Korea
E-mail: [email protected]

Daegon Cho
College of Business, KAIST
85 Hoegiro Dongdaemoon-gu, Seoul, Korea
E-mail: [email protected] (corresponding author)

Jae Kyu Lee

College of Business, KAIST
85 Hoegiro Dongdaemoon-gu, Seoul, Korea
E-mail: [email protected]

Byungtae Lee
College of Business, KAIST
85 Hoegiro Dongdaemoon-gu, Seoul, Korea
E-mail: [email protected]

Acknowledgement: This research is funded by the KAIST Bright Internet Research Center,
EEWS Research Center (Grant for Climate Change Global Hub Research), and the Korean
Ministry of Science, ICT, and Future Planning through the Graduate School of Green Growth at
KAIST College of Business.

Electronic copy available at: https://ssrn.com/abstract=2896059

 Economics of Cybercrime: The Role of Broadband

and Socioeconomic Status

Abstract
Under what conditions is the Internet more likely to be used maliciously for criminal activity?
This study examines the conditions under which the Internet is associated with cybercriminal
offenses. Using comprehensive state-level data in the United States during 2004-2010, our
findings show that there is no clear empirical evidence that the Internet penetration rate is related
to the number of Internet crime perpetrators; however, cybercriminal activities are contingent
upon socioeconomic factors and connection speed. Specifically, a higher income, more education,
a lower poverty rate, a lower unemployment rate, and a lower inequality are likely to make the
Internet penetration be more positively related with cybercrime perpetrators, which are indeed
different from the conditions of terrestrial crime in the real world. In addition, broadband
connections are significantly and positively associated with Internet crime perpetrators, though
narrowband connections are not. Taken together, cybercrime requires more than just a skilled
perpetrator, and it requires an infrastructure to facilitate profiteering from the act. A relevant
discussion is provided.
Keywords: Internet crime, cybercrime, Internet penetration, broadband, socioeconomic status,
Bright Internet
JEL classification: K42

Electronic copy available at: https://ssrn.com/abstract=2896059

Introduction

Crime has been one of the most serious societal challenges, and civilizations have regulated

malicious social conduct that threatens others property, health, safety, and moral welfare. The

Internet can provide an additional space to commit crimes with a possibly low degree of

detection due to its anonymity. Crimes have increasingly occurred in cyberspace as Internet use

grows rapidly so-called Internet crime or cybercrime. 1 Figure 1 presents the numbers of

complainants and perpetrators of cybercrimes reported to the Internet Crime Complaint Center

(IC3)2 in the United States; the information indicates that the number of Internet crimes had an

increasing trend over the period 2000-2014. According to the IC3, Internet crime yielded losses

of 800,492,073 USD by 269,422 complaints in 2014. However, considering the fact that these

were reported voluntarily by individuals, IC3 (2014) highlighted that the IC3 estimates less than

10 percent of victims file directly through www.ic3.gov. Furthermore, most cybercrimes are

likely to lead to additional indirect losses e.g., loss of trust, missed business opportunities and

defense costs e.g., security products and services (Anderson et al. 2013).

[Figure 1 about here]

The large variation of crime rates across time and space is one of the oldest puzzles in

economics and sociology (Glaeser et al., 1996). A large stream of literature has attempted to

disentangle the determinants of general crimes i.e., demographic and socioeconomic factors

(e.g., Chan et al., 2016; Glaeser and Sacerdote, 1999; Kelly, 2000; Sampson et al., 1997).

1
Since electronic networks are mainly connected via the Internet, the term, Internet crime, emphasizes where cybercrime occurs;
hereafter we use the two terms interchangeably.
2
IC3 (http://www.ic3.gov/) was founded as a joint effort between the National White Collar Crime Center and the U.S. Federal
Bureau of Investigation. IC3 provides the public with a reliable and convenient reporting mechanism to submit Internet crime
information and refers complaints to law enforcement agencies on behalf of the filing individuals.

Electronic copy available at: https://ssrn.com/abstract=2896059

Similarly, Broadhurst et al. (2013) suggest that cybercrime can be considered as a socio-

technological phenomenon, by highlighting the necessity to understand the characteristics of

people conducting malicious behaviors on the Internet. While there are substantial numbers of

studies on either terrestrial crimes or Internet crimes, we have a limited understanding of how

offline socioeconomic status and infrastructure impact the malicious use of the Internet for

criminal activities. Thus, to bridge this gap in the previous literature, the current paper aims to

investigate the role of socioeconomic status and broadband connections in cybercriminal

offenses by addressing the research question: Under what conditions is the Internet more likely

to be used maliciously for criminal activity?

While prior literature documents the effects of socioeconomic status on general terrestrial

crimes including violent crime and property crime, the socioeconomic factors are not necessary

to have same influences on cybercrime. Notably, virtual space with its unique social environment

differentiates cybercrime from traditional terrestrial crime, though less attention has been paid to

the relationship among socioeconomic status, Internet penetration (including broadband), and

cybercrime offenders. In this study, we aim at extending the economics of crime, originating

from Becker (1968), into the context of cybercrime in virtual space, and comparing it with

terrestrial crimes in terms of the role of socioeconomic status. Specifically, the differing effects

of socioeconomic status on cybercrime and terrestrial crime may stem from the inherent features

of the Internet and information goods, skill premium, and initial costs to commit cybercrime.

Using comprehensive state-level U.S. data on cybercrimes and time-varying attributes

during 2004-2010, we empirically examine which factors are associated with the occurrence of

Internet crimes. First of all, our results confirm that Internet crime targets are distributed more

equally across the country, whereas the origins of cybercrime perpetrators are more concentrated,

4
compared with terrestrial violent crimes. This finding implies that ICTs may serve as a force

multiplier, enabling individuals with minimal resources to generate potentially large negative

effects on widely dispersed people on the web.

For the main empirical analyses, we control extensively for state-level demographic and

socioeconomic factors and estimate specifications that include state effects, year effects, and

state-specific time trends. According to our findings, there is no clear empirical evidence that the

Internet penetration rate is directly related to the number of cybercrime perpetrators; however,

cybercriminal activities are contingent upon socioeconomic factors and connection speed.

Specifically, socioeconomic factors have differential moderating effects with Internet penetration

on the growth of cybercrime perpetrators, rather than having direct effects. Our findings suggest

that a higher income, more education, a lower poverty rate, a lower unemployment rate, and a

lower inequality are likely to make the Internet penetration be more positively related with

cybercrime perpetrators, which indeed differ from the conditions of terrestrial crimes. In addition,

contrary to narrowband connections, broadband connections are significantly and positively

associated with Internet crime perpetrators, highlighting the dark side of advanced information

and communication technologies (ICTs). Taken together, cybercrime requires more than just a

skilled perpetrator, and it requires an infrastructure to facilitate profiteering from the act.

This paper contributes to the literature in two directions. First, we show the distinct

geographical distribution of perpetrators and victims of Internet crime, compared with terrestrial

crimes. We provide comparisons between cybercrime and terrestrial crimes, which have not been

extensively explored in previous studies. Second, we find the conditions in which Internet use is

significantly associated with cybercrime activities, which can provide implications for

prescriptively preventing potential Internet crime. Notably, we highlight the moderating roles of

5
socioeconomic status and broadband in the malicious use of the Internet for criminal activities.

Our findings call for the extension of the economic theory of crime into the context of

cybercrime, which is quite different from terrestrial crimes.

Literature Review

With the wide and rapid prevalence of Internet use, online and offline spaces closely interact

more often with each other. As a result, a growing number of studies focus on examining the

positive or negative societal impacts caused by ICTs and widespread online access (Ayyagari et

al., 2011; Bhuller et al., 2013; Chan et al., 2016; Chan and Ghose, 2014; Ganju et al., 2016;

Greenwood and Wattal, 2016; Lee, 2015). Focusing on the impact on crimes, Chan et al. (2016)

provide empirical evidence of positive spillover effects of Internet use on racial hate crimes in

the United States. Bhuller et al. (2013) show that Internet use is associated with a substantial

increase in sex crimes and argue that this phenomenon is mainly driven by a direct effect on sex

crime propensity, possibly due to an increased consumption of pornography. Our study is in line

with this stream of studies, though we closely look at the relationship between Internet

penetration and cybercrime perpetrators. Further, we examine the moderating roles of

socioeconomic status in the malicious use of the Internet, which has been largely neglected by

the literature.

Cybercrime is defined as computer-mediated activities which are either illegal or

considered illicit by certain parties and which can be conducted through global electronic

networks (Thomas and Loader, 2000). Note that cybercrime (or Internet crime) is different from

cyberattacks that can be defined as computer-to-computer attacks undermining the

confidentiality, integrity, and/or availability of computers and/or the information they hold

6
(Kim et al., 2012). In other words, Internet crime can be regarded as a more comprehensive

concept, including computer-assisted crimes (e.g., credit card fraud) and computer-focused

crimes (e.g., hacking).

A stream of literature (e.g., John et al., 2009; Lakhina et al., 2005) and cybersecurity

industry research (e.g., Symantec, 2016) has emerged to deal with the attack origins of

cybercrime. For instance, John et al. (2009) implement Botlab, a real-time botnet monitoring

platform, and find that six botnets are responsible for 79% of spam messages. While the

technical side of cybercrime origins has been studied from many perspectives, the human side

the perpetrators who commit cybercriminal activities has received relatively little attention

from scholars, presumably because it is not easy to identify cybercrime offenders within an

anonymous environment. However, Broadhurst et al. (2013) highlight the notion that cybercrime

can be considered as a socio-technological phenomenon, and requires an understanding of the

characteristics of the people committing these crimes. In this respect, our paper focuses on the

unexplored relationship among socioeconomic factors, Internet penetration, and Internet crime

perpetrators.

The large variation of crime rates across time and space is one of the oldest puzzles in

sociology and criminology (Glaeser et al., 1996), and prior literature suggests that crime rates are

influenced by demographic and socioeconomic factors including population, age, income,

educational attainment, racial distribution, number of people in poverty, and unemployment

(Chan et al., 2016; Chiu and Madden, 1998; Edmark, 2005; Glaeser and Sacerdote, 1999; Kelly,

2000; Lochner and Moretti, 2004; Patterson, 1991; Sampson et al., 1997). However, less

attention has been paid to the determinants of Internet crime, which is the focal topic of this

paper.

7
 Kigerl (2011) examines the determinants of high cybercrime countries; in doing so, his

study is the closest to ours. Using country-level cross-sectional data, Kigerl (2011) suggests that

the GDP and Internet penetration rate are positively associated with cybercrime activity, while

the unemployment rate positively mediates the effect of Internet users on cybercrime activity.

However, our study is indeed different from Kigerl's (2011) in several directions. First, while

observed and unobserved confounding effects may not be captured in a country-level cross-

sectional analysis, our state-level panel analyses within a single country enable us to better

examine the impact of important state-level socioeconomic attributes on Internet crimes. Second,

while Kigerl (2011) accounts for spam and phishing-related activity in his analysis, we use a

more comprehensive measure of Internet crime. Third, it is worth emphasizing that we account

for spatial separation by differentiating cybercrime perpetrators from victims, which has not been

properly addressed in previous studies.

Theoretical Background

The recent discussion emphasizes the salience of Internet crime. Many researchers argue that

virtual space with its unique social environment differentiates cybercrime from traditional

terrestrial crime, and the emergence of a new breed of crime requires the development of a new

and distinctive criminological approach (Capeller, 2001; Yar, 2005). However, less attention has

been paid to the economics of cybercrime in virtual space. According to the economic theory of

crime, originating from Becker (1968), potential offenders attempt to maximize their expected

utility, which is echoed by the expected return associated with a given target, costs to commit

crime, and a perceived risk (Cornish and Clarke, 2014). In this section, we discuss how

cybercrime can be differentiated, in terms of the role of socioeconomic status, from offline

8
terrestrial crimes, such as violent crime and property crime, through the lens of the economic

theory of crime.

First, the inherent features of the electronic communication medium can confer a benefit on

malicious conduct or criminal activities in virtual space. Being unconstrained by the normal

barriers of physical distance, the Internet can act as a force-multiplier by helping individuals

with minimal resources to generate potentially large negative effects on widely dispersed people

via electronic networks (Yar, 2013). The high-speed broadband connections provide even better

conditions to commit cybercrimes, with more returns to cybercrime being expected.

Furthermore, the Internet has significantly reduced the marginal cost of producing and

distributing digital information goods. The advent of digital information goods with very low

marginal cost coincides with the emergence of economy of scale and large-scale bundling

strategies in Internet businesses (Bakos and Brynjolfsson, 2000, 1999; Brynjolfsson et al., 2010),

leading to winner-take-all markets (Arthur, 1989). In a similar vein, malware or other tools for

cybercriminal activities are a kind of digital information goods with low marginal cost. Thus, the

more people offenders can infect, the bigger the return, leading to more concentrated distribution

of cybercrime offenses than that of terrestrial crimes (Clayton et al., 2015). On the other hand,

cybercrime victimization would be democratized, that is, the geographic distribution of

cybercrime victims can be more evenly dispersed across locations than that of terrestrial crime

and cybercrime offenders.

Second, virtual space can make neighborhood-based factors to motivate criminal behavior

less important. For instatance, strain theory argues that, when individuals view the relative

success of their acquaintances, people in the lower classes are likely to feel frustration (Merton,

1938), and they are more likely to commit crime in response. Munyo and Rossi (2013) insist that

9
frustration is followed by a spike in violent crime. In addition, Kelly (2000) suggests that areas

of high inequality place poor individuals who have low returns from market activity next to high-

income individuals who have goods worth taking, thereby increasing the returns to time allocated

to criminal activity. However, the income inequality or relative poverty compared to their

neighborhood may not influence the cybercriminal activities which target offending unspecified

people online beyond their physical boundary.

Third, skill premium to ICTs can change the dynamics of criminal behavior on the Internet.

A large stream of literature has covered skill-biased technological change led by ICTs, which is a

shift in technologies that favor skilled over unskilled labor by increasing its relative productivity

and thus its relative demand (Acemoglu, 1998; Autor et al., 1998; Bresnahan et al., 2002;

Machin and Van Reenen, 1998; Michaels et al., 2014). In the context of criminal behavior,

Lochner (2004) develops a model of crime and human capital investment, which explicitly

models decisions to work, to commit crime, and to make costly investments in human capital.

According to the model, the opportunity costs of crime should generally rise with education; thus

education should be negatively correlated with unskilled crimes, such as violent crimes.

However, Lochner also suggests that skills may increase the returns to crime, especially for

white collar crimes, such as fraud and embezzlement, possibly leading to a positive correlation

between education and crime. Given that Internet crime usually requires very high level of

technical skills or know-how, we conjecture that education would be positively moderate the

relationship between Internet penetration and cybercrime offenders.

Finally, the Internet use and cybercrime require high initial costs, which act as barriers to

commit cybercrime. According to the economic theory of crime, low income, high level of

poverty, and increased unemployment rate lead to higher crime rates, especially for property

10
crime by providing more economic incentives to use their free time to commit crimes, compared

with benefits from work (Chiu and Madden, 1998; Edmark, 2005; Patterson, 1991; Raphael and

Winter-Ebmer, 2001). Corts et al. (2016) argue that the negative economic shocks, specifically

the crash of Ponzi schemes, increase property crimes with relatively weak judicial and law

enforcement institutions and with little access to microcredit for consumption smoothing. In

addition, low economic conditions have been identified as a significant factor that weakens the

networks of social control (Kelly, 2000; Patterson, 1991; Sampson et al., 1997), increasing crime

rates due to social disorganization (Shaw and McKay, 1942). Contrary to terrestrial crimes,

however, cybercrime commitment involves high initial costs, including Internet service

installation and costly investment in Internet- and cybercrime-related skills, which can act as

barriers to commit cybercrime for individuals with low economic conditions. Thus, we

conjecture that Internet use and cybercrime perpetrators would be more positively correlated in

areas with a higher income, a lower poverty rate, and a lower unemployment rate.

Data

To examine the relationship among Internet penetration, Internet crime, and socioeconomic

factors, we combine state-level data collected from various official sources in the United States,

including the IC3, Census Bureau (CB), Bureau of Labor Statistics (BLS), Bureau of Economic

Analysis (BEA), and Federal Communications Commission (FCC). We obtain Internet crime

data from the IC3s annual report, Internet Crime Report, which provides comprehensive

statistics of Internet crime. Specifically, the report includes the number of individual victims

(complainants) and perpetrators from the 50 states and the District of Columbia (D.C.). Note that

we restrict our focal period to the years 2004-2010, because in the early years, a small number of

11
complaints were received annually possibly due to low awareness for the IC3 and it has not

reported information about perpetrators since 2011 (see Figure 1).

In the IC3, a complainant self-reports her/his location and most importantly, the location of

the perpetrator as represented to her/him by the perpetrator. It is worth noting that the IC3 (2009)

reports that as important as it is to understand the prevalence and monetary impact of

cybercrime, it is also vital to gain insight into who the typical perpetrators are. This can prove to

be difficult in the world of cybercrime, where a mask of anonymity can impede law enforcement

efforts; the gender of the perpetrator was reported only 35.1% of the time, and the state of

residence for domestic perpetrators was reported only 38.0% of the time. This means that only

around one third complainants had some idea about the location of the perpetrators. While it is

very difficult to collect the actual number of crimes on the Internet, we believe that reported

figures by the IC3, which are only available nation-wide longitudinal statistics on cybercrime

perpetrators, can proxy for actual incidents.

Wall (2003) and Yar (2005) report a taxonomy of cybercrimes by four established legal

categories: i) cyber-trespassing: crossing boundaries into other peoples property and/or causing

damage (e.g., hacking, defacement, viruses); ii) cyber-deception and theft: stealing (e.g., credit

card fraud, intellectual property violations); iii) cyber-pornography: activities that breach laws on

obscenity and decency; and iv) cyber-violence: doing psychological harm to, or inciting physical

harm against others, thereby breaching laws pertaining to the protection of the person (e.g., hate

speech, stalking). According to the 2010 Internet Crime Annual Report (IC3, 2010), high-ranked

types of Internet crime include computer crime, non-delivery payment/merchandise, identity

theft, and various types of fraud (see Table A1 in the Appendix). Thus, it appears that our dataset

12
from IC3 may be limited to two categories of cybercrime i) cyber-trespassing and ii) cyber-

deception and theft.

To identify Internet penetration varying across states and years, we use the number of

Internet providers who offer connections over 200 kbps that are reported in the FCC Form 477.3

A number of previous studies have also used the same source as a proxy for Internet penetration

in the United States (e.g., Chan et al., 2016; Kolko, 2010). It is also worth noting that the FCC

has reported the percentage of connections by downloading speed since 2008, which allows us to

separate out the effect of broadband adoption on cybercriminal activities. We define broadband

connections as residential connections over at least 3 Mbps downstream. 4 Narrowband, broadly

defined as the opposite of broadband, is calculated by subtracting broadband connections from

all residential connections over 200 kbps.

Regarding socioeconomic factors, previous studies have identified education, household

income, poverty, unemployment rate, and income inequality as main indicators, and they also

suggest that these factors influence criminal activities (Chan et al., 2016; Chiu and Madden,

1998; Edmark, 2005; Glaeser and Sacerdote, 1999; Kelly, 2000; Lochner and Moretti, 2004;

Patterson, 1991; Sampson et al., 1997). Accordingly, we obtain data on education, the median

household income, and poverty rate from the U.S. Census. Here, we define education rate as the

percentage of the population over 25 years old with a high school diploma or higher, and the

poverty rate as the percentage of the population below the poverty level. The unemployment rate

3
The current population survey (CPS) conducted by the CB reports the number of individuals living in households with Internet
access in each state. However, the CPS does not include the actual usage data on a year-to-year basis, thereby preventing us from
conducting a panel analysis.
4
While the FCC defines broadband as being 4 Mbps down and 1 Mbps up, the FCC form 477 does not report the corresponding
statistics for each state.

13
is defined as the percentage of unemployed individuals out of the employable population,

provided by the U.S. BLS. For the inequality measure, we use the Gini coefficient for each state

from the work of Frank (2009); a high-level Gini score is associated with an unequal distribution

of income.5

While we cannot directly observe the spatial and temporal factors triggering crime events,

prevailing crime levels can capture the various determinants of criminal acts (Chan et al., 2016).

To control for crime-related factors, we obtain general crime data from the FBIs annual report,

Uniform Crime Report. In this report, the FBI provides counts of general crimes classified as

violent crimes and property crimes, by the 50 states and the Washington, D.C. Violent crimes

refer to offenses of murder, forcible rape, robbery and aggravated assault; property crimes refer

to offenses of burglary, larceny-theft and motor vehicle theft. In addition, we combine focal

variables with demographic variables and size of various industry sectors that could influence the

crime rate (see Chan et al., 2016). Table 1 summarizes the variables used in this study and their

sources.

[Table 1 about here]

Empirical Model

Figure 2 summarizes our research model. To examine under which conditions Internet crimes are

more likely to occur, we consider the following empirical model:

( ) = + ( ) + + + + + ,

5 The data are publicly available on http://www.shsu.edu/eco_mwf/inequality.html. A score of zero on the Gini coefficient
represents complete equality, i.e., every person has the same income. A score of one represents complete inequality, i.e., where
one person has all of the income and others have none.

14
where X is a set of control variables including crime-related, demographic and socioeconomic

factors. represents state dummies to account for time-invariant state-level factors such as the

established state laws or regulations. represents year dummies in considering the average time

trends across states e.g., nationwide cybercrime reporting trends, possibly due to the awareness

of IC3 or external shocks, or common effects to all states (e.g., federal laws). represents a

random error not explained by the model. Some regressions also include state-specific linear

time trends ( ) (Wooldridge, 2010). Previous studies employ this specification to further

account for unobservable state-specific (Raphael and Winter-Ebmer, 2001) or country-specific

trends (Cho et al., 2016). As a result, the identification in the main model arises from variation in

Internet penetration and cybercrime perpetrators within states, after netting out state-specific

time trends.

To investigate the moderating effects of socioeconomic status, we extend the baseline

model by including interaction terms between the focal independent variable (( ))

and socioeconomic factors, such as the education rate, median income, inequality, poverty rate,

and unemployment rate. Note that we log-transform the nominal variables (e.g., population),

excluding the percentage variables (e.g., unemployment rate), given that the distributions of

crime, Internet users, and the population are skewed.

[Figure 2 about here]

Since our dataset represents a longitudinal panel structure, we need to address potential

econometric problems. In this case, pooled ordinary least squares regression is inefficient, and

the standard errors are not correct when there exist heteroskedasticity and autocorrelation

(Greene, 2011). To address these potential issues, we employ a fixed-effects regression with

Driscoll and Kraay standard errors (Driscoll and Kraay, 1998), where the error structure is

15
assumed to be heteroskedastic, auto-correlated up to some lag, and possibly correlated between

the panels.

Results

Geographical Distribution of Internet Crime

Before presenting our empirical analysis, we observe geographical distributions of Internet crime

and terrestrial crimes for a comparison. Figure 3 shows the U.S. states from where Internet

crimes mostly originated during 2004-2010.

[Figure 3 about here]

To further understand the degree of the geographical concentration of crime, we apply

several inequality measures and the results are reported in Table 2. The findings suggest large

variations and differences across perpetrators and victims of Internet crime and terrestrial crimes.

Specifically, property crimes are distributed most sparsely across states, and the occurrence of

violent crimes shows higher concentration. More interestingly, the degree of the concentration of

Internet crime perpetrators is consistently higher than that of violent crimes in the terrestrial

world (see Clayton et al. 2015 for anecdotal evidence of cybercrime concentration).

[Table 2 about here]

In addition, perpetrators of Internet crime seem to be more concentrated than Internet crime

victims, and the geographic distribution of cybercrime victims is more evenly dispersed across

states than that of terrestrial crime. These salient observations may imply that cybercriminal

activities could be conducted by a small number of perpetrators who are force-amplified by the

inherent features of the Internet, and all Internet users are potential victims of cybercrime,

16
regardless of where they live. In other words, while benevolent activities can be conducted and

undeterred by time and space on the Internet, at the same time, the Internet space provides

opportunities for criminals to commit malicious conduct by targeting faraway individuals. Thus,

it is necessary to separately investigate the factors affecting the occurrence of both cybercrime

origins and destinations. In the next subsection, we report the main findings of our empirical

analysis.

Effect of the Internet and Broadband Penetration on Internet Crime

We estimate the empirical model by using our panel data of 357 observations, which pools 51

states (including Washington, D.C.) in the United States during 2004-2010. Table 3 reports the

main results, and each column in the table corresponds to a different regression. Note that our

dependent variable is either Internet crime perpetrators or Internet crime victims for a

comparison.

In Columns 1 to 3, our findings suggest that Internet penetration is not significantly

associated with cybercrime perpetrators, even after controlling for industry size and state-specific

time trends. It implies that malicious offences on the web may not be directly affected by the

degree of Internet penetration. In contrast, Internet penetration is positively and significantly

correlated with Internet crime victims (Columns 4 to 6). Notably, these discrepant findings may

be in accordance with the model-free observations in the previous subsection. Victims of

cybercrimes are more likely to be distributed than the origins of cybercrimes and are more likely

to be proportional to Internet penetration.

[Table 3 about here]

17
 Among socioeconomic factors, education seems to have negative effects on cybercriminal

offenses and victimization in Columns 1-2 and 4-5, respectively. In a nutshell, our results

correspond to findings from previous studies that emphasize the importance of education in

mitigating crime (Lochner and Moretti, 2004). In addition, cybercriminal offenses are more

likely to be activated in regions with low poverty rate, whereas household income, inequality

(Gini coefficient) and unemployment rate are not statistically significant. However, the effects

become weaker or even in the opposite direction, after controlling for state-specific time trends

(Columns 3 and 6). We conjecture that state fixed effects and state-specific time trends

subtantially remove variation that would have been captured by independent variables without

these effects, which is commonly reported by previous studies using aggregate level with a

relatively small number of observations (Cheng and Nault, 2012; Han et al., 2011; Stiroh, 2005).

The estimated coefficients of crime-related factors show various aspects regarding the

differences between Internet and terrestrial crimes. Interestingly, both the perpetrators and

victims of Internet crime are not associated with property crimes, but significantly with violent

crimes. The origins of cybercrimes are positively associated with violent crimes. By contrast,

there are more victims of Internet crimes living in safe regions with fewer violent crimes,

possibly resulting in more evenly distributed victims of Internet crime. According to routine

activity theory (RAT), which explains the cause of crimes, criminal acts require the

convergence in space and time of likely offenders, suitable targets, and the absence of capable

guardians (Cohen and Felson, 1979). From the viewpoint of RAT, our finding suggests that

Internet crimes can be distinct from terrestrial violent crimes regarding suitable targets, while

the motivated offenders are largely homogeneous between violent crimes and Internet crimes.

Consistent with Yar (2005)s theoretical argument, this study confirms the salience of

18
cybercrime by providing empirical evidence of the similarity between terrestrial violent crimes

and Internet crimes perpetrators and the difference between violent crimes and cybercrime

victims, which has not been explored by previous studies.

To further examine whether Internet connection speed matters, we divide Internet

connection into broadband and narrowband, as described in the Data section. Table 4 presents

the estimation results after separating broadband and narrowband over the period 2008-2010.

Columns 2-3 and 5-6 highlight the differential effects of connection speed on cybercrime

perpetrators and victims, respectively. In Columns 2 and 3, while narrowband connections have

no significant relationship with cybercriminal offenses, an increase in broadband connection is

positively associated with Internet crime perpetrators, indicating that the ability of potential

offenders appears to be amplified by a faster Internet connection. The results imply that

cybercrime requires an infrastructure to facilitate profiteering from the act, highlighting the dark

side of advanced ICTs. On the other hand, Columns 5 and 6 suggest that both broadband and

narrowband connection significantly contribute to the increase in cybercrime victims, and even

the effect of narrowband appears to be larger than that of broadband.

[Table 4 about here]

Moderating Role of Socioeconomic Status

Our findings in Table 3 suggest that there is no clear evidence that Internet connection type is

directly associated with the occurrence of cybercrime perpetrators. Therefore, we further

investigate the possible moderating role of socioeconomic conditions, where the Internet is

related to cybercriminal offenses. Table 5 presents the estimation results for the moderating

effects of socioeconomic status i.e., the education rate, median income, poverty rate,

19
unemployment rate, and inequality. Note that we include each interaction term separately due to

multicollinearity. While we control for all variables included in the main model, we present the

estimation results only for the main and interaction terms of the focal variables due to space

constraints.

In Columns 1 through 10, our results demonstrate that all socioeconomic factors

significantly moderate the impact of Internet penetration on cybercrime perpetrators in both

specifications without and with state-specific time trends. One of the most notable findings is

that the perpetrators of cybercrimes are likely to increase with Internet penetration in highly

educated areas, unlike the finding that education in itself may be negatively associated with

cybercrime occurrence in Table 3, implying that there is a high skill premium to cybercrime

commitment. In addition, consistent with our theoretical conjecture, the household income

positively moderates, but poverty rate, unemployment rate, and inequality negatively moderate

the relationship between Internet penetration and Internet crime perpetrators. In sum, the Internet

use and cybercrime perpetrators are more positively related in areas with more education, a

higher income, a lower poverty rate, a lower unemployment rate, and a lower inequality.

[Table 5 about here]

Discussion and Conclusions

Given the fact that individuals who can access the Internet rose from 122.6 million (43.1% of the

population) to 279.8 million (86.8%) in the United States for the period 2000-2014,6 Internet

crimes are far from negligible societal challenges in the 21st century. Thus, a variety of online

6
http://www.internetlivestats.com/internet-users/united-states/

20
activities have been increasingly affecting real-world society in many aspects. Cybercrime can

be one of the most salient dark-side impacts from the Internet, which can also be regarded as a

socio-technological phenomenon. However, despite the prominent importance of cybercrime,

there are surprisingly few empirical studies on socioeconomic factors and cybercrime. Our study

may bridge this gap. In particular, this paper is one of the first studies to investigate the roles of

broadband and socioeconomic status in cybercriminal offenses. Our findings are of interest to

ongoing academic efforts to document societal challenges introduced by widespread online

access and ICT-enabled platforms.

Our findings and their implications are threefold: First, we highlighted the distinct

geographical distributions of perpetrators and victims of Internet crime, compared with terrestrial

crimes. Our results demonstrate that Internet crime targets are distributed more equally across the

country, whereas the origins of cybercrime perpetrators are more concentrated, compared with

terrestrial violent crimes. This finding implies that cybercriminal activities may be conducted by

a small fraction of perpetrators. ICTs may serve as a force multiplier, enabling individuals with

minimal resources to generate potentially large negative effects (Yar, 2013), increasing expected

payoffs for cybercriminal offenders. With an increase of such payoffs, network effects and

economies of scale on the web may be associated with the concentration of cybercrime origins

(Clayton et al., 2015).

Second, this study echoed the dark side of advanced ICTs. In the last several decades,

telecommunication technologies have been markedly advanced, and, according to the Pew

Research Center, home broadband users have increased from 1% in 2000 to 67% in 2015.7 High-

7
http://www.pewinternet.org/2015/12/21/home-broadband-2015/

21
speed connection provides various benefits to individuals, communities, and businesses

(Communications Workers of America, 2009), though it may introduce additional threats in

terms of cybercrime. In fact, the level of trust in advanced technologies is quite low. For example,

according to Dapp (2014), the payment method that provides the highest perceived security is

payment by invoice (84% of survey participants), and only 9% of participants perceive that

instant online bank transfers are secure. Digital technologies are constantly growing

(Brynjolfsson and McAfee, 2014); therefore, our findings suggest that the digital economy calls

for an effective strategy with respect to cybercrime prevention.

Third, we found the conditions in which Internet use is significantly associated with

cybercrime activities, which can provide implications for prescriptively preventing potential

Internet crime. Our findings confirm that socioeconomic status plays a critical moderating role in

the relationship between the Internet and cybercriminal activities, rather than having direct

effects on Internet crime perpetrators. In fact, ICTs have played a critical role as a driver of

productivity (Brynjolfsson and Hitt, 1996) and local economic growth (Kolko, 2012), potentially

leading to an increase in household income and a decrease in the poverty rate. Internet access has

also provided various benefits to individuals and communities, such as more opportunities for

education and employment (Atasoy 2013; Lenhart et al. 2010). On the contrary, our findings

highlight the dark side of economic development and societal benefits from Internet use. As

people become more educated and experience less poverty and more employment, the Internet

tends to be more positively associated with cybercrime activity. Interestingly, these findings

appear to indeed differ from the conditions of terrestrial crimes. It would be a fruitful area for

future research to extend the economics of crime into the context of cybercrime, quite different

from terrestrial crimes which have been main focuses in the economics literature.

22
 Our findings call for actions to mitigate the negative effects of malicious Internet use in

order to reconcile the contradictory consequences of Internet penetration and economic growth.

In particular, situational crime prevention strategies aim to reduce opportunities for crimes

arising from routines (Clarke, 1983). The first step in reducing opportunities for crimes is to

understand the conditions in which offenders are more likely to be situated. These strategies have

been proposed as viable responses to Internet crime (Newman and Clarke, 2012). In addition,

Clayton et al. (2015) suggest that cybercrime concentrations are amenable to effective

interventions in social problems; therefore, it is necessary to identify how such viable

interventions would work. In this regard, proper intervention strategies are required, such as

preventive education and campaigns to lessen the motivations of Internet crime, particularly for a

prosperous village with high education rate, high income, low poverty rate, and low

unemployment rate.

Moreover, given the different level of concentrations of Internet crime origins and

destinations, it is essential to address spatial separation when cybercrime prevention policies are

developed. Our results suggest that it is necessary to treat offenders and victims separately:

specifically, cybercriminal acts may be conducted by a small number of malicious perpetrators,

but widely dispersed virtuous Internet users are vulnerable to these crimes. This spatial

separation makes it difficult to apprehend and detect cybercriminals. In this sense, Lee (2015)

proposes the principle of origin responsibility to build an ICT-enabled Bright Society,

indicating that the originating perpetrators are responsible for the consequences of malicious

offenses. Thus, it is important to develop technologies and policies of identifiable anonymity and

the issuance of digital warrants through global collaboration.

23
 Naturally, our study is not without limitations, and these can be promising future research

topics. Due to the limited data available, we analyzed only the reported number of cybercrime

perpetrators and victims from the IC3. While it is very difficult to collect the actual number of

crimes on the Internet, we believe that reported figures by the IC3 can effectively proxy for

actual incidents. In addition, looking at the detailed dynamics of cybercrime may also be

interesting, given that not all cybercrimes yield an equal amount of damage. Future research can

replicate our approach by using data that include either more comprehensive crime information

or financial damage. While we control for state-level heterogeneity by including state fixed

effects, another important direction for future research would be to analyze policy effects on

cybercrime if some states enact anti-cybercrime regulations.

24
References

Acemoglu, D., 1998. Why Do New Technologies Complement Skills? Directed Technical
Change and Wage Inequality. Q. J. Econ. 113, 10551089.

Anderson, R., Barton, C., Bohme, R., Clayton, R., Van Esten, M.J., Levi, M., 2013. Measuring
the Cost of Cybercrime, in: The Economics of Information Security and Privacy. Springer,
Berlin.

Arthur, W.B., 1989. Competing Technologies, Increasing Returns, and Lock-in Historical Events.
Econ. J. 99, 116131.

Atasoy, H., 2013. The Effects of Broadband Internet Expansion on Labor Market Outcomes. ILR
Rev. 66, 315345.

Autor, D.H., Katz, L.F., Krueger, A.B., 1998. Computing Inequality: Have Computers Changed
the Labor Market? Q. J. Econ. 113, 11691213.

Ayyagari, R., Grover, V., Purvis, R., 2011. Technostress: technological antecedents and
implications. MIS Q. 35, 831858.

Bakos, Y., Brynjolfsson, E., 2000. Bundling and Competition on the Internet. Mark. Sci. 19, 63
82.

Bakos, Y., Brynjolfsson, E., 1999. Bundling Information Goods: Pricing, Profits, and Efficiency.
Manage. Sci. 45, 16131630.

Becker, G.S., 1968. Crime and Punishment: An Economic Approach. J. Polit. Econ. 76, 169217.

Bhuller, M., Havnes, T., Leuven, E., Mogstad, M., 2013. Broadband internet: An information
superhighway to sex crime? Rev. Econ. Stud. 80, 12371266.

Bresnahan, T.F., Brynjolfsson, E., Hitt, L.M., 2002. Information Technology, Workplace
Organization, and the Demand for Skilled Labor: Firm-Level Evidence. Q. J. Econ. 117,
339376.

Broadhurst, R., Grabosky, P., Alazab, M., Bouhours, B., Chon, S.K., 2013. Crime in Cyberspace:
Offenders and the Role of Organized Crime Groups. SSRN Electron. J.

Brynjolfsson, E., Hitt, L., 1996. Paradox Lost? Firm-Level Evidence on the Returns to
Information Systems Spending. Manage. Sci. 42, 541558.

Brynjolfsson, E., Hu, Y.J., Smith, M.D., 2010. Long Tails vs. Superstars: The Effect of
Information Technology on Product Variety and Sales Concentration Patterns. Inf. Syst. Res.
21, 736747.

25
Brynjolfsson, E., McAfee, A., 2014. The Second Machine Age: Work, Progress, and Prosperity
in a Time of Brilliant Technologies. W.W. Norton & Company, New York.

Capeller, W., 2001. Not Such a Neat Net: Some Comments on Virtual Criminality. Soc. Leg.
Stud. 10, 229242.

Chan, J., Ghose, A., 2014. Internets Dirty Secret: Assessing the Impact of Online Intermediaries
on HIV Transmission. MIS Q. 38, 955975.

Chan, J., Ghose, A., Seamans, R., 2016. The Internet and Racial Hate Crimes: Offline Spillovers
from Online Access. MIS Q. 40, 381403.

Cheng, Z. (June), Nault, B.R., 2012. Relative Industry Concentration and Customer-Driven IT
Spillovers. Inf. Syst. Res. 23, 340355. doi:10.1287/isre.1100.0345

Chiu, W.H., Madden, P., 1998. Burglary and Income Inequality. J. Public Econ. 69, 123141.

Cho, D., Smith, M.D., Zentner, A., 2016. Internet Adoption and the Survival of Print
Newspapers: A Country-Level Examination. Inf. Econ. Policy 37, 1319.
doi:10.1016/j.infoecopol.2016.10.001

Clarke, R. V., 1983. Situational Crime Prevention: Its Theoretical Basis and Practical Scope.
Crime and Justice 4, 225256.

Clayton, R., Moore, T., Christin, N., 2015. Concentrating Correctly on Cybercrime
Concentration, in: 14th Annual Workshop on the Economics of Information Security.

Cohen, L.E., Felson, M., 1979. Social Change and Crime Rate Trends: A Routine Activity
Approach. Am. Sociol. Rev. 44, 588608.

Communications Workers of America, 2009. Speed Matters: Benefits of Broadband.

Communications Workers of America, Washington, D.C.

Cornish, D.B., Clarke, R. V., 2014. The Reasoning Criminal: Rational Choice Perspectives on
Offending. Transaction Publishers, Piscataway.

Corts, D., Santamara, J., Vargas, J.F., 2016. Economic Shocks and Crime: Evidence from the
Crash of Ponzi Schemes. J. Econ. Behav. Organ. 131, 263275.

Dapp, T., 2014. Fintech The Digital Revolution in the Financial Sector. Deutsche Bank
Research, Frankfurt am Main.

Driscoll, J., Kraay, A., 1998. Consistent covariance matrix estimation with spatially dependent
data. Rev. Econ. Stat. 80, 549560.

Edmark, K., 2005. Unemployment and Crime: Is There a Connection? Scand. J. Econ. 107, 353
373.

26
Frank, M.W., 2009. Inequality and growth in the united states: Evidence from a new state-level
panel of income inequality measures. Econ. Inq. 47, 5568.

Ganju, K.K., Pavlou, P.A., Banker, R.D., 2016. Does Information and Communication
Technology Lead to the Well-Being of Nations? A Country-Level Empirical Investigation.
MIS Q. 40, 417430.

Glaeser, E.L., Sacerdote, B., 1999. Why Is There More Crime in Cities? J. Polit. Econ. 107,
S225S258.

Glaeser, E.L., Sacerdote, B., Scheinkman, J.A., 1996. Crime and Social Interactions. Q. J. Econ.
111, 507548.

Greene, W.H., 2011. Econometric Analysis, 7th ed. Prentice Hall, New Jersey.

Greenwood, B.N., Wattal, S., 2016. Show Me the Way to Go Home: An Empirical Investigation
of Ride-Sharing and Alcohol Related Motor Vehicle Fatalities. MIS Q.

Han, K., Kauffman, R., Nault, B., 2011. Returns to Information Technology Outsourcing. Inf.
Syst. Res. 22, 824840.

IC3, 2014. 2014 Internet Crime Report. Internet Crime Complaint Center.

IC3, 2010. 2010 Internet Crime Report. Internet Crime Complaint Center.

IC3, 2009. 2009 Internet Crime Report. Internet Crime Complaint Center.

John, J.P., Moshchuk, A., Gribble, S.D., Krishnamurthy, A., 2009. Studying Spamming Botnets
using Botlab, in: NSDI09 Proceedings of the 6th USENIX Symposium on Networked
Systems Design and Implementation. pp. 291306.

Kelly, M., 2000. Inequality and Crime. Rev. Econ. Stat. 82, 530539.

Kigerl, A., 2011. Routine Activity Theory and the Determinants of High Cybercrime Countries.
Soc. Sci. Comput. Rev. 30, 470486.

Kim, S.H., Wang, Q.-H., Ullrich, J.B., 2012. A Comparative Study of Cyberattacks. Commun.
ACM 55, 6673.

Kolko, J., 2012. Broadband and local growth. J. Urban Econ. 71, 100113.

Kolko, J., 2010. A new measure of US residential broadband availability. Telecomm. Policy 34,
132143.

Lakhina, A., Crovella, M., Diot, C., 2005. Detecting distributed attacks using network-wide flow
traffic, in: Proceedings of FloCon 2005 Analysis Workshop.

Lee, J.K., 2015. Research Framework for AIS Grand Vision of the Bright ICT Initiative. MIS Q.
39, IIIXII.

27
Lochner, L., 2004. Education, Work, and Crime: A Human Capital Approach. Int. Econ. Rev.
(Philadelphia). 45, 811843.

Lochner, L., Moretti, E., 2004. The Effect of Education on Crime: Evidence from Prison
Inmates, Arrest, and Self-Reports. Am. Econ. Rev. 94, 155189.

Machin, S., Van Reenen, J., 1998. Technology and Changes in Skill Structure: Evidence from
Seven OECD Countries. Q. J. Econ. 113, 12151244.

Merton, R.K., 1938. Social Structure and Anomie. Am. Sociol. Rev. 3, 672682.

Michaels, G., Natraj, A., Van Reenen, J., 2014. Has ICT Polarized Skill Demand?: Evidence
from Eleven Countries over 25 Years. Rev. Econ. Stat. 96, 6077.

Munyo, I., Rossi, M.A., 2013. Frustration, Euphoria, and Violent Crime. J. Econ. Behav. Organ.
89, 136142.

Newman, G., Clarke, R., 2012. Superhighway Robbery: Preventing E-Commerce Crime. Willan
Press, Cullompton.

Patterson, E.B., 1991. Poverty, Income Inequality, and Community Crime Rates. Criminology 29,
755776.

Raphael, S., Winter-Ebmer, R., 2001. Identifying the Effect of Unemployment on Crime. J. Law
Econ. 44, 259283.

Sampson, R.J., Raudenbush, S.W., Earls, F., 1997. Neighborhoods and Violent Crime: A
Multilevel Study of Collective Efficacy. Science (80-. ). 277, 918924.

Shaw, C., McKay, H., 1942. Juvenile Delinquency and Urban Areas. University of Chicago
Press, Chicago.

Stiroh, K.J., 2005. Reassessing the Impact of IT in the Production Function: A Meta-Analysis
and Sensitivity Tests. Ann. Econ. Stat. 529561. doi:10.2307/20777587

Symantec, 2016. Internet Security Threat Report 2016. Symantec.

Thomas, D., Loader, B., 2000. Cybercrime: Law Enforcement, Security and Surveillance in the
Information Age. Routledge, London.

Wall, D., 2003. Crime and the Internet. Routledge, London.

Wooldridge, J.M., 2010. Econometric Analysis of Cross Section and Panel Data, 2nd ed. MIT
Press.

Yar, M., 2013. Cybercrime and Society, 2nd ed. SAGE Publications, London.

Yar, M., 2005. The Novelty of Cybercrime: An Assessment in Light of Routine Activity
Theory. Eur. J. Criminol. 2, 407427.

28
Figure 1: Internet Crime in the United States (2000-2014)

Figure 2: Research Model

29
 Figure 3: Top States of Internet Crime Origins in the United States

Notes: The top 15 states ranked by Internet crime perpetrators are dark-shaded; Ranks are in parentheses; See
Table 2A in Appendix for the detailed statistics.

30
 Table 1: Descriptive Statistics (2004-2010)
Variables Mean Std. Dev. Min. Max. Description Source
Internet Crime
Total number of individual perpetrators where state is
Internet Crime Perpetrators 1,467.2 2,260.4 41.2 15,952.5 IC3
known
Total number of individual complainants where state is
Internet Crime Victims 3,897.5 5,045.5 174.9 39,755.5 IC3
known
Internet Penetration
Total number of providers of Internet connections over
Internet 59.0 34.5 7 204 200 kbps in at least one direction for residential use FCC
(2004-2010)
Broadband Connections Residential connections over at least 3 Mbps downstream
964.1 1,120.5 60.2 6,405.6 FCC
(in thousands) (2008-2010)
Narrowband Connections Residential connections below 3 Mbps downstream and
1,248.3 1,629.1 64.5 11,638.4 FCC
(in thousands) over 200 kbps in at least one direction (2008-2010)
General Crime
Offenses of murder, forcible rape, robbery and
No. of Violent Crimes 26,695.9 34.791.2 504 198,070 FBI
aggravated assault
No. of Property Crimes 191,560.3 220,777.9 11,895 1,227,194 Offenses of burglary, larceny-theft and motor vehicle theft FBI
Demographic Factors
Population (in thousands) 5,905.2 6,589.2 509.1 37,349.4 Population estimates for each state CB
Nonwhite Rate 19.4 13.5 3.5 74.4 Ratio of non-Whites to total population CB
Median Age 36.8 2.2 27.8 42.8 Median age of population CB
Socioeconomic Factors
Percentage of population over 25 years old with a high
Education Rate 86.1 3.6 77.3 92.3 CB
school diploma or higher
Median household annual income (2014 CPI-U-RS
Median Income 55,896.4 8,427.4 38,706 79,915 CB
adjusted dollars)
Poverty Rate 12.9 3.0 6.6 21.7 Percentage of population below the poverty level CB
Percentage of unemployed individuals out of the
Unemployment Rate 6.1 2.3 2.4 14 BLS
employable population in November (seasonally adjusted)
Gini coefficient for each state, ranging from 0 (complete
Gini Coefficient 0.61 0.03 0.54 0.71 CB
equality) to 1 (complete inequality)

31
 Table 1: Descriptive Statistics (Continued)
Variables Mean Std. Dev. Min. Max. Description Source
Industry Sector Size
Relative GDP of Utilities Percentage of GDP from utilities industry with NAICS
1.94 0.53 0.61 3.65 BEA
Industry code 22 to Total GDP for each state
Relative GDP of Percentage of GDP from information industry with NAICS
3.74 1.63 1.36 9.83 BEA
Information Industry code 51 to Total GDP for each state
Relative GDP of Finance Percentage of GDP from finance and insurance industry
6.15 4.04 1.95 27.00 BEA
and Insurance Industry with NAICS code 52 to Total GDP for each state
Relative GDP of
Percentage of GDP from professional, scientific, and
Professional, Scientific,
5.90 2.95 2.38 22.90 technical services industry with NAICS code 54 to Total BEA
and Technical Services
GDP for each state
Industry

32
 Table 2: Inequality Measures for Crimes
Internet Crime Internet Crime
Measure Violent Crime Property Crime
Perpetrators Victims
Relative Mean Deviation 0.203 0.126 0.197 0.103
Coefficient of Variation 0.791 0.526 0.533 0.244
Gini Coefficient 0.275 0.178 0.269 0.137
Kakwani Measure 0.083 0.039 0.067 0.018
Theil Entropy Measure 0.213 0.091 0.122 0.030

Notes: All inequality measures are scale-free; all measures are calculated based on the number of crimes per
100,000 for each year and are averaged for 7 years.

33
 Table 3: Estimation Results for the Main Effects (2004-2010)
DV: ln(Internet Crime Perpetrators) ln(Internet Crime Victims)
(1) (2) (3) (4) (5) (6)
0.049 0.040 -0.019 0.206*** 0.201*** 0.027*
ln(Internet)
(0.054) (0.051) (0.112) (0.038) (0.037) (0.016)
0.461*** 0.418*** -0.180 -0.227*** -0.218*** -0.012
ln(Violent Crime)
(0.040) (0.045) (0.174) (0.062) (0.061) (0.051)
0.094 0.115 0.021 -0.001 0.000 -0.010
ln(Property Crime)
(0.094) (0.095) (0.026) (0.050) (0.049) (0.022)
4.129*** 4.112*** -1.277 1.447*** 1.460*** 1.124
ln(Population)
(0.687) (0.738) (3.165) (0.522) (0.509) (0.705)
-0.178*** -0.170*** 0.124 0.013 0.012 -0.104**
Nonwhite Rate
(0.046) (0.046) (0.158) (0.011) (0.010) (0.044)
-0.176*** -0.165*** -0.204** 0.009 0.005 -0.034
Median Age
(0.049) (0.034) (0.081) (0.029) (0.029) (0.035)
-0.079*** -0.079*** 0.042*** -0.023*** -0.021*** -0.018**
Education Rate
(0.025) (0.021) (0.014) (0.005) (0.005) (0.007)
0.489 0.529 0.047 -0.132 -0.106 0.036
ln(Median Income)
(0.384) (0.423) (0.348) (0.127) (0.120) (0.101)
-0.043* -0.045* 0.047* -0.047*** -0.048*** 0.001
Poverty Rate
(0.022) (0.024) (0.023) (0.015) (0.015) (0.012)
0.004 0.001 -0.011 0.003 0.001 -0.008
Unemployment Rate
(0.009) (0.009) (0.011) (0.008) (0.008) (0.010)
0.829 0.892 -1.313*** -0.597 -0.596 -0.619
Gini Coefficient
(1.196) (1.139) (0.145) (0.545) (0.558) (0.370)
Industry Controls No Yes Yes No Yes Yes
Year Fixed Effects Yes Yes Yes Yes Yes Yes
State Fixed Effects Yes Yes Yes Yes Yes Yes
State Specific Trends No No Yes No No Yes
Observations 356 356 356 356 356 356
Within R-sq. 0.801 0.803 0.912 0.929 0.929 0.969

Notes: Driscoll-Kraay standard errors are in parentheses; * p<0.1, ** p<0.05, *** p<0.01.

34
 Table 4: Results for the Role of Broadband Connections (2008-2010)
DV: ln(Internet Crime Perpetrators) ln(Internet Crime Victims)
(1) (2) (3) (4) (5) (6)
0.380* 0.439***
ln(Internet)
(0.208) (0.047)
0.422** 0.619*** 0.192*** 0.061
ln(Broadband)
(0.167) (0.095) (0.045) (0.039)
-0.004 -0.074 0.311*** 0.263***
ln(Narrowband)
(0.151) (0.076) (0.023) (0.010)
Other Controls Yes Yes Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes Yes Yes
State Fixed Effects Yes Yes No Yes Yes No
State Specific Trends No No Yes No No Yes
Observations 153 153 153 153 153 153

Notes: All other variables shown in Table 3 and industry controls are included; Driscoll-Kraay standard errors are in
parentheses; The model including both state fixed effects and state-specific trends fails to achieve convergence,
possibly due to relatively short time period; * p<0.1, ** p<0.05, *** p<0.01.

35
 Table 5: Results for the Role of Socioeconomic Status (2004-2010)
DV: ln(Internet Crime Perpetrators)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
-2.224*** -1.151*** -12.89*** -6.946* 0.564*** 0.231 0.171*** -0.078 0.750** 0.682**
ln(Internet)
(0.184) (0.376) (1.854) (4.106) (0.056) (0.166) (0.036) (0.142) (0.364) (0.306)
-0.154*** -0.004
Education Rate
(0.019) (0.024)
ln(Internet) * 0.026*** 0.013***
Education (0.003) (0.004)
-4.182*** -2.418*
ln(Median Income)
(0.944) (1.241)
ln(Internet) * 1.173*** 0.631*
ln(Median Income) (0.167) (0.374)
0.151*** 0.126**
Poverty Rate
(0.043) (0.048)
ln(Internet) * -0.047*** -0.021**
Poverty (0.005) (0.009)
0.130*** -0.051
Unemployment Rate
(0.019) (0.045)
ln(Internet) * -0.032*** 0.010
Unemployment (0.004) (0.010)
5.578*** 3.473
Gini Coefficient
(1.405) (2.236)
ln(Internet) * -1.196* -1.170**
Gini Coefficient (0.624) (0.555)
Other Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
State Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
State Specific Trends No Yes No Yes No Yes No Yes No Yes

Notes: All other variables shown in Table 3 and industry controls are included; Driscoll-Kraay standard errors are in parentheses; * p<0.1, ** p<0.05, *** p<0.01.

36
Appendix

Table A1: Top Ten Types of Internet Crime

Percentage in
Complaint types Description
2010
Non-Delivery Purchaser did not receive items purchased, or seller did not receive
14.4
Payment/Merchandise payment for items sold
FBI-Related Scams Scams in which criminals pose as the FBI to defraud victims 13.2
Identity Theft Unauthorized use of victims personally identifying information to
9.8
commit fraud or other crimes
Computer Crimes 1) Crimes that target computer networks or devices directly or 2)
9.1
crimes facilitated by computer networks or devices
Miscellaneous Fraud Variety of scams meant to defraud the public, such as work-at-home
scams, fraudulent sweepstakes and contests, and other fraudulent 8.6
schemes
Advance Fee Fraud Criminals convince victims to pay a fee to receive something of value,
7.6
but do not deliver anything of value to the victim
Spam Mass-produced, unsolicited bulk messages 6.9
Auction Fraud Fraudulent transactions that occur in the context of an online auction
5.9
site
Credit Card Fraud Fraudulent, unauthorized charging of goods and services to a victims
5.3
credit card
Overpayment Fraud An incident in which the complainant receives an invalid monetary
instrument with instructions to deposit it in a bank account and to send
5.3
excess funds or a percentage of the deposited money back to the
sender

Source: 2010 Internet Crime Annual Report

37
 Table A2: Internet Crime and Terrestrial Crime Statistics by State
Internet Crime Internet Crime
Violent Crime Property Crime
State Victims per Perpetrators per
per 100,000 (rank) per 100,000 (rank)
100,000 (rank) 100,000 (rank)
ALASKA 317.99 (1) 29.87 (11) 648.48 (7) 3244.35 (25)
COLORADO 103.07 (2) 26.00 (13) 358.73 (26) 3230.78 (26)
NEVADA 92.17 (3) 66.54 (2) 686.04 (4) 3656.07 (16)
DISTRICT OF COLUMBIA 90.17 (4) 172.11 (1) 1409.54 (1) 4828.95 (1)
WASHINGTON 87.57 (5) 41.87 (4) 334.94 (28) 4197.86 (3)
NEW JERSEY 87.54 (6) 24.04 (16) 333.85 (29) 2246.03 (47)
MARYLAND 86.37 (7) 21.91 (20) 641.40 (9) 3401.75 (21)
ARIZONA 84.65 (8) 30.50 (10) 466.39 (18) 4371.70 (2)
OREGON 83.27 (9) 21.28 (22) 273.85 (36) 3641.66 (17)
FLORIDA 79.22 (10) 39.80 (5) 671.09 (5) 3971.86 (8)
CALIFORNIA 75.77 (11) 32.05 (9) 507.08 (14) 3036.15 (29)
UTAH 75.48 (12) 32.80 (8) 224.24 (46) 3540.51 (19)
VIRGINIA 74.82 (13) 17.89 (34) 258.08 (41) 2504.77 (40)
NEW HAMPSHIRE 72.11 (14) 18.08 (32) 151.26 (48) 2005.98 (49)
HAWAII 70.71 (15) 21.72 (21) 267.67 (39) 4083.96 (6)
WYOMING 70.64 (16) 25.86 (14) 227.82 (45) 2878.89 (33)
IDAHO 69.33 (17) 18.08 (33) 238.03 (44) 2320.49 (46)
DELAWARE 67.46 (18) 42.58 (3) 647.45 (8) 3349.48 (22)
VERMONT 65.34 (19) 20.83 (25) 127.15 (50) 2348.35 (44)
KANSAS 63.41 (20) 16.73 (39) 402.78 (25) 3556.33 (18)
MONTANA 63.09 (21) 39.71 (6) 271.50 (38) 2734.69 (34)
NEW MEXICO 62.39 (22) 13.23 (47) 650.67 (6) 3869.92 (14)
MAINE 61.50 (23) 28.86 (12) 115.50 (51) 2443.61 (41)
INDIANA 60.91 (24) 17.55 (36) 325.55 (30) 3321.05 (23)
MISSOURI 60.28 (25) 17.38 (37) 502.53 (16) 3684.47 (15)
CONNECTICUT 59.89 (26) 20.67 (26) 282.20 (35) 2439.56 (42)
TENNESSEE 59.18 (27) 21.28 (23) 709.28 (3) 4036.20 (7)
SOUTH CAROLINA 59.04 (28) 15.43 (43) 728.19 (2) 4197.34 (4)
MASSACHUSETTS 58.77 (29) 18.29 (30) 452.41 (20) 2380.06 (43)
PENNSYLVANIA 58.64 (30) 18.10 (31) 406.88 (24) 2345.98 (45)
TEXAS 58.51 (31) 21.21 (24) 506.61 (15) 4116.10 (5)
NORTH CAROLINA 57.34 (32) 18.50 (28) 441.86 (21) 3943.27 (9)
RHODE ISLAND 57.24 (33) 22.45 (18) 244.56 (43) 2688.64 (36)
NEW YORK 57.09 (34) 36.48 (7) 415.90 (23) 2029.86 (48)
OHIO 56.23 (35) 18.32 (29) 340.31 (27) 3485.37 (20)
MINNESOTA 56.23 (36) 16.51 (41) 272.83 (37) 2900.43 (32)
GEORGIA 56.19 (37) 25.85 (15) 453.84 (19) 3935.87 (10)
MICHIGAN 55.83 (38) 17.79 (35) 518.51 (12) 2987.63 (30)
WEST VIRGINIA 55.44 (39) 14.01 (45) 283.43 (33) 2516.13 (39)
ILLINOIS 54.37 (40) 22.09 (19) 518.13 (13) 2938.91 (31)

38
WISCONSIN 53.66 (41) 13.48 (46) 257.97 (42) 2693.01 (35)
OKLAHOMA 53.26 (42) 16.73 (38) 501.90 (17) 3275.82 (24)
ALABAMA 52.55 (43) 16.26 (42) 430.27 (22) 3885.25 (12)
NEBRASKA 50.77 (44) 19.82 (27) 292.09 (31) 3108.46 (28)
ARKANSAS 49.57 (45) 12.63 (49) 519.14 (11) 3879.88 (13)
LOUISIANA 49.37 (46) 12.51 (50) 640.79 (10) 3918.32 (11)
KENTUCKY 47.98 (47) 15.11 (44) 266.74 (40) 2539.83 (38)
IOWA 47.12 (48) 12.88 (48) 282.41 (34) 2589.90 (37)
NORTH DAKOTA 43.55 (49) 22.74 (17) 148.57 (49) 1911.47 (50)
SOUTH DAKOTA 41.08 (50) 16.66 (40) 191.89 (47) 1742.76 (51)
MISSISSIPPI 36.15 (51) 9.23 (51) 285.62 (32) 3146.70 (27)

Note: All measures are based on the average No. of crimes per 100,000 for 7 years; states are sorted by the ranking
of Internet crime victims per 100,000.

39