One way to think about criminal organizations is as social networks. A network is a set of discrete entities connected to one another. By using a network-based approach to analyze the links among criminals, scholars have developed a better appreciation for the relationships that underpin organized crime. Research has begun to capitalize on this progress and suggests specific strategies by which law enforcement can understand, prevent, disrupt, and demolish criminal organizations based on the nature of the underlying networks.
Criminal Organizations as Networks
Criminal organizations operate as businesses with the objective to make money. However, unlike people or firms engaged in legal activities, perpetrators of illegal activities cannot access the many institutions that lower the costs of doing business and allow engagement in normal economic transactions. Formal institutions such as the legal system or marketplaces in which goods and services are bought and sold are central to profitability through the efficient enforcement of contracts and transmission of information these institutions allow. Those engaged in criminal activities have limited access to formal institutions and are therefore more likely to rely on comparatively costly interpersonal channels of communication and enforcement of agreements.
Unlike market institutions in which a person can participate with minimal revelation of his or her characteristics—a corner store hardly cares about one’s personal qualities beyond the ability to pay for one’s vegetables or soda pop—illegal transactions require greater knowledge about the participants on both sides of the transaction as well as trust between them. The seller and buyer in an illegal transaction need to worry about the quality and nature of the good or service being bought or sold and the likelihood that enforcement of the informal contract between the parties is possible at sufficiently low cost. Contrast, for example, a legal transaction at the pharmacy for aspirin using a credit card versus an illegal purchase of cocaine from a street dealer using cash. Not only are the participants forgoing the use of credit or other convenient means of payment that would leave a trail, but they are also subject to the added anxiety about whether one of them might be a police agent.
A shorthand characterization of the key difference between the required information about participants in legal and illegal activities is to say that markets allow for transactions among anonymous individuals who need only to know the going price to make a purchase, whereas illegal markets require intricate trust and knowledge among the participants. Social networks intrinsically emphasize the individual actors and their relationships to one another and are consequently a valuable way to model and study relationships among criminals.
What Constitutes a Network?
A social network is defined as a set of individuals and the connections among them. Each individual may be thought of as a node from whom connections (links) spring. The connections can be unidirectional or bidirectional and can have an intensity that is potentially different in each direction. There are two basic approaches to model the structure of connections in a network. The first approach assumes that the connections are preexisting or determined outside the scope of the analysis. Prime examples are networks based on kinship that are extensively used in anthropology and sociology. In the context of crime, an example would be crime families. The second approach assumes that networks form as a consequence of purposeful behavioral choices. The participants in a network are better-off through their group behavior than as stand-alone criminals. While the cosa nostra (mafia) may focus on family ties, there are numerous criminal organizations ranging from outlaw motorcycle gangs to street gangs to the yakuza that display no such family associations.
Network Persistence
Typically, criminal networks display persistence— the particular structure of links and nodes remains relatively stable over the period of time relevant for analysis. The persistence of criminal networks allows them to create and preserve valuable information and trust among their members. In some cases, this may be a consequence of familial ties or common interests, but more generally, the persistence arises as the outcome of adaptive behavior to a stable environment. Network analysis assumes a set of external or environmental conditions that can be viewed as inputs into the behavior of the criminals pursuing their interests. The underlying stability of the players and links that form the network is an important result of this behavioral process.
Examples of Criminal Networks
Criminal networks have been used as a tool for analyzing a wide variety of applications. Human trafficking; illicit substance distribution including tobacco, alcohol, cocaine, heroin, and other drugs; the manufacture and provision of illegal goods; delinquency; and other illegal activities have all been studied through the lens of network theory. These studies have emphasized the ways in which criminals are connected to each other and the specific configurations of connections that appear to lead to particular benefit to the network. The studies frequently pinpoint individuals who are centrally located or are key players in the network’s structure. The ambit of research on criminal organizations also naturally extends to some aspects of terrorism. The structure of at least some terrorist organizations has many of the characteristics of a network, albeit often more ideologically motivated than traditional criminal networks.
Some dimensions of observed networks are easily characterized. For example, a study of heroin dealing finds that the star structure in which each dealer is connected to a number of clients fits the data well. Suppliers, however, are also connected to each other as well as to their dealers. The overall shape of the organization that puts heroin on the street is hence not well mapped to a traditional top-down hierarchical structure but is better described by a network. Numerous other criminal networks have been observed, but part of the role of recent research is to identify what network structures researchers should expect to observe. Thus, the role of network analysis is not only descriptive but also prescriptive, as it may suggest network characteristics useful for policy.
The Network Approach to Organized Crime
Network theory studies the consequences of the links among participants. These links are simultaneously a source of benefit and a source of risk or other costs. Compared to acting individually, the benefit from being connected to other criminals arises from an underlying activity that allows the network members to obtain a greater gain by participating in the network. By being connected through people who are also connected, an individual’s benefit is drawn from the entire network not simply from one’s direct connections. Similarly, the costs of being part of a network may be also related to the extent of one’s connections: There is a risk of discovery or capture by law enforcement as well as costs of establishing and maintaining links with other criminals. The interplay of the costs and benefits both determine and circumscribe the scope and structure of links in the network.
Early analysis of criminal organizations tended to use organizational theory to describe criminal groups. Such theories emphasize relationships among patrons and clients in the context of a structured crime business. The network-based approach places the emphasis elsewhere. For example, some crime-related network theories highlight the importance of lacunae among subgroups of participants who are linked by a small number of agents who act as brokers of information and providers of services between more tightly organized groups. These brokers or key players are seen as important to the overall criminal enterprise because they raise the income generated by the entire network.
Others argue for a functional rather than an individual-based approach to characterizing criminal networks. For example, a study of human trafficking has looked at the tasks associated with recruitment, transportation, and exploitation. Certain fundamental roles have to be filled within the trafficking network, but there is no need for a single structured organization or specific individuals. Individual actors can form temporary connections with a variety of other players, but the roles and the links among them remain the same.
The network approach is distinct from alternative approaches that emphasize associations based on the aggregation of individuals. These peer effect approaches assert that average or aggregated characteristics or actions by members of the group affect individual behavior. Typically, all group members are treated equally, without regard to the precise structure of social connections among them. In contrast, network theory generates its own set of concepts related to the multiplicity of possible configurations of links among the members. Ideas such as network density, characterizing the frequency of links among players; network centrality of particular players (how many paths among members pass through specific nodes); and distance (the number of links separating one player from another) play important roles. For example, greater delinquency from peer association has been found to increase with network density. Consequently, the specific structure of a crime network is relevant for characterizing criminal behavior beyond simple peer effects. Location in the network—which can be quantified using the measures of density, centrality, and the number of links associated with a node—can provide a better predictor of delinquent involvement than simply knowing the aggregate attitudes of a criminal’s peers.
Optimal Crime Networks
The network theories characterizing criminal organizations discussed thus far are traditional and descriptive. They are characterized by an extant organization made up of interpersonal links about which a description is constructed of centrality, connectedness, distance, among others. More recent functional approaches point to the origin of the networks themselves. Networks arise as a consequence of underlying intentional and strategic behavior by their participants. For example, by using an assumption common to economists—that individuals maximize their wellbeing (or utility)—the costs and benefits from criminal activity arise from the underlying parameters of individual payoffs and the economic and criminal environment. This in turn determines the observed or predicted structure of the network, in addition to the activities of its members. Thus, the characterizations of centrality, density, key players, and the like can be viewed as outcomes of rational choice instead of mere statistical network descriptors.
Sharing this common assumption, one strand of research takes the network structure as determined prior to the analysis (but possibly unknown) and analyzes the outcomes achieved in a strategic equilibrium across different network structures. By strategic, it is meant that players are aware of or make behavioral assumptions about the actions of other players in the network. For a particular network, the incentives of individuals to engage in crime depend on the structure of links among them. Each criminal chooses a level of effort (criminal activity) assuming, for example, that the effort levels of all other members in the network are outside of his or her control. The chosen effort level determines the criminal’s net payoffs: the rewards less costs. The latter might include being captured and penalized if caught, the psychic costs of being a criminal, or out-of-pocket costs.
The individual criminal efforts and the aggregate crime level are determined in a Nash equilibrium. This is a game theory concept in which each player treats the strategies or actions of the other players as fixed and chooses his or her best response to them. In the equilibrium outcome, no player can gain by unilaterally changing his or her own strategy. This framework allows a characterization of individual activity in a crime network as a function of the Bonacich centrality measure, according to which each network member is assigned a weighted score based on his or her direct and indirect links to others in the network. This in turn allows identification of the key player in a crime organization—the individual whose removal would reduce criminal activity by the largest amount. A policy implication is that law enforcement policy that directly targets such individuals would be more successful than a policy targeting those who simply do the most crime or a policy targeting criminals in an indiscriminate (random) way.
A second approach to modeling crime networks takes the network itself as an object of choice. Network formation is viewed as the result of a strategic optimization process. Links are formed at the discretion of the individuals involved, and game theory is used to predict what stable networks emerge. This approach retains the idea of maximizing behavior by the participants and assumes Nash equilibrium, but unlike efforts in which the network is predetermined and outside the scope of the analysis, the network itself is also the outcome of optimization. The latter can either take a decentralized form—the network is the end result of the actions of individuals forming links strategically but bilaterally—or a centralized form by assuming, for example, a criminal godfather who controls the network and shapes it in a way to maximize profits. Terrorist activity can also fit into this framework. Finally, the optimal network structure could also emerge as the result of natural selection through the process of competition (e.g., gang wars) with other criminal organizations.
Implications for Law Enforcement Policy
Network theory provides a useful framework for understanding both the visible and hidden links in a criminal organization. If the socioeconomic or statistical model of a criminal network is sufficiently precise and the appropriate data exist, then a prediction about the structure of the criminal network—who is linked to whom—provides valuable guidance to law enforcement about where to look for what is missing among the observed connections.
Law enforcement is particularly interested in what strategies would do the greatest harm to a criminal organization. The network approach offers clear prescriptions about this. Given limited resources, identifying the key player, the person whose removal will cause the greatest harm to the network in terms of criminal activity, is important. Measures of network centrality are shown to provide a useful guide to identifying the appropriate target. Researchers have extended the same logic to target optimal groups of key players. Law enforcement and antiterrorism efforts also explore issues such as destabilizing crime or terrorist networks and preventing them from reforming.
Considering both the network structure and the activities of its members as outcomes of optimizing behavior provides additional ways to identify and solve the problems facing law enforcement. For example, if criminals understand and expect that enforcement would target the key player, then the network that forms may not be the same network that would arise in the absence of such law enforcement policy. Criminals may choose organizational structures that are the most robust to the particular enforcement environment and strategy. Failing to understand that criminal networks may anticipate the crime fighting policy environment can thwart crime-combating efforts. What may appear to be the best anticrime or counterterrorism strategy planned for a network structure that is observed before the anticrime strategy is adopted, may be substantially different once the criminal organization reacts to the policing strategy. This observation is relevant to both criminal and terrorist organizations.
Empirical Research
Empirical studies have spanned a wide variety of criminal networks related to the market for art and antiquities, pornography, cybercrime, among many others. Since it is already digitized, to the extent that data collection can be automated, identifying large networks in the Internet environment is relatively easy—at least in contrast to gathering data about street-level networks. This has led to a series of studies on Internet crime including phishing, banking fraud, theft of account information, propagation of malware, computer viruses, among others.
Although as of 2018 it still remains relatively rare, some police departments and other law enforcement agencies are using network models to help identify the key players, the most active players, or other analytically significant players to watch or arrest. Describing crime networks in visually suggestive ways has been an important technological advance since it allows investigators to systematize their knowledge and transmit it efficiently. Beyond graphics, continued development of network analytics—computing centrality, density, and key players—and borrowing from graph theory in mathematics provides a richer understanding of the network characteristics to help evaluate and guide policy.
There is substantial complexity in using historical data to predict the outcome of an intervention by law enforcement because large and sophisticated networks of criminals or terrorists may be well aware of the actions of the authorities. If the crime network is taken as predetermined, then any action by the authorities would be a surprise to network participants. Finding the key player, for example, would locate the person whose removal would do the most damage to the preexisting network. Many networks may fall prey to such an analysis. However, if the network itself is an object of choice by the organization, then the criminals can use their knowledge of police strategy to adjust the network structure as a function of the expected policy. Thus, the historical characterization of the network may no longer be relevant if the police strategy is known in advance. A more sophisticated behavioral model that accounts for the criminals’ expected reaction to the new law enforcement policy will be necessary to ensure success. Failing to do so is inherently flawed and can have unintended consequences.
Although applying network theory to crime has progressed over the years, as a systematic crime fighting tool, it still has a way to go. Many current applications are primarily based on descriptions of extant networks. Progress will come with applications that are increasingly able to model the underlying costs and benefits to the participants in criminal organizations and are consequently able to anticipate and take into account their responses to enforcement policy and environmental changes.
References:
- Ballester, C., Calvo-Armengol, A., & Zenou, Y. (2006). Who’s who in networks: Wanted the key player. Econometrica, 74, 1403–1417.
- Burt, R. (1992). Structural holes: The social structure of competition. Cambridge, MA: Harvard University Press.
- Easton, S. T., & Karaivanov, A. (2009, February). Understanding optimal criminal networks. Global Crime, 10 (1 & 2), 41–65.
- McIllwain, J. (1999). Organized crime: A social network approach. Crime, Law and Social Change, 32, 301–323.
- Morselli, C., & Tremblay, P. (2004). Criminal achievement, offender networks and the benefits of low self-control. Criminology, 42, 773–804.
- Xu, J., & Chen, H. (2005). CrimeNet explorer: A framework for criminal network knowledge discovery. ACM Transactions on Information Systems, 23(2), 201–226.
More about Criminal Organizations and Networks: