The document discusses how forensic data analytics (FDA) can help banks better detect money laundering risks and meet evolving anti-money laundering (AML) regulations. FDA uses advanced analytics on both structured and unstructured data to identify hidden relationships and behaviors that may indicate money laundering. This allows banks to more accurately predict and prevent AML risks. Major banks are already using FDA to enhance transaction monitoring and develop risk scores to flag potentially suspicious activities for investigation.
1. Forensic Data Analytics:
The changing regulatory landscape within the financial services industry
has made recent years extremely complex and difficult for both
domestic and international banks to operate efficiently. In attempts to
assimilate current business operations to the ongoing market changes,
banks are devoting more resources to Know Your Customers (KYC),
Anti-Money Laundering (AML), fraud detection and prevention, as well
as sanctions compliance.
Recent AML penalties and fines handed down to major banks around the world verify that one or
two dimensional customer screening or transactions monitoring programs are simply not
sufficient to properly manage the compliance risks present today. In order to meet the ever-
evolving regulatory environment and address the AML regulatory environment, the best possible
solution for banks is to use a scalable, multi-dimensional data analytics technology to identify
hidden abnormal behaviors within mass amounts of random data, predict risks based on real world
scenarios and deliver intelligence business insights to make strategic decision, as well as to
continue improving AML processes and procedures based on data and behavior rather than just
updating traditional rules in hopes of identifying the modern day criminal.
Traditional AML transaction monitoring is a rule based monitoring system (two dimensional
monitoring) that satisfies basic regulatory and business requirements. However, we need to ask
ourselves if a rule based monitoring system based on standard policies and procedures is
impenetrable or is it applicable and encompassing to your global operations? The global financial
environment and the way people conduct business with it or within it is ever changing; yet, AML
policies and procedures have remained mostly unchanged. As such business growth requires more
advanced and innovative strategies to be competitive, yet remain compliant in today’s
environment.
About EY’s Fraud
Investigation &
Dispute Services:
Dealing with complex
issues of fraud,
regulatory compliance
and business disputes
can detract from efforts
to succeed. Better
management of fraud
risk and compliance
exposure is a critical
business priority — no
matter what the
industry sector is. With
our more than 3,000
fraud investigation and
dispute professionals
around the world, we
assemble the right
multidisciplinary and
culturally aligned team
to work with you and
your legal advisors. We
work to give you the
benefit of our broad
sector experience, our
deep subject matter
knowledge and the
latest insights from our
work worldwide.
The future of detecting, predicting
and preventing Anti-Money
Laundering risks
2. Forensic Data Analytics: The future of detecting, predicting and preventing Anti-Money Laundering risks 2
Analytics as a part of the business discipline has existed for
decades within the banking industry. Its application and
acceptance have increased recently due to a number of
reasons, including but not limited to: the pace and scale at
which data is accumulating from different sources, the vast
amounts of structured and unstructured data, the speed of
access, the rapid reduction of storage costs and the
sophistication of enabling tools and technology. Forensic Data
Analytics (FDA) combines highly specialized and skilled people,
advanced technology and years of forensic experience to help
banks and their management make quicker, more intelligent
and well-informed business decisions. FDA now sits at the top
of the AML monitoring agenda for many leading financial
institutions looking for more effective ways to identify outliers,
hidden transactions, unknown relationships and abnormal
activities that cannot be seen at the transactional level and
could now be linked together with FDA.
FDA takes the data collected and analyzes it at a whole new
level, a multidimensional analysis combining structured and
unstructured data that identifies hidden relationships and
correlations. By comparing each transaction, understanding the
behavior of the transactions, and applying WHO, WHAT,
WHERE, WHEN, WHY and HOW technics to all transactions, we
are not only able to identify suspicious activities, but we are
also capable of finding loop holes in the internal controls,
policies and
procedures that require constant fixing and improvement.
Furthermore, FDA allows banks to predict and score day-to-day
transactions as high/medium/low risk activities based on the
past behaviors at the bank and historical industry benchmarks.
Changing and customizing policies and procedures based on the
data that has been collected and analyzed for each individual
bank is essential, because the data pinpoints and tells you what
is really happening at each bank and what the areas are that
need to be strengthened. Generic and traditional rules can no
longer be used to apply to global banking operations when in
fact, each market is unique and its peoples’ behavior differs due
to diverse customs.
The heart and the soul of FDA is within the risk indicators that
are tailored to each bank and its multiple business operations.
These quantifiable risk indicators are developed from risk
scoring algorithms based on historical unethical behavior data
and weights calculated for each risk factor. These indicators
should be dynamically changed and updated based on data
behaviors, business growth strategies, and regulatory changes.
Assigning mathematical risk scoring to each transaction, person
and company gives banks a 3-dimensional data analytics
capability to predict potential high risk AML/ CFT non-
compliance issues before they occur.
Text
Picture
Email
Sentiment Analysis
Customer Relationship
Management System
Database
Accounting
System
80%20%
Source: Gartner consulting company
Structured Data Unstructured Data
In the past, structured data has served banks well
in the AML space, as it has created opportunities
to manage a mind-bending array of information
and data. However, it is important to note that
around 80% of corporate data is now unstructured
or quasi-structured, including but not limited to
social media, emails, texts, audio, video, instant
messages, photos, metadata, blogs and much
more, and this percentage will only increase over
time. Thus, it creates the need for businesses and
management operating in today’s data
environment to have an entirely different mindset
and approach when it comes to analytics and to
go outside of the box when it comes to evaluating
current, traditional two dimensional analytics
programs.
3. Forensic Data Analytics: The future of detecting, predicting and preventing Anti-Money Laundering risks 3
Behavior and prediction - FDA combines the extensive
use of big data and statistical and qualitative analysis in
conjunction with explanatory and predictive models to
guide and identify AML/CFT violations and areas
warranting further review. Our fact-based evidence
drives actionable business decisions, focuses
investigative efforts where it matters and optimizes
outcomes. FDA comprises proactive and reactive
methodologies that leverage the information contained
in large-scale, structured and unstructured data sets.
This allows banks to effectively detect and prevent fraud,
to identify instances of error, ML/TF typologies and
misconduct, or to address to a regulatory response.
Network analytics – There are three key advantages to
network analytics. Firstly, the technique itself
helps identify high risk stakeholders, key
entities, sensitive files and keywords. Secondly, it
ensures that data sources for monitoring includes due
diligence databases, hidden network access of social
media networks, blogs, forums and high risk IP
addresses. Lastly, it identifies certain behavior, including
cyphers, negative emotions and topic modelling.
The benefits of using FDA:
1
2
3
4
Ability to detect hidden behaviors – These behaviors
cannot be found using standard and traditional two
dimensional models. The FDA approach incorporates
targeted model-based mining and visual analytics tools
that allow the data to ‘speak for itself’. When deployed
over large data sets, our analytics can be a powerful tool
to identify large and unusual transactions or anomalies
derived from multidimensional attributes within a bank’s
transactional data. Model-based mining shifts the focus to
high-risk areas where controls may not necessarily exist or
perhaps even bypassed. FDA has the ability to detect
abnormal behavioral changes in the data, as it compares
all data sources against each other.
Predictive modeling and detection – With historical
corporate big data, unethical behavior models and
statistics, and data from numerous past and real world
scenarios, FDA can be transformed into a predictive
analytical platform to detect potential risks or threats
before they occur. With today’s technological advances,
the future of fighting crime and exceeding regulatory
requirements is the ability to predict and mitigate risks
before they occur.
4. Forensic Data Analytics: The future of detecting, predicting and preventing Anti-Money Laundering risks 4
3
Improper Transaction-Network Analytics - Visualization to
identify rogue accounts or potentially improper transaction
Major Banks are currently using FDA in the following ways:
Banks are implementing FDA risk scoring model into their existing transaction monitoring systems, which focus on
identification of suspicious patterns of transactions that may result in the filing of Suspicious Activity Reports (SARs) or
Suspicious Transaction Reports (STRs). FDA transforms a traditional, two dimensional outlier identifier into a 3D analytics
platform. FDA risk scoring models also allow banks to convert and profile reports into high risk population of people/entity
for further data mining.
Bank’s current challenges
► One person may have multiple credit card accounts with
different names, ID, and address. How do they identify
which credit card account belongs to which customers?
► Identify and compare account level behavior for each
account in cash deposits to the credit card and cash
withdraws to the credit card
► Identify potential outliers in credit card account
transactions and continue improving AML policies and
procedures through collected data
FDA Solution
► Leveraging current banking system infrastructure to
collect relevant data and identify bank’s own Risk
Indicators
► Leveraging cluster analytics to identify which person may
potentially have multiple credit card accounts and the
behavior of those accounts
► Leveraging network analytics to determine the
relationships between each account and its behavior
towards cash deposit and cash withdraw activities
► Leveraging predicted modeling to identify
high/medium/low credit card transactions that relate to
AML
► Develop rule-based risk scores for each and every
transaction, account and business to identify potentially
high risk activities
► Developed interactive and visual dashboards that allow
senior management to quickly see where the high risks
are in their bank and to easily navigate through the
transactional data
A major Chinese bank with strong credit card business is leveraging FDA to detect and identify
potential AML activities outside of the traditional rule based monitoring system.
Forensic Data Analytics: The future of detecting, predicting and preventing Anti-Money Laundering risks 4
1
5. Forensic Data Analytics: The future of detecting, predicting and preventing Anti-Money Laundering risks 5
Combining multiple data sources to find hidden patterns
and trends
A global leading international bank is leveraging FDA to detect AML activities and predict
high/medium/low risk transactions
Bank’s current challenges
► Existing client’s IT infrastructure was neither designed to
process today’s magnitude, complexity or workload of
data nor was it designed to effectively prevent and detect
fraud
► The top two client challenges using Forensic Data
Analytics were “getting the right tools and expertise for
forensic data analysis” and “combining data across
various IT systems”
► To view global transactional data in one dashboard with
the capabilities of identifying outliers, viewing transaction
relationships and predicting high risk transactions from
rule based to statistical prediction
FDA Solution
► Developed interactive, visualized dashboards for senior
management to navigate through their transactional data.
► Leveraged predicted modeling to identify
high/medium/low transitions that relates to AML by
region, by country, by department and by resources
► Developed rule based risk scores for each and every
transaction, account and business to identify high risk
activity
► Leveraged network analytics to understand internal
resource relationships, behavior of the customers ( Know
your Customers) and 3rd party relationships
External
news and
data feeds
Social
media
Transaction
data
(e.g., sales)
Email and
instant
messages
ERP and
financial
systems
EY Counter
Fraud Analysis
5
2
Developing an FDA strategy and putting the systems, software and tools in place to execute is critical. But more
importantly, an effective FDA program requires conversations across multiple departmental lines to understand the
business and its data from all angles. As such, it is vital to cultivate the right thinking and skillsets within the business and
define new positions that defy traditional job titles and responsibilities. At EY, we understand the significance of Big Data
and FDA, and as such we have invested over USD500MM into our data analytics strategy to help our clients develop and
implement successful FDA programs by combining state-of-the-art technologies with our people and our rich experience in
banking, AML/CFT, sanctions and fraud risks.