ACIP: Industry Perspectives – Adopting Data Analytics Methods for AML/CFT
recently best practices on the use of data analytics for AML/CFT in financial institutions. The
ACIP set up the Data Analytics Working Group for member banks to share their respective journeys of data analytics adoption and implementation in AML/CFT, and to provide practical insights for financial institutions at different stages. The working group is chaired by the Head of Group Legal, Compliance and Secretariat, DBS, and members from commercial banks operating in Singapore and other organisations including: Exiger, Delta Capita, Fircosoft, IBM, and SAS.
A key AML/CFT measure is the screening of customer names, along with the connected parties of customers and counterparties to transactions, against money laundering and terrorism financing information sources - including sanctions watchlists, politically exposed persons lists, and adverse news sources. Data analytics can be used to improve these screening and name matching capabilities. One of the more common applications is fuzzy logic, where algorithmic processes are used to detect and evaluate near matches. Natural language processing technology is another, and is used to screen against mulitple adverse media sources, often involving the analysis of unstructured data. These use cases have the potential to improve the accuracy of matching, thereby reducing false positives and increasing detection rates.
More advanced data analytics solutions can also be used for the automatic disposition of false hits, together with explanation of how the hits are disposed. This enables institutions to reduce false positives and have a level of explainability in their use of analytics. (page 11)
Use Cases – 4 Overarching Types:
As demonstrated, there is a wide range of use cases for data analytics. Some may be more suitable for advanced institutions, while other simpler solutions may be appropriate to consider when institutions start their analytics journeys. In each case, this will depend on a number of factors, including the individual institution's broader strategy, existing in-house skillsets, risk profile, and appetite for the use of AI or automation.
The [Working Group] highlights that the above organisation of data analytics use cases by themes is not the only useful approach to thinking about the available technology.
Exiger instead suggests categorising technology into 4 overarching types:
- Feature Creation:
Applications that define features for modelling
- Data Transformation:
Applications that facilitate the execution of data assessment or transformation
- Decision Analytics
Applications that allow you to replicate or model decisions
- Witness Models
Applications that monitor for data or model issues
Most of the use cases described above may utilize more than one of these types of technology. (page 13)
Want to read more?
Download the full paper here >
You may also be interested in: