Now’s the Time for Fraud and AML Convergence
January 8th, 2021
Financial institutions maintain two large and complex operational regimes dedicated to fighting financial crime. Both are laden with technology operated by armies of professionals dedicated to rooting out criminal activity. Why not combine them and reap the benefits in increased efficiency, lowered costs, and improved analytical insights?
Convergence of fraud and anti-money laundering (FRAML) technology and operations has been somewhat of a Holy Grail for financial institutions for at least a decade. While the concept has its promoters and detractors, FRAML makes sense to many in the industry. A number of technology providers have offered FRAML features in their financial crime solutions for some time already.
Now is a good time to take stock of where we are with FRAML. Regulators have come to understand that fraud and money laundering are intrinsically connected and are starting to ask financial institutions to coordinate their fraud, AML, and even cybersecurity efforts. The pandemic year has seen significant increases in fraud volumes and a proliferation of new laundering techniques. This has all played out against a backdrop of increasing technological sophistication on the part of fraudsters and money launderers to exploit the criminal opportunities that abound in digital financial services channels.
A Fraud and AML Approach
FRAML approaches do have traction. Some financial institutions, particularly in Europe, moved the management of their fraud, AML, and security operations into consolidated financial crime divisions years ago, although this has not translated into technology or operational convergence. Many firms have enterprise-level dashboard reporting for centralized tracking of fraud and money laundering risk indicators. Various technology providers support integrated risk scoring and management of AML and fraud alerts to facilitate enhanced detection of criminal activity; and can ingest data feeds from multiple AML systems for this purpose.
Fraud and AML convergence at the operational level is a tougher nut to crack. There are a number of barriers to FRAML in terms of organizational, operational, and technology issues. First, fraud and AML are fundamentally different beasts. Fraud is traditionally a line of business problem, since fraud losses are direct hits to the bottom line. AML, on the other hand, is a compliance area subject to a strict regulatory regime and well-developed procedures. As a result, fraud and AML have developed as separate functions within financial institutions, with siloed operations, technology, staff, and data.
Fraud and AML Operational Requirements
There are deep differences in operational requirements for fraud and AML, with implications for technology. Fraud is all about preventing leakage of funds in real time, before the money goes out the door — or out of the country. Firms are also in a position to verify when fraud has occurred because they own all the relevant data and can use this information to develop and refine analytic models to detect future frauds. The existence of this feedback loop led to the development of artificial intelligence solutions for anti-fraud as early as the 1990s.
The urgency of stopping fraud in its tracks has also led to a good deal of information sharing between financial institutions to identify fraudsters when they walk in the door or come over the digital transom. Lists of fraudsters and other information are compiled by various industry consortia and also provided as commercial offerings by technology and data providers.
AML, by contrast, has traditionally been about looking at customer and transactional activity retrospectively and over time to detect behavior that may indicate money laundering (“suspicious activity”). This information is then packaged up into regulatory reports such as SARs and sent to financial investigation agencies to support government efforts to combat money laundering and terrorist funding. AML monitoring software, as a result of its retrospective nature and a comparative lack of operational urgency, typically runs in batch mode.
Moreover, because they are not privy to information generated by government investigations, banks have lacked the data to power feedback analytics. As a result, AML has been reliant on using rule sets to detect known typologies (“scenarios”); and AI has been a late entrant to the AML toolkit. The sharing of information on money launderers between financial institutions is also limited and subject to regulatory constraints, a stark contrast to the information tools available on the fraud side.
Despite these existential differences between fraud and AML operations, there are areas of commonality. While behavior analysis for AML has been a batch processing affair, real-time analysis is a necessity for fulfilling regulatory obligations around sanctions — in order to prevent wire transfers or other payments from being sent by or to sanctioned entities.
Digital financial services are also broadening the overlap between fraud and AML. The instantaneous nature of digital transactions is leading firms to reconfigure their AML monitoring on a real-time footing. Improvements in advanced analytics are enabling the successful application of AI to the AML domain. The regulatory need to run KYC checks on their digital customers gives financial institutions the opportunity to consolidate their ID verification, fraud, and KYC checks into one seamless process. Incidentally (!), this is also leading to the full automation of formerly very manual KYC processes.
The combined effects of an increasing regulatory emphasis on coordinated fraud and AML investigation, the operational implications of digital financial services, and advances in AI and analytic technology are all, well, converging, to move FRAML forward. There has never been a better time for financial institutions to take a good look at their fraud and AML operations with an eye to increasing efficiencies, enhancing analytic insights, reducing costs, and catching more bad guys.