Future Trends in Sanctions
Can automation, artificial intelligence and outsourcing resolve inefficiencies?
Financial institutions are struggling under the challenges brought by today’s sanctions regimes. Could these inefficiencies be resolved by developments such as automation, artificial intelligence and outsourcing? A panel at Sibos 2017 discussed.
Compliance at Sibos - Future Trends in Sanctions
- Challenges include pulling together the necessary data to accurately process alerts in a consistent and efficient way.
- Possible opportunities to use technologies such as analytics and machine learning to improve sanctions screening.
- Potential barriers to adoption include the need for regulatory acceptance and concerns about relying on machines for decision making.
Challenges and inefficiencies
- Multiple systems. Larger banks may have ten or more different systems that are “attacking the same problem.” It can be challenging to understand, consolidate and create consistency.
- Data. It is “messy” trying to pull together the names, addresses and entity names needed to do a clean match against the relevant lists.
- List management. Large global organisations struggle to understand the different lists and how to manage updates.
- Technology. Having the technology and workflow needed to process alerts in a consistent way.
Panellists pointed out that in the case of a single cross-border transaction, several different banks will screen the same payment – so efficiency is clearly lacking.
Lorraine Lawlor argued that it is “naïve” to think that a sanctioned party would use their real name in a transaction. As a result, “the field itself needs to rethink” how customers and transactions are screened. Screening is moving towards an AML model, with a focus on understanding customers, who those customers are dealing with and the associated risks. The panel highlighted the issues that arise when dealing with FinTechs, which may not be subject to the same KYC requirements as banks. So, when banks ask other parties for information, such as a customer’s date of birth, those parties may not have the information, “simply because they are not required to collect it.”
“Technology has evolved, but innovating in a heavily-regulated environment is always challenging.”
Pascal Aerens, Business Technologist, Entrepreneur and Financial Crime Expert, iXendar
Also under discussion was whether the model now being used for KYC utilities could work for sanctions. One view is that data sharing is a critical success factor, as a utility that holds banks’ data in “watertight silos” doesn’t bring much value.
While there is a lot of potential in the idea of a trusted identity and the ability to share that trusted entity with multiple organisations, “the ability for banks to really share data is still limited.”
New technologies could help to address these challenges. Vikas Agarwal said that banks are using more sophisticated analytics to tune their models for matching purposes. And leading institutions are taking this further by using machine learning to improve the quality of the information going through the systems.
The panel also said that anomaly detection and machine learning algorithms allow banks to address the complexity associated with some foreign language names. Other areas of focus include automating the research that has to be carried out before a decision is made.
“When you think about these topics of AI and robotics and blockchain, these are things that are transforming not only the front office but the back office, too.”
Vikas Agarwal, FCU Tech and Analytics, PricewaterhouseCoopers LLP
Obstacles to adoption
Adoption of these technologies is currently limited. One of the challenges seems to be regulatory acceptance. “People aren’t even sure how they would validate these models and then explain them to their regulators,” said one of the experts. Complexity was also raised as an impediment to adoption, including concerns about relying on machines to make decisions for complex transactions. The panel highlighted the difference between using machines to gather underlying information versus “having the ultimate decision being made by a computer.”
Where sanctions programmes are concerned, the industry will likely see more measures targeting specific activities, “like we have seen with Venezuela and Russia,” as opposed to a broad trade embargo.
Jack Jared referenced the trend of applying sanctions against ownership interests, arguing that this requires a registry of ultimate beneficial owners in order to succeed. He also argued that governments should consider “stopping the creation of shell corporations.”
Lawlor predicted that there will be more interaction with law enforcement and sanctions agencies. Adding, “somehow we have to figure out how to merge the AML disciplines and the monitoring, coming up with typologies and sharing that information back with the government.”