Financial sanctions and natural language processing: supporting evolving client needs through innovation
By Sinead Mitchell, Citco (Canada) Inc.
Published: 28 November 2022
International financial sanctions are nothing new for the industry. The status of many regions around the world is often in flux, forcing investors such as hedge funds and other alternatives managers to think carefully about where they invest.
However, since the turn of the century, and specifically since 9/11, these sanctions have become increasingly targeted.
As the name suggests, this new breed of sanctions focus on specific individuals or companies. Following the invasion of Ukraine earlier this year a raft of sanctions was published by various international organisations and countries - including the European Union (EU), United Nations (UN) and United Kingdom (UK) - impacting Russian and Belarusian individuals and companies.
Adhering to these sanctions is one of the important tasks which asset servicers handle on a day-to-day basis for clients, and the industry has well-established and effective processes for this. These sanctions feed into the ongoing screening process of the Citco group of companies (Citco).
Using techniques such as fuzzy matching logic (which finds two similar - but not identical - elements of text or information), asset servicers can compare the individuals or companies named on sanctions lists against customer databases (i.e. clients, investors and their associated parties).
As it pertains to investors, the screening process is fed by the data collected via Citco’s Anti-Money Laundering Customer Due Diligence (AMLCDD) process.
While different AML regimes have different identification and verification requirements, many AML regulators have sought to address transparency issues in recent years by requiring asset servicers and their clients to fully understand ownership and control of non-individual investors – i.e. ultimate beneficial ownership (UBO).
This has resulted in a much enriched investor data set for screening, which has been helpful in trying to pre-prepare for potential sanctions.
However, amid the invasion of Ukraine, we have seen a shift this year. Clients have moved to get ahead of the official listings by understanding their potential exposure to Russia and Belarus, both in terms of their underlying investors and their investments.
This presented a new challenge; a shift from monitoring the ‘who’ to a focus of the ‘where’ in regard to domicile and nationality/citizenship.
In response to client requests to know the ‘where’ of their investors specifically, additional, bespoke extracts and reports are now being utilised to collate this data, and this is where technology is playing a growing role.
Can natural language processing help manage sanctions?
As sanctions can be implemented quickly, it is extremely helpful to be able to easily extract data from an asset servicer’s systems.
In the case of exposure to Russia and Belarus, Citco expanded its data set by applying natural language processing (NLP) to its document repositories. This process looked for specific data points on a document, for example, if the place of birth recorded on any passport is Moscow, USSR, etc. While the outputs required manual review, it is clear that this additional technology, once fully developed and tested, will offer an additional layer to Citco’s data collection tools.
However, it is important to understand that NLP is not a standalone solution. This type of data extraction has its limitations within the AMLCDD process. For example, in the case of Russian sanctions, a ‘Russian’ person could present a passport from a different jurisdiction, with a non-Russian nationality and non-Russian registered address.
As a result, the data captured, while compliant, would not present any nexus to Russia. In this scenario, the next stage is to use the name matching capabilities of Citco’s established screening process which would identify an actual sanction target, albeit only after the sanction is published.
So far we can see how NLP can be a useful tool rather than a one-stop-shop solution when it comes to sanctions. However, as demand for better data solutions grows, we will likely see NLP and other technologies used even more in this space.
Clients now want real-time, flexible reporting and analytics, and that means tools such as NLP – combined with existing processes – will take center stage in the battle to adhere to financial sanctions in the future
Being able to automate tasks such as financial sanction screening is undoubtedly a step forward for asset servicers when it comes to delivering ever more value to their clients.
NLP is also developing at a rapid rate. Trends seen in the space this year include an increasing ability to automatically complete parts of documents accurately based on less and less initial input.
While NLP is predominantly used to enhance the interpretation of text currently, it could well expand its reach into areas such as unstructured data, again bringing further benefits to asset servicers and their clients.
However, the future success of this is dependent not just on technological advancement, but also in the adoption of these technologies by all industry participants.
The more widespread the use of these tools across both investors and asset servicers is, the more streamlined the process will become as the various systems used by different organizations manage to talk to each other more clearly.
One other thing is also clear; now that this technology is being used, there will be no stepping back from it. Administrators looking to deliver the best possible service to their clients will need to incorporate these technologies more and more.