Laura Hutton, Executive Director at Quantexa
Financial crime is a wide-reaching and prolific issue that banks are struggling to tackle. Laundered money is known to be funding illegal activities, including terrorism, which places banks under immense pressure to identify the source of such funds. All banks have Anti-Money Laundering (AML) systems in place but they are crippled by a variety of different inefficiencies that are allowing criminal activity to remain undetected. In many cases, organised criminals are systemically probing the various weaknesses within these AML systems and are actively capitalising on them in order to turn the profits of crime into ostensibly legitimate assets.
At first, financial institutions put systems in place to detect money laundering within their retail book, with other product lines left unprotected. Under increasing regulatory pressures to change this, banks hastily repurposed these systems into other areas of the business. Inevitably, these have not been suited to the product line that they are trying to protect, with systems struggling to cope with the complexity of the schemes perpetrated by organized criminals. With access to only a fraction of the available data needed, they use simplistic analysis methods, looking only at typologies dictated by the regulator.
Perhaps the biggest impact of this is that systems are ineffective, producing large numbers of low-quality alerts. Essentially, they detect an enormous amount of seemingly anomalous behavior that, when investigated, is found to be unsuspicious. These systems are providing levels of false positives that are in excess of 99%, and this is where the inefficiencies arise.
The problem is that under compliance legislation, banks are legally obliged to investigate every alert created by their AML system, regardless of how unlikely they may seem. This leads analysts to distrust the detection systems. Often the investigation process is outsourced to lower cost resources, employed to do basic box-checking processes, thus removing the financial crime knowledge from the investigation process. This is something that criminals have been able to actively capitalize on, and as such, is becoming an absolute priority for banks in terms of improving their systems.
The use of basic analytics and limited data has prevented current systems from being able to make the sort of judgements required to identify well-hidden activity. The data sets that are available to these banks are enormous and, when utilized effectively, can provide useful insight into illegal activity. Yet, this data is not yet made use of – analysis is conducted across a thin slice of data, solely at the transactional or account level, leaving masses of potentially incriminating data uninterpreted.Systems look for simple patterns like large cash deposits in short time periods. Money launderers will intentionally hide illegal activities within businesses that have a lot of international transactions going on all the time. The analytics in current systems are simply not sophisticated to detect activities that have been deliberately concealed.
Aside from their various issues of inefficiency, repurposed AML systems are inflexible and hard to adaptunable to support the addition of new data points, typologies and products. This means they are difficult to improve, withreconfiguration requiring laborious and time-consuming effort. The inflexibility of these systems leads to an eventual and inevitable degradation of model performance.
In order to combat these issues, top tier banks are embracing new technologies that can improve efficiency and effectiveness through use of intelligent analytics and network based solutions.
A new approach
Money launderers are ultimately people and companies, not transactions and accounts. Banks have thus been able to successfully combat the problems discussed by employing entity resolution and network analysis techniques.These advanced analytical processes are able to interpret vast data sets, contextualizing seemingly isolated incidents and relationships within wider networks which ultimately helps to unearth cases of intentionally hidden money.
Connections can be extracted from internal and external data, derived from information such as names, contact information, company structures and transactional money flows.These newer analytics are supported by AI and machine learning and are therefore able to recognize patterns amongst confirmed cases of criminal activity and use them to inform interpretations in the future. These new AML systems are proving to be far easier to rework and are therefore far more reactive to additions of new data points or typologies. This has allowed leading banks to employ AML solutions that are personalized and adaptable to their individual needs.
AML systems in place across the banking sector are outdated and unfit for purpose. An overhaul of these systems is the only way that banks can hope to effectively tackle the issue of laundering. The current political climate is putting more pressure on banks to find out how nefarious activities are being funded and how it is so often going undetected. High levels of erroneously positive alerts are inundating analysts with cases they are legally obliged to investigate which is eating away at the time they can spend investigating the cases that matter. Similarly, as political pressures mount, money launderers are becoming increasingly intelligent in how they work and the systems in place are simply not sophisticated enough to keep up with them. Banks that have taken a proactive stance to these glaring issues, embracing a more holistic and flexible approach towards anti-money laundering,are finding it is the obvious solution to a dangerous problem.