How to make faster and more secure decisions on new customer applications?

Since the Anti-Money Laundering (AML)/Counter Terrorism Financing (CTF) changes to the Bank Secrecy Act requirements in 2003 there has been no let up in the requirements for financial institutions to effectively screen customers for risk.  During the same time, data services have expanded over the internet manifold.  This has exposed security risks and breaches that have led to the loss of billions of individual identity records (4.5 billion IDs lost in 2018 alone).

It’s safe to assume that virtually every individual has had some part of their identity compromised – at least once and possibly many times.  The fraudulent use of this gold mine of information is completely in the hands of criminals to use whenever they choose. Finally, criminals have evolved many new and effective means of creating fraudulent identities.

For instance, “Synthetic Identity Fraud” attacks pose a threat across the current process of verifying new customer identities and determining risks.  Financial institutions are at greater risk than ever for complying with BSA requirements and for employing best business practices that mitigate fraud losses.

There is wide potential to use artificial intelligence and machine learning in customer risk assessment to offer new means of increasing risk detection and decision aiding, and in reducing critical labor costs.

What Is Machine Learning?

Machine learning is a process that makes use of engineering, statistics, and computer science to solve problems. It can be used to interpret data by detecting meaningful patterns and take decisions accordingly.

The application of machine learning in the customer risk assessment can help pave the way for improved risk management and better decision making.

How does this benefit financial institutions?

The bottom line for banks, credit unions and financial institutions is that you are able to make decisions on new customer applications faster, and still lead to improved outcomes in terms of customer risk.  This leads to higher longer term growth, and fewer customer problems, all the while decreasing labor costs.

1. Analytical Speed

The use of machine learning and big data can help improve the efficiency with which banks process customer data. A major problem in implementing effective KYC protocols – both for ID verification and for BSA risk compliance – is that there is a lot of data and many compliance requirements.

Read here to find how ClayHR is helping the Banks, Credit Unions and other Financial Institutions.  Also, this 2010 paper on the topic of consumer credit risk models using machine learning models is a good read.

Machine learning and big data analytics can help increase the speed of processing this data and allow for better decision making. Machine learning can also detect meaningful patterns in the data provided and identify any unusual activity.  This can help prevent any illegal transactions from taking place.

‍

2. Improved Risk Assessment

Machine learning tools can detect warning signs that may be missed in traditional risk screening processes. These methods, while capable of doing many tasks traditionally requiring human analysis, can also increase repeatability – so both consistency and reliability are improved.  

Over time these tools collect ‘experience’ in a way that can guard against shifts in productivity and quality when personnel changes occur. Their ability to deal with more data faster is also an advantage in aiding small organizations with critical labor allocation and workflow planning.

3. Improved AML Compliance

Improvements in risk assessment of customers using ML can automatically contribute to better anti-money laundering (AML) procedures and better preparation for audits. For instance, identity or BSA risk characteristics can be scored according to the importance and the decisions of bank personnel monitored with respect to different types of risk.  The ML system can keep track of many combinations of risks versus decisions and provide the user with assistance by recommending safer decision options. Similar ML methods can be used to assess the individual and relative risks of the AML-compliance data collected during onboarding.  

Risk factors like customer age, occupation/employment, citizenship, residency, offshore relationships and planned use of products and services can be automatically analyzed to detect high-risk customers and suggest them for Enhanced Due Diligence. Machine learning can help build a stronger defense system that protects financial institutions against illicit activities. ML can be a significant benefit to smaller banks and credit unions where the costs of certified AML experts is prohibited but where compliance with BSA requirements is still required.  

To Sum It Up

The use of Machine Learning in the risk assessment of customers produces a paradigm shift in the reduction of cost and improvement of efficiency of processes that protect against financial crimes.  More data can be analyzed faster and in greater detail.

The decisions and actions taken by analysts can be processed in conjunction with the analysis of risk and strong recommendations developed for repetitive decision processes.  This paradigm can change levels of skills needed and reduce the demand on valuable supervisor time. Soon, ML will be a vital part of the data analysis and decision-making backbone of every financial institution’s fight against financial crimes.

Also, read how you can use data analytics to strengthen your internal processes and workforce challenges.  While every change is hard, you can start implementing machine learning in customer risk assessment today.  Take the first step and contact one of ClayHR experts.

Bob Cofod is an AML and compliance expert with specialization in business intelligence. He is also the president of ClayHR and leads all key initiatives with the financial sector.  In his spare time, mentors young entrepreneurs and helps them overcome their challenges!