Financial crime: Four essential rules for effective transaction monitoring
Money launderers are, by definition, a devious bunch. They quickly learn the rules that banks set to detect suspicious transactions, and adapt. They are, in other words, dynamic – so transaction monitoring should be equally so.
In our previous blog we talked about the problems with many legacy transaction monitoring (TM) systems, which tend to be rules-based. That’s not to say that rules-based systems are obsolete; but in the complex world of global finance, they’re not enough on their own. If banks are to efficiently root out complex illegal trades they need to know far more about the connections between parties to transactions. Fortunately, advanced technology and analytics allow us to do this.
Advanced TM systems should apply a combination of four elements:
Rules. The application of rules centred around customer segmentation and threshold criteria are and will always be an essential part of TM, but new technology resolves some of their limitations. Machine learning, for example, makes sure that the thresholds that are set are always appropriate and consistently optimised; rules are updated as the systems learn. Assuming the data is available, technology gives banks the power to manage many more meaningful segments.
Network or link analysis. This examines the key characteristics of parties to transactions, looking at beneficial ownership (often using third party data sources) to identify relationships beyond those that are direct and immediately apparent. This allows banks to better understand the connections and interactions between parties and could, for example, help to identify the relationships between companies involved in money laundering schemes.
Behavioural analytics. This allows banks to monitor segmented customers constantly for anomalous behaviour and, assuming the data is captured, also allows for segmentation at a more granular level (for example, by comparing the behaviour of similar businesses in a certain location). Segmentation itself is a blunt tool; it’s not what a customer is that’s important, it’s how they behave (especially when compared to their peers, or against their own past behaviour). A nail bar, for example, that makes bulk purchases of pre-paid credit cards – something that nail bars in the same region don’t normally do – could be an indication of people trafficking activity.
Adaptive machine learning. Artificial intelligence is the next important step in TM; machine learning feeds back what the system has learned (for example, if a particular transaction was flagged but then turns out to be innocent) and uses that knowledge to refine its analysis going forward.
Each of these elements could work together to dramatically increase the power of TM. We’re no longer at a stage where thousands of transactions are flagged and each has to be manually investigated – these four elements together allow banks to understand the transactional activity of customers over time compared to their peers, in a more nuanced and intelligent way than simple heuristic approaches. The system can spot when information has been manipulated to open multiple accounts, and it can even join together poor quality data on a single customer to create a more complete picture. This is exactly the type of ammunition banks need to combat modern-day financial crime.
If you would like to discuss these issues, or the impact of emerging technology or data and analytics on your industry, then contact our Data & Analytics team.
