Blog post written by Ido Nir, Oracle Financial Services Analytical Applications, Financial Crime and Compliance Management - Asia Pacific.
Enhanced enforcement to be carried out in light of upcoming Mutual Evaluation by the Financial Action Task Force (FATF) planned for October/November 2019.
Tokyo-based financial institutions have established a longstanding ecosystem to address anti-money laundering. This ecosystem has served for many years as the standard dictating operational best practices across Japan, as well as the standard for customer due diligence procedures and the detection of suspicious activity. In recent years, we have witnessed constant change in the direction of global regulation, focusing mostly on proper KYC processes, identification of the ultimate beneficiary, Trade Based Money Laundering (TBML), and other areas. As a result, we see a distinct shift in the focus of global banks and a re-evaluation of best practices for transaction monitoring, customer due diligence, and screening.
The Japanese government has acknowledged this shift and has recently introduced changes to the Act on Prevention of Transfer of Criminal Proceeds (effective October 2016), which set new standards for both banks and other financial institutions operating in Japan for customer on-boarding and monitoring (KYC, CDD, and ECDD), the verification of foreign correspondent bank relationships, monitoring and detection of suspicious transactions, as well as implementation of additional in-house AML measures.
In previous years, the FATF has criticized the Japanese government for inadequately addressing AML issues. The last official statement was given in 2014 where the FATF contemplated adding Japan to its "grey list" of high-risk jurisdictions. The 2016 amendments take Japanese AML legislation one step closer to fully addressing the FATF's criticisms and establishing a more robust AML regime as seen in other developed nations.
As a direct result of the new legislation, and in light of the upcoming FATF mutual evaluation planned for October or November 2019, the Financial Services Agency (FSA) began a nationwide evaluation of multiple banks, securities firms, and other financial institutions, surveying the measures they are taking to prevent money laundering. The aim of this survey is to better tackle the growing problem of money laundering and financial crime among both global and regional financial institutions.
At each of the financial institutions surveyed, the FSA is examining the precise steps being taken to prevent money laundering across a multitude of risk areas, and whether these policies have been properly adopted by individual branches and subsidiaries. Furthermore, bank-wide knowledge and understanding of AML as well as management's understanding of related risks and exposure is also being evaluated.
It is clear that the FSA will continue its effort to bring proper AML controls to the forefront and will vigorously require both global and regional financial institutions to adopt proper measures.
Financial institutions are expected to update their current AML measures and policies to ensure compliance with the recently updated regulatory framework. This change in policies, operations, and procedures should be applied with minimal impact to the existing business process - ideally it should enhance the existing process. Financial institutions should set a goal to establish a holistic compliance program that handles the identification of customers, screening of customers and transactions, as well as monitoring of suspicious activity in a cohesive, effective and most importantly in a unified way, with minimal impact to business processes.
By arming themselves with the right tools and technology, Japanese financial institutions can ensure they implement the level of scrutiny that regulators are demanding today, as well as ensuring their ability to meet both the business and regulatory challenges of tomorrow.
Financial institutions must comply with and observe the changes in a set of standards imposed by regulators and, as indicated above, they are regularly audited to verify their compliance. Financial institutions who are found in breach of their duties face legal consequences as severe as being stripped of their banking license and the threat of hefty fines rising potentially into the hundreds of millions of dollars. Good examples of this include the case of 1Malaysia Development Berhad and the case of Deutsche Bank's USD 41M fine for AML lapse imposed as part of their May 2017 settlement with the U.S Federal Reserve. There are numerous other examples where fines reach more than USD 1B, which clearly reflects the regulators' intentions.
To avoid hefty fines, financial institutions must look at the banking products they offer, the markets in which they operate, and the regulations that apply in those markets to further understand their risks. Then they must implement controls and solutions, such as detection scenarios. Many financial institutions have followed a "rule-based technology approach." This approach has worked well for many years, but while it will trigger alerts that catch bad guys, it may also result in false positive detection of legitimate activities. False positive detection is a known problem across the industry that routinely requires analysts to spend time reviewing more alerts, incurs a larger workload and commensurate costs on AML operations.
Machine learning algorithms are promising tools for reducing such false positives. Algorithms can be developed using training data, and then customer-specific data to be fine-tuned, resulting in higher detection accuracy and increased performance. However, when it comes to compliance, showing results is not enough for regulators. Banks are required and must be able to explain how they arrived at their detection results.
This is one of the key challenges with models trained using Machine Learning techniques, the advanced and more sophisticated algorithms are essentially a black-box, and the inability to explain the algorithm in a black-box has been a major roadblock for industry adoption of this technology. However, the stakes are too high to give up on Machine Learning.