Transaction Monitoring: combatting Financial Crime

According to United Nations Office on Drugs and Crime (UNODC), money laundering amounts to 2-5% of the world’s GDP per year, this is in the range of £600 billion to £1.6 trillion annually. The National Crime Agency (NCA) report confirms that money laundering costs the UK over £100 billion a year. Despite law enforcement’s efforts, they still face enormous challenges, and therefore tackling this issue is a top priority for Financial Institutions (FIs), government agencies and regulators.

The Dutch bank ING was fined €775 million (£680 million) in September 2018 after admitting they failed to notice unusual transactions and that criminals had laundered money through the bank’s accounts. Other firms such as Danske Bank and Nordea have faced similar challenges. $4.8 billion in penalties were paid for financial crime irregularities in 2018. The UK FCA has fined £102.2 million to Standard Chartered Bank this year for not having adequate AML controls, which is on the back of the penalties the bank faced for facilitating payments to Iran and breaching international sanctions, and UBS was fined £27.6 million for failing to report over 135.8 million transactions. A record $7.7 billion in international penalties were paid for AML irregularities between January and April 2019.

Transaction monitoring is an area of specific risk, often seen with inadequate systems and controls.

There are challenges with existing transaction monitoring systems within FIs. These include legacy systems and infrastructure, bad data, rigid rules-based systems and high level of false positive resulting in the need for additional resources and effort. Investigating individual transaction is not enough in understanding the patterns related to complex transactions. FIs often do not have a clear procedure that details how alerts of suspicious transactions should be identified and often it is quite difficult to source information on the customers as some transactions do not require them to enter enough details to potentially identify suspicious activity.

We have seen that Artificial Intelligence (AI) and more specifically Machine Learning (ML) algorithms integrated into transaction monitoring systems can help FIs to investigate through patterns of transactions, help reduce false positives and enable analysts to focus their efforts on investigating the truly suspicious activities. There is also an opportunity for law enforcement to apply these technologies to assist with their investigations into SARs/STRs which have been submitted. There are many examples of banks conducting proofs of concept (POCs) with AI/ML in relation to transaction monitoring. Recently RegTech solutions have been piloted by OCBC Bank and Standard Chartered in Singapore. Some of the practical applications of Cryptography (the method of writing or solving codes) in the FCA tech sprint this year confirmed how the latest technologies such as the use of MPC protocols, a subfield of Cryptography, can be used to analyse complicated networks and transactions.

Additionally; Natural Language Processing (NLP) can be deployed to manage unstructured data within the process. AI and ML solutions can be used to search and collate information on individuals in the wider network. The unstructured data can then be analysed in a more structured format, which in turn can reveal the details of the parties involved in the end-to-end transactions.

With the emergence of these new technologies it is extremely important that staff are trained to understand the results produced alongside clear and maintained policies and procedures. Clearly an upskilling of staff is required including those with responsibility to explain the processes to the regulator. Understanding, explaining and documenting what the algorithms are doing is a vital part of the design of any system that is implemented.

As Elbert Hubbard said “One machine can do the work of fifty ordinary men. No machine can do the work of one extraordinary man”.

How Lysis Can Help: