By Greg Pinn
The quality of name matching is the single most important factor in reducing risk, cost, and turnaround time in the know your customer (KYC) process. Relying on outdated or simplistic matching techniques (e.g., only using exact, partial, and phonetic matching) results in significantly higher false positive and false negative rates, reduces the efficiency of compliance efforts, and opens financial institutions (FIs) to regulatory fines and reputational damage.
Manual, human remediation of screening matches is the largest area of cost associated with anti-money laundering (AML) compliance efforts. A recent study found that 40% of financial crime compliance costs is related to internal labor costs with an additional 20% for outsourced costs. With an average tenure of 12 months, financial institutions must continually hire and train new AML analysts to cover both attrition and growing compliance needs. These costs continue to grow both in absolute terms and as a percentage of overall compliance spend.
For many FIs, AML screening solutions are so integrated into existing workflows that any replacement, regardless of efficiency gains would be cost and time prohibitive. Such changes often require onboarding new teams to manage the technology, logistics, and regulatory issues that come with such a fundamental change to the organization’s risk modeling. While these changes may be beneficial from a risk mitigation and operational efficiency perspective, getting buy-in for such a potentially disruptive project can be difficult.
Simple screening solutions as described above cause a significant increase in human remediation effort. As an example, the OFAC Sanction List Search provides a mechanism for interested parties to screen against both the SDN and Non-SDN lists. A search for Hasan Khan returns over 45 results with a score of 80% or higher:
|KHAN, Rahat Hasan||100|
|HASAN DOUKHAN, Abdul Fattah||90|
|HASAN DUKHAN, Abdul Fattah||90|
|HASAN, 'Ali bin||90|
|HASAN, Hameed ul||89|
|HASAN, Kamal 'Ali||89|
|HAFIZ, Said Khan||87|
|HASAN, Shaikh Daud||87|
|HASAN YUSUF, Ahmad||85|
The table above displays all search results with a confidence score of 85 or greater, a score for which most institutions would require review. However, it’s clear even with a superficial glance that most of these names are not a match. The OFAC screening service returns a confidence score of 100 for Rahat Hasan Khan even though it is obviously not an exact match. While it may be the same individual, it certainly is not a 100% match to the search term. Other names simply match because of the term Hasan, a first name in our query, but a last name in most of these results. With over 4,200,000 people with the surname Hasan, returning such a strong match score for only the last name causes a huge, unnecessary manual remediation workload.
Recognizing that replacing existing AML solutions may not be feasible for many organizations due to budgetary limitations, compliance teams must identify solutions that supplement existing systems and reduce the manual remediation required for watchlist screening.
Rosette® Name Indexer by Babel Street is designed to solve this problem. By using Rosette’s pairwise matching capability on the results from an existing matching system, financial institutions can quickly and automatically pare down the results requiring manual review by using a high-speed, configurable resolver.
From our example above, here are the results when the names returned by OFAC to match Hasan Khan are put through Rosette pairwise matching.
|NAME||ORIGINAL SCORE||ROSETTE SCORE|
|Rahat Hasan KHAN||100||82.6|
|Abdul Fattah HASAN DOUKHAN||90||75.5|
|Abdul Fattah HASAN DUKHAN||90||71.5|
|'Ali bin HASAN||90||53.2|
|Hameed ul HASAN||89||58.1|
|Kamal 'Ali HASAN||89||55.8|
|Said Khan HAFIZ||87||51.6|
|Shaikh Daud HASAN||87||57.9|
|Ahmad HASAN YUSUF||85||49.5|
Rosette Name Indexer is highly configurable and supports 24 languages across multiple scripts enabling high-quality matches regardless of the data source. The above example relies on our highly tested and optimized default settings.
In this situation, we have used Rosette’s pairwise matcher to post-process the results retrieved from the OFAC Sanction Screener. This is simply an example, and this post-processing can work against any matching engine. Analysis of these results demonstrates that Rosette has reduced the number of results down to a single result that would require human intervention: Rahat Hasan Khan. The level of granularity provided by Rosette enables compliance teams to make targeted judgments on where to draw the line for manual remediation.
Implementing Rosette Name Indexer for second pass remediation can save teams significant manual effort in name matching processes. Our name matcher is the culmination of 20+ years of research and development, creating the most sophisticated, configurable name matcher available on the market.
All names, companies, and incidents portrayed in this document are fictitious. No identification with actual persons (living or deceased), places, companies, and products are intended or should be inferred.