Subscribe / Unsubscribe Enewsletters | Login | Register

Pencil Banner

Outsmarting money launderers with data science

Scott Zoldi, Chief Analytics Officer, FICO | Oct. 3, 2016
With money laundering becoming not only an economic issue, but also a political one, it is imperative to fight against money laundering intelligently and effectively.

Min Zhu, Deputy Managing Director of the IMF, was absolutely right when he called action to prevent terrorist financing and money laundering both an 'economic need' and a 'moral imperative.' So, what more can be done to tackle money laundering?

AML detection and prevention needs to be smarter, quicker, and more efficient. It's not just about reducing costs or showing compliance, but ensuring that the right investigative resources are focused on the right problems. Rules-based systems focused on compliance alone are no longer sufficient. Criminals move fast, so those working to stop them must too. Thankfully, new technologies are emerging that can make the fight against money laundering more effective, if deployed efficiently. In Asia, financial services companies are expected to spend a record amount of $300 million (approximately S$420 million) on technology and infrastructure in 2017, as they boost efforts to banish money-laundering. 

Analysing customer behaviour

One such technology is self-learning analytics, which offer an important solution to one of the biggest challenges banks face: discerning a customer's true intentions.

A lot of legacy AML systems are built around KYC, which segments customers into different risk categories and applies rules in order to be compliant with regulatory AML requirements. However, relying on a rules-based approach makes it too easy for criminals to fake their way through the KYC process. For a bank, this means that what it knows about its customer may not be the full story, despite the fact that it has demonstrated AML compliance. In contrast, because self-learning analytics evaluate constantly evolving transaction data, they can pick up instances a rules-based system is likely to miss.

Continuous analysis of a customer's data not only ensures they have been correctly categorised, but that any future changes in behaviour are detected and can be evaluated. For example, if Customer 1 is initially placed in Segment A and subsequently continues to behave similarly to other Segment A customers, the bank can be confident that they have been grouped appropriately. If, however, Customer 1's behaviour begins to change, machine learning can detect this, and may recommend that Customer 1 be treated with enhanced customer due diligence. This process, which runs automatically, helps banks to act quickly when they are alerted to the true intentions of a customer.

Assessing the threat level

Another important technological development is the use of intelligent, automatic scoring to rank customers on behavioural risk to facilitate AML diligence.

Rules-based systems flag any behaviour previously defined as "suspicious", so AML teams end up buried under a pile of suspicion reports without the guidance needed to prioritize their evaluations.  Scoring systems rank-order threats by giving each a number based on its level of severity. This allows banks to focus their efforts on the most significant threats, meaning they can not only work smarter, but more quickly and efficiently too.


Previous Page  1  2  3  Next Page 

Sign up for CIO Asia eNewsletters.