This is the first post in a series addressing inefficiencies in current AML processes and practical solutions to the most prevalent issues.
I spend my days speaking with AML Officers, FIU Directors, and the investigators and analysts that spend their days completing EDD reviews, transaction monitoring analysis, and case investigations.
I can distill those hundreds of conversations over the past year to three words:
The cost of inefficiency is incalculable, and the most concerning impact of inefficiency is the opportunity cost. What isn’t getting done because of outdated technology and the heap of manual work steps that keep piling up?
Inefficiency Harms the Real Purpose of AML
I have been in AML since 1996 (yes, AML is that old). Over those 20+ years I have seen a lot, and sometimes it is easy to feel a bit cynical. But I still believe those in AML serve a vital purpose. We are part of a global effort to detect and help law enforcement combat evils like narcotics trafficking, human trafficking, crimes against children, and terror financing.
But when does this get done amidst all the checking of boxes, endless new work steps, and struggling with clunky old systems?
How AML Inefficiency Can Be Fixed
So much of AML is inefficient and there are a lot of reasons for this, but instead of dwelling on those reasons, let’s discuss what can be done now to reduce inefficiency. One part of AML at a time.
Why one part at a time?
It would be nice if all the steps in complex AML processes like detecting suspicious activity could be solved at once. There is a lot in today’s AML media suggesting that we are on the cusp of magical Artificial Intelligence (AI) that will do just that. “My CEO said we need to wait for AI and then we can cut our AML costs,” is a sentence I hear often from AML officers.
That ain’t going to happen anytime soon.
Big problems, like big puzzles, get solved one piece at a time. As an example, let’s look at a big problem Uber solved and how the same framework applies to solving AML’s big problems.
For Uber to exist, it relies on the success of multiple technology applications. To click an app on your phone and watch a car drive to you, Uber first needed a lot of things to fall into place. To name just a few:
- The internet
- Reliable Wi-Fi
- Mobile phones
- Mobile phones with Wi-Fi capability
- GPS to work on handheld devices
- Mapping software that plot precision points on a map app.
- Mobile payment systems
- Payment security
It took years of trial and error to make each of these parts of the “stack” work.
Solving AML’s problems is similar. Solutions to improve quality and reduce inefficiency of the many multiple parts of AML must be created, refined, and perfected a piece at a time.
AML is Riddled with Inefficiency
AML processes for Enhanced Due Diligence, transaction monitoring alert analysis, and suspicious activity investigation each involve similar, and often the same, steps. Broadly speaking those steps include:
- Reviewing transaction or other triggering alerts.
- Obtaining, documenting and reviewing customer KYC information.
- Reviewing transaction records.
- Reviewing prior alerts and SARs when they exist.
- Searching public record information.
- Searching negative news, sanctions, watch list, and PEP information.
- Conducting analysis.
- Making risk management decisions.
- Recording findings.
- Creating a file of supporting evidence.
- Having that file reviewed by management
- Having that file reviewed (perhaps) by Internal Audit or the regulator.
Completing EDD, alert, and case reviews can take anywhere from a few minutes to many hours. That time is riddled with inefficiency for many reasons, one of which is that investigators must access multiple systems to search, find and retrieve needed information.
For anyone who does this work, having to flip from one system to another is monotonous and eats away at time. Believers in Artificial Intelligence say that will no longer be necessary.
One day, those believers will be right, but to get to that day, each of these AML steps must first be substantially improved before the AML stack functions in unison.
Fixing Negative News Searching
In this five part series, Bringing Efficiency to AML, we will dissect one of these required AML steps – searching for negative news. We will illustrate how new approaches and new technology like Machine Learning (a necessary precursor to AI) are now being used to substantially reduce inefficiency and improve quality of work in this area.
In dissecting the problems with negative news searching today and how it can be improved, we will publish four additional articles, each addressing one of the elements of negative news searching. These four elements are:
- Searching for negative news;
- Reviewing and analyzing results;
- File documentation and record keeping; and,
- Quality control, Internal Audit, and examiner review of work product.
Look next week to read about the inefficiency and risk of negative news searching and how solving these problem is now possible.