Fraud, a multi-structured and multi-layered phenomenon, poses a big challenge to financial institutions. The accelerating rate and complexity of financial frauds demand better and more effective defense solutions with robust Machine Learning, data analytics and predictive capability.

Effectively mitigating financial fraud will help banks protect its customers, employees and reputation, while also enhancing the resilience of the financial system.

Some of the key challenges banks face in today’s context are:

• Newer regulatory and compliance reporting
• Internal challenges often caused by legacy issues
• Disjointed fraud systems and teams
• Lack of cost synergies
• Silo views of client’s risk
• Duplication of core risk and control activities
• High dependence on archaic rule based systems

Bank’s Risk Management is characterized by reactive, rules-driven detection approach to provide response to regulation. However, maturing landscape and advancement in technology offers the opportunity to improve outcomes and efficiency in threat combat as well as to mitigate the key challenges addressed above.

A recent industry survey identified that fraud and compliance monitoring is increasingly the focus of senior management¹. This is due to the rising frequency and value of fines being issued for non-compliance. Firms typically spend 4% of their total revenue on compliance, but that could rise to 10% by 2022². Banks spend most on enhancing their transaction monitoring, fraud and payment systems³. In addition to system upgrades, banks are actively looking for applying advanced analytics to augment resilience in their compliance function.

A proactive approach to fraud management

As banks embark on the journey towards proactive, analytics-driven capability, they look to adopt a robust real-time approach to identify threats using data analysis and service-oriented architecture to define the target approach.

Some of the core capabilities where analytics is combined with existing legacy architecture which have proven beneficial to the banks are:

Advanced Segmentation - Most banks leverage multiple technologies as part of their fraud control activities. Segmentation is a fundamental component of anti-money laundering (AML), and is concerned with the grouping of customers based on similar transactional attributes. By leveraging advanced data mining and aggregation techniques, banks are able to transition from a small number of high-level segments based on hypotheses-driven segmentation to lower-level behaviour-driven segments through data-driven segmentation.

Data-driven segmentation involves the following steps:

a) Identify data population

b) Create an analytical base table for profile uniqueness

c) Build topological model algorithms

d) Validate model using test and training data sets

e) Generate segments

Artificial Intelligence (AI) and Robotic Process Automation (RPA) – As traditional processes and silo teams do not support an agile response to customer and market needs, banks have modelled new ‘Ways of Working’ to include AI and RPA. Intelligent automation in RPA has created growth through a set of features enhancing traditional automation solutions. Banks consider RPA as the starting point in their journey to unlock tangible benefits delivered by AI.

Some main use cases for AI are:

a) Intelligent product pricing

b) Automated advanced tasks by incorporating NLP and image recognition

c) Next best action through enhanced judgement

d) Chat-bots and virtual agents

e) Malware prediction

f) Voice recognition

Dynamic Transactional Monitoring - Traditional Transactional Monitoring (TM) solutions rely on static rules comparing base transaction and Know Your Customer (KYC) data against pre-defined TM configuration settings. This results in large case volumes and a high proportion of false positives through ‘alerts’. These alerts are generated using a fixed batch process, usually on a monthly basis, instead of real time. Also, there is a reliance on ‘judgement calls’ on how alerts are handled, which is not a consistent and effective process.

By leveraging analytics and advanced data science, banks are able to implement automated ‘Alert Hibernation’ process, which helps to detect anomalies and thereby reduce false positives.

There are a number of factors influencing the growth for analytics-based TM systems. They are:

a) Sustainable data ingestion approach and improved understanding of data

b) Streamlined alert monitoring through the ability to discriminate recurring false positives

c) Optimum usage of Level 1 and 2 alert handling workforce

d) Effective Level 3 process for filing Suspicious Activity Report (SAR)

e) Standardized backlog remediation and TM system

Reduction of Customer Periodic Reviews - As part of the KYC process, customer reviews are conducted manually, either periodically (based on customer rating) or triggered (based on risk). Banks find that periodic reviews do not change the customers’ risk rating; on the contrary, they increase operational costs by involving manual checks. For cost savings, banks are moving to ‘Risk event based reviews’ wherein data from internal systems (transaction monitoring, name screening, alerts, events) as well as external data (adverse media, court rulings, government source) are integrated to compute Risk Score for each customer and reviews are conducted only for profiles that have a revised risk score based on event changes.

Banks are automating client-on boarding activities, leading to decreased process time, average handling time, and end-to-end costs, helping them to reduce service backlogs.

Event-based reviews involve the following steps:

a) Identify material and administrative triggers

b) Build a model scoring algorithm combining internal and external datasets

c) Generate risk score for customer profile based on algorithm

d) Conduct reviews based on score

e) Integrate score with existing workflow for case prioritization

Workforce Analytics - Banks use Workforce Analytics to create the capability to understand the turnaround time and efficiency of its AML agent workforce. Analytics of agent performance is based on a combination of segmentation and predictive modelling techniques applied to agent analytical record which serves as a single view for AML agents operations.

Through Workforce Analytics, banks are able to identify four dimensions that play a crucial role in the successful running of AML operations. They are:

²¹)ÌýAlert volumes - Provides an insight to understand the alert volumes generated in Name Screening, Transaction Screening and Transaction Monitoring systems and the number of workforce required to investigate the alerts in respective areas.

b) Demand management - Helps to understand the operational impact by being able to re-allocate teams and prioritization.

c) Operational efficiency - Provides deep insight into agent’s performance leading to higher team efficiency, thereby reducing the number of agents needed to process a certain number of alerts.

d) Quality criteria – Helps to model quality performance and the factors that impact quality leading to greater operational capacity.

Regulatory reporting – The impact of regulatory changes and the multi-geography nature of financial institutions have left processes fragmented with multiple hand-offs across operational teams. Regulatory breach possess significant risks / costs to firms, including monetary losses, regulatory fines, and negative reputational impacts.

 Evolving regulatory landscape makes regulatory reporting a constant challenge for banks, making some banks focus on analysing disparate data points and gathering insight through their integration

The focus areas of Regulatory Analytics are:

a) Unauthorized trading

b) Market manipulation and abuse

c) Laundering crime

Adopting ‘Regulation as a service’, which provides banks with an analytical and a service model framework, brings innovative thinking to Regulation and Compliance, thereby driving excellence. As Regulation’s strategic role continues to grow, bank’s scale of responsibility is expanding. The resultant broader enterprise risk management can be addressed by fostering digital and analytical innovation to sustain and excel operations in meeting regulatory reporting demands.

To conclude

In order to deal with the rising demands of the fraud and risk ecosystem, banks have made significant progress towards self-learning, intelligent and optimized services via adoption of advanced innovative tools and technologies such as Machine Learning, RPA, Big Data, API, Blockchain and Cloud technologies. This adoption has enabled banks to drive critical business transformation outcomes by augmenting the existing framework leading to an increase in operational efficiency and increased risk management.

References

1.

2.

3.