The Business Problem: Financial Crime, Penalties, and Operational Costs
Money laundering is a global criminal enterprise of stunning size. Recently, the United Nations published its study of previously classified government documents that detail money laundering activities. By their estimates, a small number of documents indicated that some 90 financial institutions had experienced between $800 million to $2 trillion in suspicious activity on an annual basis.
According to Bloomberg News, “Without an urgent, concerted political effort, criminals — from drug dealers and terrorists to human traffickers — will keep the upper hand.”1
Money laundering creates financial and regulatory headaches for banks. On average, penalties imposed for proven money laundering amount to $145 million per case. In 2019, that added up to $8 billion, twice the amount of 2018. With this level of regulatory scrutiny, financial services institutions have established complex and resource-intensive transaction-monitoring capabilities to demonstrate and report regulatory compliance. These frameworks are usually built around rule-based detection systems that generate alerts of potential Suspicious Activity Reports (SARs) that must be manually reviewed and reported. Often, though, there are a considerable number of false positives. Some studies suggest between 95-99% of alerts are false positives. This requires banks to hire skilled investigators and use expensive resources to review alerts. Even then, banks may miss anomalous changes in money laundering behaviors. On average, a bank spends $48 million annually on these compliance programs for anti-money laundering (AML), across people, processes, and technology.
The cost of reviewing an alert ranges between $30-$70. For a bank that receives 100,000 alerts a year, the reduction in false positives could result in between $600,000 – $4.2 million in savings.
The traditional rules-based detection systems must cope with a large volume of data. These systems often set conservative threshold tuning, because compliance teams would rather review alerts (even likely false positives) than allow money laundering to pass through lower thresholds. These systems cannot detect anomalous changes in behavior or nascent money laundering patterns before they spread. Monitoring is time-consuming and labor-intensive. It forces banks to spend time chasing false positives and hunting for investigators’ notes.
Creating a Powerful Partnership
DataRobot has partnered with Automation Anywhere, a Robotic Process Automation (RPA) company headquartered in Silicon Valley, to apply intelligent automation to uncover financial crimes and help banks reduce costs and regulatory risk.
A New Approach to Detecting Money Laundering
DataRobot delivers powerful AI and automated machine learning, to accelerate the model development, deployment, and monitoring of models at scale. DataRobot uncovers insights within data that would be impossible for even expert humans to detect.
Automation Anywhere is a global leader in Robotic Process Automation (RPA), making employees more productive by partnering them with “digital coworkers” that take care of repetitive, rule-based tasks.
The powerful combination of AI and RPA to create intelligent automation helps financial institutions prevent the damage that money laundering can do to their organizations, while reducing operational costs through automation.
The approach to deterring financial crime is based on Intelligent Process Automation (IPA). This next-generation combination of tools assists financial services organizations by automating repetitive and routine tasks, such as data gathering, and populating SAR reports on money laundering cases. IPA mimics activities carried out by humans. It even learns to do them better, because it can eliminate errors caused by human oversight. AI and RPA together enhance traditional automation and decision-making capabilities to increase efficiency and output and improve customer experience.
There are three steps in this process:
Data collection. Automation Anywhere logs in to third-party systems, collects data (such as transactional, customer due diligence, and adverse news), and prepares the data for the machine learning model.
Machine learning. A DataRobot model is overlaid over the rules-based monitoring system and scores the alerts to reduce the number of false positives (on average, by 20 to 60%).
Analytics and RPA. Suspicious cases are generated and prioritized. The RPA bot sends an automated email to the case handlers for review. Regulatory reports can be automatically populated and sent for review, with prediction explanations that outline the key drivers of each prediction being made.
Together, DataRobot and Automation Anywhere deliver a financial services solution that helps banks:
- Better respond to regulatory scrutiny.
- Reduce the costs of financial crime compliance programs.
- Reduce processing costs.
- Reduce false positives.
An Affordable Solution
The DataRobot and Automation Anywhere solution develops models tailored to banks’ requirements, deploys and monitors those models, and measures the value the models provide.
DataRobot also updates its models to correct for drift in accuracy and to ensure that their performance meets customers’ expectations.
For more information about this new approach to reducing money laundering, contact your account teams and book a demonstration with DataRobot and Automation Anywhere.
1. “The World Is Losing the Money Laundering Fight,” Bloomberg Opinion, Sept. 21, 2020
The post Applying Intelligent Automation to the $2 Trillion Global Problem of Money Laundering appeared first on DataRobot.