Over the last decade, the compliance offices of the global banks have been heavily hit twice. First time, by money launderers, who have become more and more creative in inventing methods to use the financial system in their operations. Later, by the regulators, which have addressed the money laundering risk by imposing strict legislation and have elevated their oversight on the financial sector which resulted in the number of heavy fines imposed on the banks. The era of don’t ask don’t tell policy has come to an end and financial institutions became aware, that understaffed and underfinanced anti-money laundering (AML) programmes may cost them more than acceptable, both in terms of financial fines and reputational damage.
Until recently, many banks have responded to regulatory challenges by increasing the headcount of compliance teams, while maintaining the existing processes. If these are not maintained and overseen properly they ended having costly and ineffective programmes. The next step, which will have to be taken is an investment in technology, which can help improve the AML processes. New technologies are becoming more and more accessible, due to wide-spread availability and decreasing costs of implementation. Technologies that were only associated with science fiction movies a decade ago, are now becoming widely available. There are multiple ways the manual process can be enhanced by technology.
Robotic Process Automation (RPA)
Robotic Process Automation (RPA) can emulate human actions in a predefined sequence. Such bots can open defined website or application, log in, and perform simple actions, like copying and pasting or extracting sets of data. Robots can perform manual, repetitive tasks the same way humans do. The difference is they can do quicker and operate 24/7 increasing efficiency and optimizing cost-effectiveness at the same time. A perfect implementation of RPA would be in case of gathering data through websites, regulators’ databases or commercial data aggregators.
Machine Learning, a subset of Artificial Intelligence, which allows performing tasks with no direct instructions provided. Algorithms are fed with the training data set, which then is used as a base to make decisions or provide predictions. Machine Learning algorithms can be a great improvement in risk scoring process used in Know Your Customer (KYC) procedures. Another potential implementation of such a solution is in generating transaction monitoring alerts. PricewaterhouseCoopers (PwC) report estimated that 90%-95% of all alerts generated by current tools are false positives, which tremendously increases costs of the manual review. Implementation of machine learning algorithms which can analyze large data sets and prior reports may substantially improve the efficiency of the process.
Machine learning algorithms are being successfully implemented in other banking departments. Software House Altkom Software & Consulting has created a dedicated solution which supports one of our clients in processing sales of installment loans. It automatically identifies industry, based solely on the name of the product in the customer’s shopping cart.
Natural Language Processing (NLP)
Natural Language Processing (NLP) is another branch of Artificial Intelligence which can boost the effectiveness of AML programmes. It allows understanding the text they way humans do and is able to determine what are its implications. In other words, the NLP can learn to read between the lines, similar to how humans do. It is a potential partner in analysis of the adverse media checks, performed in both KYC and transaction monitoring processes. It can serve as the first stage of review and filter out the majority of articles while allowing the analysts to focus on the non-obvious ones.
Proper implementation of new technologies into the AML operations offers extended possibilities of cost savings and process streamline, however, there might be a few speed bumps in integrating them into the compliance programmes. First is the investment needed in the development. One size does not fit all, the solution must be tailored to the scale and characteristics of individual bank customer portfolio to effectively improve the compliance efforts. Not all institutions recognize this investment as necessary, as it does not translate directly into the growth of revenue. However, well-thought investment in technology may reduce the costs of employment and increase productiveness. Moreover, the regulatory pressure on financial institutions is growing and in the upcoming years, the use of efficient and powerful technology will be the easiest and the cheapest method of maintaining compliance with regulations.
Another obstacle is the fact that the regulatory bodies often require full documentation of how the systems work and how the risk is assessed and decision are made. Explanation of the mechanisms defining the logic in case of advanced AI algorithms may be unsatisfactory to the regulatory bodies and the audit trail must be typically stored for documentation purposes.
The possibilities of increasing the effectiveness of the AML programmes are inevitably associated with the vision of the bank management and thorough understanding of all the benefits of such solutions. At the same time, the regulatory bodies must ensure that the implementation of new technologies will not be slowed down by the procedural limitations and the banking sector needs to be supported in this process. Only both these factors combined can trigger a revolution in banking AML programme.
Some of these technologies are being considered and some are already being implemented in the large global banks. Coming soon, we will present the latest trends in the AML field and demonstrate the financial institutions’ efforts to incorporate the advanced technologies into their compliance operations.
- From source to surveillance: the hidden risk in AML monitoring system optimization
- Difference between Machine learning and Artificial Intelligence
Author: Maciej Dudkowski, Junior Business Analyst