Blog Assignment: Implementing SNA in fraud detection and prevention in banking business

Social network is critical to business success, especially banking business too. Social network analysis (SNA) enables banks to exploit the structure of customers, as it not only reveals insight into finding targeted customers, but also locates fraud behaviors.

For example, when customers are applying for loans, banks can analyze the loan applicants’ social networks to see if there are contacts relevant or guaranteed to the new business. Then, banks can combine that with an analysis of customers’ credit card transactions to take a look at business interactions relevant to the new business. Even more, value network analysis can help banks to understand their business partners, to see who has met performance commitments and has contributed a valuable role in the network.

Problem:
For most banking business, profit loss and law issues have become the common reasons for banks to invest in fraud detection and prevention. Financial crimes have evolved in more complex effort chains to attack on businesses, thus, banks and financial organizations are working together to detect and prevent financial crimes from occurring. Identity network fraud, anti-money laundering, and denial of terrorist financing are the main areas of fraud detection and prevention where SNA could be applied for banks to improve.

Solution:
Implementing SNA for fraud detection and prevention, observers would be able to detect transaction patterns within and across financial businesses as a potential crime circle sign, saving banks’ individual or corporate customers from financial losses as the crime further develops. This is where the SNA’s visual and analytical capabilities can help in fraud detection and prevention, by analyzing both traditional channeled and web-based transactions. Banks need to better understand the dynamics of fraud networks.

The overall question that the SNA approach can address would be asking customers about the purposes of their transactions and loans. Then banks could monitor and compare the actual transactions flow.

However, a lot of what banks are doing has been behind-the-scenes analysis. The data can be found after linking all banks’ transaction records, along with sharing customers’ identification database. Within law’s permissions, collecting these data is not a difficult issue. Then, the SNA approach focuses on the patterns of relationships between/among individual and corporate customers, as shown in the following flow structures:

 

 

 

Measurements:
Density measures are significantly useful in detecting potential fraud hotspots between banks from a maze of account transactions and applied control measures. Credit card transaction monitoring and money-laundering are typically two areas where density measures would trigger the necessity for the deep investigations.

Tools:
Desktop applications like UCINET can be used for small scale fraud hotspots detections. Enterprise-level systems like SAS Fraud Framework which contains an SNA server can help banks implement end-to-end SNA based fraud prevention.

Challenge:
· SNA approach is inherently retrospective. For example, customer can only react to a fraudulent incident after it has occurred.
· SNA approach is also affected by different regulations in various countries under the influence of globalization. Data protection laws etc., could weaken the effect of implementing SNA, and could even render investigations futile when they confront with authorities. Thus, SNA investigation workflow processes should be checked for cross-border regulatory conformity.
· While banks would like to tie social network users to real transaction customers, that direct connection is actually out of reach. For example, linking online personas to bank customer records is a mission impossible, mostly due to the anonymous nature of the typical social network.

- by Yuyang Cai