Money Launderers Know How to Network
Hidden Relationships and Networks: Financial Institutions at Risk
Link: http://www.oracle.com/us/industries/financial-services/045950.pdf
In class, our discussion of networks has relied on a few key assumptions: the nodes are clearly defined, the relationships (edges) are known or reasonably accessible, and a few basic theories can be applied to our analysis of a given network (mainly the idea of triadic closure and the concept of balanced versus unbalanced networks thus far). However, what happens to the analysis of a network when the nodes are ambiguous and the relationships are not known, or in fact intentionally hidden? In the Oracle White Paper referenced above, the authors provide an insightful investigation of money laundering networks and display some of the complications that arise in the real-world application of network theory.
Money laundering is intrinsically dependent on networks. In fact, it is the creation of highly complex, seemingly nonexistent networks that makes it so difficult for enforcement agencies to uncover money-laundering rings. And in fact, one of the leading organizations in the pursuit of money-laundering criminals, the Financial Action Task force on Money Laundering (FATF), openly admits that an approximation of the total value of laundered funds is impossible to date. So what makes it so difficult to track down these criminals?
The network of accounts, holding companies, aliases and compliant and non-compliant parties that are involved in these typical money-laundering schemes create extremely tangled pathways and circular loops that significantly complicate the pursuit of fraudulent sources. Imagine that a local bank identifies an account that appears to be receiving suspiciously large funds, perhaps at irregular frequencies or without clear line of sight. The natural next step for law enforcement would be to uncover and validate the source of the money. If it were my personal bank account in question, the investigation would be easy. They would contact me, review my personal income and other gains with the IRS, as well as identify any additional accounts under my name and, with a little algebra, determine if the numbers align. During the investigation of a money-laundering scheme, officials may hypothetically find that the suspicious funds were transferred from a different account at the same bank, which was transferred from an account at a different bank, which was held by a company located across the country whose profit-making activities are unclear, which is owned by an individual with no traceable identity. So as you might imagine, the task that lays ahead for our anti-money-laundering agents is to create a network based on misleading nodes and unclear edges that vary across several dimensions. While in our class, we analyze two-dimensional networks (nodes connected by edges that all represent a given characteristic – i.e. a network of people (nodes) and their friendships (edges)), you can see how quickly these money-laundering networks become three-dimensional, extending through networks that are connected with edges that represent different characteristics. On top of the complexity of the networks, the relationships are generally hidden or falsely represented – it may be extremely difficult to establish the relationships. In the Oracle White Paper, the authors give the example of six accounts at one specific bank. Investigators might find that account one shares a phone number with account two, but no other indentifying information. Account two then shares a home address with account three, and so on with all the additional accounts and a continuum of different characteristics. In this way, the network of each characteristic and the links between those networks must be established.
The financial services software company, Oracle, has taken a leading role in offering products to help financial institutions begin to spot trends and build networks between fraudulent accounts. A few examples of the software techniques that Oracle uses include: Link Analysis and Sequence Matching. Through Link Analysis, the Oracle product compiles all hidden links (or normally unnoticed links) between accounts and builds networks of all interrelated accounts. This allows investigators to analyze accounts as group, a network, which is often helpful in unmasking the sources and uses of funds. Through Sequence Matching, the Oracle software refers to a database of commonly used, sequential transactions; it identifies and marks them. Again, the software compiles networks and seeks to identify nodes that are involved in multiple, high-risk sequences.