Transaction monitoring systems have progressed over time to meet evolving regulatory standards, but gaps still exist for criminals to exploit.
In our webinar in partnership with Protiviti, we were joined by a world-class panel to dive into the depths of transaction monitoring systems and discuss big topics including the challenges, gaps, and future requirements:
- Joshua Heiliczer, (former) Managing Director at Protiviti
- Samuel Lok, Head of AML Compliance at OSL
- Natalie Hall, General Manager Financial Crimes Compliance at Commonwealth Bank
- Janet Bastiman, Chief Data Scientist at Napier
- Robin Lee, Head of APAC at Napier (Moderator)
You can catch up with the webinar below or read on for a summary of the key points discussed.
Gaps in transaction monitoring systems open the door to financial crime
A transaction monitoring coverage gap assessment is the foundation of any transaction monitoring programme as it enables the identification and risk rating of any gaps in coverage, in line with a risk-based approach. Following the assessment, it’s possible to identify what solutions are necessary to close the gaps.
The following are commonly found, persistent gaps faced by many regulated organisations in the financial sector:
The data gap
Access to complete and accurate data is make-or-break for the functioning and success of transaction monitoring systems, and is the primary factor that can make both rule-based and artificial intelligence (AI)-enhanced systems more effective.
Access to data is also the biggest downfall of many transaction monitoring systems, which are plagued by false positives because of poor quality data.
Crucially, next-generation transaction monitoring – the kind that features machine learning and dynamic risk scoring – can’t be achieved without data, as AI firstly needs to be trained with data and secondly, requires context to identify anomalies.
Since AI can only provide information on the data it’s been given, it’s important to consider what’s not in the typology as much as what is.
To combat criminal activity, high quality data should be used and shared internally and externally as widely as possible within the constraints of data privacy. Sharing data helps to ensure it doesn’t fall through the cracks and unintentionally give criminals to place to hide.
The rules gap
False positives and false negatives are the single biggest issue faced by regulated entities and rule-based alerts can significantly exacerbate the problem if the rules are too generic or outdated.
While rule-based systems can and do work effectively in the retail banking space, they can also be abused by criminals who know those rules and therefore how to work the system. For example, by making several payments below a certain threshold so that they don’t trigger an alert.
The effectiveness of rules also diminishes as the complexity of the client’s operations increases. This is particularly the case for commercial and institutional banking where it becomes harder to understand what the customer is doing and what is ‘typical’ or expected behaviour.
How AI can fill in the gaps
Where the effectiveness of existing rule-based systems has plateaued, it’s becoming increasingly evident AI is the next step to addressing these challenges.
This is because AI brings unprecedented suspicious activity detection capabilities and efficiencies by finding patterns and the unknown unknowns in transaction data, as well as streamlining compliance processes with intelligent automation to reduce workloads.
AI can detect well-known criminal typologies, those that are only partially known, and also unknown or emerging typologies. Unknown typologies are those that have not yet been defined but may account for most of the financial crime.
Use case 1: augmenting rules
Since AI can detect partial and unknown typologies, complex cases that are difficult to catch with basic rules-based systems can be spotted by AI. For example, AI can identify an increase in transactions correlated with a decrease in payments over a specific period.
Network analysis can also be beneficial for grouping behaviour and reducing false positives.
Use case 2: monitoring behaviour
Customers can be categorised into one risk level at onboarding, but go on to trade in a way that is different than expected. By comparing behaviour over time, AI can spot criminals who behave differently to what’s expected for their segment and raise alerts.
The business value in AI rests in the automation it facilitates. By allowing AI to do the repetitive pattern identification and behaviour grouping work, analysts’ time can be better and more cost-effectively focused on investigating material risks.
How to future-proof transaction monitoring systems
Some regulators are recommending the adoption of AI-enhanced transaction monitoring systems but crucially, these systems must be explainable. Futureproofing with AI that’s explainable – which can be done in the form of a bolt-on enhancement for legacy systems.
Additionally explainable AI is important because it helps end users to understand AI decisions so they can be used in business-as-usual activities. While some people believe you must choose between explainable AI and accurate AI, this is not true: you can have both.
A platform-based, holistic approach to compliance is also essential and so the next generation of transaction monitoring should feed into other key financial crime compliance modules such as client screening and transaction screening, and other functionalities like client risk scoring and client activity reviews.
A centralised, comprehensive platform can provide unsurpassed client insights and enable perpetual KYC and 360-degree client reviews.
Discover next-generation financial crime compliance technology
If you are looking to learn more about the next generation of transaction monitoring, book a demo of our solutions, or get in touch to find out how Napier can rapidly strengthen your anti-money laundering defences and compliance capabilities.
.png)
