In a volatile AML-CTF landscape it is critical to improve the speed-to-compliance. However, achieving a faster response can be self-defeating if it introduces compromises to compliance effectiveness. Simply making existing processes faster might yield short-term tactical gains but can lead to a primary focus on supporting the status quo. This approach can lead to a lesser resource capacity for continuous compliance improvement. Innovation in screening programmes must therefore also consider how the standards of compliance effective can be raised in the short-term whilst also delivering capability and capacity that is ready for future challenges.
Screening capabilities for all target types
Compliance screening used to be a relatively binary process: governments issued lists of sanctioned geographies, persons, or entities; these lists were then screened, with the outcome of a determination if a target matched against a client record or transaction. Today the task is more complex. Screening requirements now include a broader range of risk types, for example, beneficial owners, persons of significant ownership, family members and professional associates, or even certain capital markets instruments, financial services, or manufactured goods. Furthermore, not all sanctions targets are cited on tangible lists or even pseudo-lists. This reality requires more investment in understanding screening data requirements and for the procurement of the right data. Institutions must also ensure that screening technology has a comprehensive technical capability to screen the full spectrum of risk, regardless of type. To be successful, appropriate screening methodologies must be implemented and maintained for each type of sanctions target.
Matching techniques for diverse risks
Expansive and complex sanctions requirements require a continuous review of the rules used to determine potential correspondence to risk. For example, would an alert be triggered for a sanctioned company from Russia if a Ukrainian variant of that name was transcribed phonetically for use in Germany? Phonetic name matching has been developed into many screening tools. However, achieving a deep understanding of how matching technology works – so that rules can be defined and maintained properly – is difficult if the “explainability” of matching algorithms is not a fundamental design feature.
Additional false negatives countermeasures
False positives reduction, or optimisation, is a priority for many institutions managing the inevitable consequences of screening massive client or transaction data volumes. Much progress has been made in these scenarios to deploy robotic automation or machine learning to investigate, categorise and route alerts quickly and efficiently. This push to increase operational efficiency has clear compliance benefit of releasing resources for more qualitative tasks. However, institutions could also consider implementing a second layer of controls to reduce the risk of false negatives. For example, re-screening data that did not trigger alerts in a primary screening process could identify undetected risks, notably when newer technology is used in a secondary process.
Responsive impact assessments
The intensity and velocity of AML-CTF screening – and the high cost of a compliance failure – has created operational environments that have little time to model the impact of new requirements and then to configure accordingly. This is often an issue when there is a dependency on older, less nimble technology that is difficult or time-consuming to test new rules quickly. The lack of a timely impact assessment can then create downstream operational issues. Moving to towards more responsive testing and modelling in can mitigate these issues and ensure a better focus on managing risk.
Beware – Technology Debt
In summary, technology should always be an enabler of compliance effectiveness, never an inhibitor. The focus of screening technology is, inevitably, on short-term imperatives. However, the impact of technology debt on long-term compliance should not be underestimated.
