07.22.2015

A New Approach to AML

Nigel Farmer, @SoftwareAG

Earlier this year, the European Commission adopted its proposal for a Fourth Anti-Money Laundering (AML) Directive. In the words of Věra Jourová, the EU’s Commissioner for Justice, Consumers and Gender Equality, “The new anti-money laundering rules adopted today will help us follow the money and crack down on money laundering and terrorist financing.”

However, in order to actually “follow the money”, banks and other financial service firms need to have flexible systems that can not only be tailored to accommodate changing requirements and regulations, but also to access, process and rapidly analyse data from a variety of sources.

Integrating siloed systems

In his recent blog series discussing the seven pillars of market surveillance, my colleague Theo Hildyard outlined the importance of integrating siloed platforms into a single monitoring system to enable a more correlated view of potential threats. And nowhere is this more essential than in the sphere of AML where, in order to gain a complete picture of how funds are being moved, where, and by whom, data needs to be monitored and analysed across a range of different business units within an enterprise, so that patterns corresponding to potential nefarious activity can be detected. Looking at data from individual siloed systems, rather than having an integrated view, will not achieve this.

Detecting money laundering patterns

One of the key components of any successful AML system is the ability to match patterns. And those patterns can consist of many elements, which when looked at together over periods of time, can start to paint a clear picture of where AML officers should focus their investigations. Typically, these elements include (but are by no means limited to):

  • High value transactions – those that approach or exceed a pre-set threshold, or where total transaction amounts approach or exceed daily limits
  • High balance accounts – where expected wealth levels are exceeded
  • Flow through funds – e.g. where money moves in and out of accounts very quickly
  • Over-payment or early repayment of loans
  • Accounts being opened and closed in a short time frame
  • Transnational Risk – transactions to or from high risk countries, institutions or individuals
  • Suspicious trading activities – e.g. frequent changes to settlement instructions and other static data, overriding of default settlement instructions, loss-making investments, cross-asset transfers
  • Unusual activities – deviation from historical behaviour or from peer group

The time element
Pulling all of this data together from across multiple sources to enable comprehensive analysis is just one half of the AML equation. Equally important is the ability to combine historical data with real-time streaming data, in order to uncover recognisable patterns as they occur.

One way of doing this very effectively is by caching large amounts of historical data in memory and using an event-based processing language to query that data, so that users can look for correlated patterns over days, weeks or even months to uncover ways that people try to hide illegal proceeds.

At Software AG, this is an area where we have been working closely with clients recently, using the combination of Apama and Terracotta providing respectively, streaming analytics and in-memory data management, to detect patterns across longer time windows. By providing the ability to cache large historical data sets from multiple sources, access that data on the fly and correlate that historical data with incoming events, this technology combination enables money-laundering patterns to be detected much more clearly.

Enhanced accuracy, functionality and predictive capability
This approach can provide a new degree of accuracy when performing AML tasks. Particularly in areas such as the fuzzy matching of names, addresses and account numbers to detect hidden linking of accounts, transactions and relevant documents for example. And with the appropriate relationship mapping functionality, firms can set rules and can profile relevant behaviours at individual, account, group and organization levels. Not only that, but new advances in predictive analytics can determine when and how funds might move next, so that firms can set – and more importantly, act upon – appropriate alerts. This is made possible through the integration of tools like the Zementis predictive analytics tooling into the streaming analytics engine.

Conclusion
AML in today’s world of complex, big, fast moving data is an ever-evolving challenge. But with the right technologies in place, firms can stay ahead of the curve to ensure that they not only comply with regulatory authorities’ AML requirements and report after the fact, but also actually detect and pre-empt money laundering from occurring in the first place.

Related articles