Money laundering, terrorist financing: Why we need customer risk rating

Image: Shutterstock

What is dynamic customer risk rating?
Dynamic Customer Risk Rating (CRR) entails capturing money laundering or terrorist financing (ML/TF) risk of a customer at the time of onboarding and updating that at regular intervals. It is a combination of a customer’s static as well as dynamic information. Banks are increasingly using various tools and methodologies with parameters ranging from transaction monitoring alerts, Suspicious Transaction Report (STR) filed, and linkages between customers or accounts to continuously update the risk rating of the customer.

Why do we need dynamic customer risk rating?
CRR is a critical component of a comprehensive Anti-Money Laundering (AML) programme. Customer Due Diligences (CDD) begins with establishing the customer identity, and assessing and ascertaining the risk associated with each customer. It is critical to have an effective CRR in place as it affects the ongoing transaction monitoring and periodic reviews adopted by the financial institution (FI).

Traditionally, the CRR model FIs adopted was typically based on certain elements on customer profile like nature of business and industry, geography, products and services availed, etc., which were more static in nature. However, the risk of ML/TF faced by the FIs has evolved. In the recent past, regulators have raised the bar and the FIs are now looking to move to a dynamic CRR model.

What statistical models should be used to predict the risk factors?
With technological advances, a few FIs are shifting from their traditional static CRR models to implementing dynamic risk rating model by using various statistical models and Machine Learning (ML) procedures, including network science to  get a more robust view of the customer risk.

Some of the popularly used statistical models such as linear regression model, binary or ordinal logistic regression model, decision trees, etc., must meet specific assumptions with a certain degree of statistical confidence to reveal the most predictive risk factors. ML procedures can help evaluate the results of these statistical models to weed out false positive events and fix data quality issues.

One more emerging tool, which is being deployed for assessing customer risk, is the Artificial Neural Network (ANN). One of the biggest advantages of ANN analysis is that it can help FIs in identifying patterns in transactions, which if viewed individually, may not ring any bell of suspicion. However, if these transactions are viewed as a cluster, they may provide linkages to the group behaviour of apparently associated customer accounts.

Why is there a need to move to dynamic risk rating model?
The FIs are slowly shifting towards dynamic risk rating models to assess customer risks. A few pre-requisites for implementing a successful dynamic risk-rating model include the following:

(i) Simplifying risk-rating models and ensuring consistent set of risk factors across business lines

(ii) Improving data quality being captured including conducting independent data verification to the extent possible

(iii) Continuously updating customer information by not just depending on customer inputs but proactively identifying changes to customer information from external sources

In general, the fight against financial crimes, and particularly money laundering, is getting tougher as the modus operandi used by criminals is becoming more complex. Hence, the FIs need to provide more sophisticated measures and frameworks. An advanced CRR model supported by ML and statistical analysis is the need of the hour.

The time and resources that FIs will need to devote in their journey to develop and successfully deploy these CRR models would depend on their current state. However, given the ever-increasing risk of ML/TF and the regulatory pressures, eventually, all FIs will have to take this journey.

A customer’s risk profile directly affects a FI’s ability to identify and manage AML risk by helping them understand what is usual and expected of the customer behaviour, and deploying the appropriate level of due diligence and controls. A dynamic risk-rating model will go a long way in helping organisations in this process and improving the efficiency and effectiveness of the AML framework.

KV Karthik is Partner & Soniya Mahajan is Associate Director – Forensic, Financial Advisory at Deloitte India

The thoughts and opinions shared here are of the author.