Banks forced to turn to machine learning to tackle compliance burdens

By David Beach | 12 April 2019

Banks are increasingly turning to machine learning to innovate heavily manual and costly middle and back office mission critical compliance management processes, say market participants.

“Banks are already in a period of change, mostly in the more business- driven activities in the front office but it’s time to appreciate the value that technology brings to those areas of business in the middle and back office,” says Anastasia Dokuchaeva, head of partnerships at ClauseMatch.

A recent report from JWG and Marklogic, reveals that banks will have to comply with 374 legislative initiatives between now and 2021, following on from an estimated 50,000 regulatory documents in the five years after the financial crisis. The senior executives interviewed as part of the report acknowledged that using a “project-based, highly-manual and siloed approach to regulatory implementation was not efficient nor sustainable”.

Source: JWG & Marklogic report

A similar report from Fujitsu published last year found that organisations are typically turning to technologies that automate functions in order to cut out manual processes. CTO of Fujitsu UK and Ireland Ian Bradbury said that the “complex and time consuming” nature of regulation for banks is leading many to begin digitising compliance processes, which he believes explains the “significant and growing regtech marketplace”.

Dokuchaeva, formerly of investment data management vendor, Factset, described her move into regtech as “eye opening to see how much historic unstructured text, documents and policies held mission critical information.

“Being able to tap into the large volume database [of compliance management] and capture all activities, decisions, changes and relationships is a treasure trove for data analytics,” she says.

Likewise Julien Recan, who recently joined regtech Alpha Reply as an associate partner, pins cost as the most important factor for banks to move away from their existing compliance management processes.

According to the latest Bain & Company report, governance, risk and compliance (GRC) processes account for 15-20% of the total operational cost of most major banks and roughly 40% of transformation costs, while the JWG and Marklogic report puts the cost of compliance at 4% of revenue, likely to increase to 10% by 2022.

Recan, who was previously chief risk officer at Revolut and before that, head of risk quants at RBS is all too familiar with banks’ compliance management processes, which he calls “bulky” and “heavily manual”.

“Imagine you have a regulation that drives a clause in a contract that you have with your many clients,” says Recan. “If you need to change that clause for new regulations or because you’ve changed your T&Cs, typically a firm would have to do that manually by opening up each individual contract to see if the clause needs changing or not.”

Alpha Reply groups and implements regulatory solutions at financial institutions.

“The cost and cost savings can be quite dramatic,” says Recan, “there are certain processes that are heavily manual that could be nearly completely automated. Just recently, we were talking to a client who could save 10 people and move them to more productive functions through a straightforward implementation.”

ClauseMatch, deployed in the cloud by open APIs, employs natural language processing to link clauses in documents to other segments in other documents using hashtags and metatags, allowing firms to pre-map and connect relationships between various clauses in separate documents.

Become a bobsguide member to access the following

1. Unrestricted access to bobsguide
2. Send a proposal request
3. Insights delivered daily to your inbox
4. Career development