Despite adopting digital technologies for several processes and services, most financial institutions still rely on a huge back-office with manually executed transactions supported with simple and standard rules-based processes. Moreover, the use of disparate technology platforms that are inflexible and often lack ways of integration, makes the operations increasingly manual, expensive, and inefficient. Anti-money laundering (AML) compliance, largely a back-office programme, falls in the above category and is, at present, highly manual notwithstanding the deployment of AML platforms.
On AML investigation process, Booz Allen Hamilton notes (April 2016 report) that financial firms have been increasing head count in their AML operations by 500 percent over a few years. The report states that AML analysts typically spent only 10% of their time on analysis. The bulk of their effort, close to 75%, goes into data collection and another 15% into data organization and entry.
Exploring automation opportunities in the AML investigation process
Anomalies in customer activities, which get generated as alerts using the AML transaction monitoring platform, can be monitored based on KYC profiling and transaction details. The monitoring rules can reflect a number of factors relating to a customer (for example, aggregate transactions, type, amount, frequency, and business). Hence, the conventional process begins with checking historical alert details, downloading relevant data from multiple systems, and then manually reviewing the underlying transaction details, KYC profiles of the subjects, and their transactional counterparties. Furthermore, public domain searches are performed to check if there is any negative or derogatory information about the customers or their counter-parties with respect to the transactions under review. Negative screening is done using watch-list tools. Most of the above actions are mandatory and findings of the same are compiled into an investigations summary report.
From a decision perspective, an alert is escalated (using the AML platform workflow) if the transaction in question is deemed suspicious, or closed if there is no unusual activity.
This entire process is manual. Depending on the complexity of the alert, standards of the bank, availability of information in the core systems, and the experience level of the analyst, the entire investigation process can take anywhere between 30 and 45 minutes for every alert.
The complexity of the situation gets compounded in the light of the fact that the industry average of false positives in AML alerts is 90 – 95%, which means that financial institutions spend a considerable amount of time and effort in non-critical alerts also. Additionally, manual error can lead to regulatory penalties too.
Therefore, it is imperative for financial institutions to look at automating the AML investigation process. This will prove beneficial on two fronts – (i) false positive reduction, and (ii) elimination of manual effort.
The industry has made some progress on the first aspect as a number of AML analytics solutions are being deployed at various global financial services organizations, and there has been significant reduction in the number of false positives. These new-age solutions can be implemented on top of the existing AML platforms of organizations. The analytics models constantly learn and advance the transaction behavior profile of each customer and the customer categories, thereby reducing the false positives.
On the second aspect, financial institutions can adopt robotic process automation (RPA) in their AML programmes to drive operational efficiency, improve flexibility, and enhance the overall performance or back-end processes. In our engagements with several global financial institutions, we have observed that RPA can help an organization save up to 70% of AML investigation time by automating repetitive tasks. Moreover, it can immensely improve the accuracy and efficiency of process execution. For instance, an intelligent workflow automation can seamlessly connect with all internal and external data sources to review, analyze, and generate transaction analysis reports (using analytics techniques), thereby reducing the manual intervention in the AML investigation process.
It’s always wise to adopt a phased approach to RPA implementation, which is why financial institutions should start with automating only a few steps rather than the entire process. This too can bring about visible results. For instance, if you only automate the collation of data from multiple source systems and presenting it through a single sign-on for the investigator, you’ll end up reducing the investigation time by some 10 to 15 minutes – no mean achievement!
With ’responsible innovation’ emerging as the governing mantra, financial services firms are looking to leverage fintech to improve processes in order to effectively meet the evolving needs of customers. Given the potential RPA offers, especially with regard to the regulatory compliance function, banks and financial services firms across the globe have embarked on wide-scale implementation projects. This is further fueled by the fact that AML compliance costs are exorbitant, and automation solutions can offer the much needed relief on both fronts – timely compliance and cost optimization. What is your view?