Last year KS&R conducted an annual survey of US Banks and other Financial Institutions survey commissioned by LexisNexis. Interestingly, the survey showed that although banks do integrate technological solutions and automations as part of their effort to adhere to AML/CFT regulations, the use of manual labor still constitutes well over 50% of spending on combating financial crimes. Even though the banks agree that the costs may be cut dramatically, could they automate more compliance related processes, the AML systems’ shortcomings do not allow that.
The Data Challenge
Eric Young, senior managing director at Guidepost Solutions names and explains the key reason why automation has not yet taken over the manual labor in the fields of KYC due diligence, transactions monitoring, sanctions screening, and other aspects of AML compliance. And the reason is – Data:
“Machine learning and AI need to have data integrity to analyze true and accurate data fed from multiple legacy, front office and other systems, which do not always or often talk to each other. Financial institutions are typically the amalgamation of multiple mergers and acquisitions still using legacy deposit, loan and trading systems. Therefore, they have to invest to ensure data from all these legacy systems are first cleansed and regularized or reformatted to accurately feed to machine learning systems and AI.”
What Young does not say is that systems based on Machine learning and AI also tend to produce many False Positives due to similarity of names and/or patterns, thus leading to the so-called AML Discrimination, prolonged manual processing times, legal actions, reputational damage, and additional costs (operational, legal, post-damage marketing).
Spending on human capital in compliance continues to increase. According to American Banker Association, the cost of AML compliance has more than doubled since 2019 and continues to increase. The average annual cost of financial crime compliance for a financial institution with over $10 billion of assets is approximately $32 million, of which just over 63% goes towards labor and resources, including salaries, training, and outsourced tasks.
The factors that drive increased costs include (according to banks): increasing geopolitical risks, evolving criminal threats, increasing data privacy requirements, customer demand for faster payment, and increasing AML regulations and regulatory expectations.
The greatest challenges are: identifying PEPs (politically exposed persons), sanctions screenings, regulatory reporting, customer risk profiling, identifying relationships between individuals and business entities.
To sum up, most of AML compliance automations lack one critical element, i.e., the ability to understand, cleanse, and unify Data. The Data fed to the systems is incomplete, damaged, or incompatible, so that the resulted screening can be neither accurate, nor efficient and leads to a growing demand in manual processing. Manual processing is very time consuming. Besides, human error is always an option. Considering the current trend of a wide-spread launch of Instant Payment services, banks (and other financial and non-financial institutions) need to find an adequate AML compliance solution. Add to that global nature of financial transactions. This means, that understanding Data includes, among other things, system’s ability to recognize and understand entries in multiple languages and transliterations and entries with spelling mistakes and variations. All these entries are collected from different data silos; they differ in structure, format, language… and they need to be processed in Real Time to support the Instant Payment requirements.
- Smart Data Management and Entity Resolution Engine
- Real Time sanctions screening against sanctions lists, watchlists, PEP, internal lists, etc.
- Effective & Efficient – accurate name matching with high phonetic proximity, ensuring reliable, accurate, and efficient screening. No missed hits. Significant reduction of false positives.
- Screening names in 44 different languages in original scripts (alphabets). Resolving spelling variations, transliterations, including unstructured names.
- Unprecedented reduction in alert rates: <3.6% (vs. current ~30%)
- Smart alert-suppression mechanism (automatic alert disposition) – resolves alerts automatically, reduces false positives, and mitigates operational burden.
- Automations, including (among other things): automated alert disposition, persistence mechanisms, ongoing sanctions monitoring, automatic lists synchronization, updated changes (delta) to sanction lists, and more.
- Analytics and Reporting.
For full functionality and features overview, visit the Product Page.