With the growth of illicit money being funneled across Asia Pacific, FSI companies in Hong Kong are now seeking ways to fine-tune their anti-money laundering (AML) systems to enhance their capability to detect suspicious transactions, according to Jack Jia, partner at Ernst & Young’s Fraud Investigation & Disputed Services Asia Pacific.
“Banks in Hong Kong are aware that they urgently need to resolve the AML clearing issue. A bank can have millions of transactions, and what it needs to do is run a system through to screen high-risk transactions that are potentially related to money laundering,” he explained.
The system usually triggers thousands of alerts that require closer inspection by bank personnel.
“Banks want to fine-tune their system so there are less alerts coming out, but every time one comes out it is a valid one. Usually, banks have a small team and it is difficult to clear tens of thousands of alerts and there is not enough time to clear all of them,” Jia said, adding that banks want to reduce the number of false positives.
The city’s financial firms are under pressure to make improvements in their AML systems. Last year, the Securities and Future Commission (SFC) tighten regulatory controls to combat illicit flow of funds.
The SFC inspected AML controls at major investment banks, brokers, and asset managers, finding problems with the way some firms assessed and reported suspicious transactions.
In its latest annual report, the regulatory body found a 253% increase in the number of cases of non-compliance with AML guidelines between 2013/14 (88) and 2015/16 (223). In addition, SFC fines totaled US$11.2 million in 2015/16, a 58% increase over the previous period.
Using analytics offers a proactive approach to AML
Seeing a gap in the market, Ernst & Young started offering in Hong Kong its AML Analytics to help financial firms to better comply with the stricter AML guidelines issued by local regulators.
“I just want to emphasize that the idea of fine-tuning AML systems is relatively new. I mean, it is happening in the US, but over here, people have just started thinking of deploying the system after high-profile local money laundering cases came out,” Jia said.
Data analytics is a critical component in re-tooling AML systems for better results, and it can be used as a proactive approach to solving the FSI’s sector AML clearing issue, Jia pointed out.
“We proactively go in and set up the system to manage the alerts better. It is like building a good foundation. We fine-tune the design of the system by setting a framework and imposing policies and procedures on how to identify fraud.”
He added: “It involves mathematics, visualization of the data, and involves the human element of understanding the AML issues. It involves a team of multiple disciplines and expertise to solve this problem.”
Self-learning is critical part of the enhanced AML system, Jia said
“If a bank keeps a record of transaction behaviors that indicate fraud, it could leverage that learning and design algorithms to pick up those patterns to prevent them from happening again.
“That learning from the past is definitely a very important component of designing system. Well, it is human element and some self- learning. The self-learning part can be designed into a workflow to allow the results of your investigations be feedback into the system so they can learn, so that is the automation part. But the human aspect is fine-tuning that learning process.
Ideally, Jia said the detection rate of true AML case is between 0.1% and 10% out of 100 cases. With a fine-tuned system, Ernst & Young hopes to increase the true detection rate up to 20%.