Article

Financial crime – are you missing key indicators?

Ravi Joshi Ravi Joshi

The prevention of money laundering is a key concern for every company – so is your internal audit function maximising the use of available technology to do so effectively?

Data analytics and artificial intelligence are now key tools in the detection and prevention of financial crime, but many internal audit functions are not applying these technologies fully and are potentially missing important indicators of criminal activity. Now is the time to assess your existing anti-money laundering (AML) processes and identify key areas where data analytics can strengthen your control framework.

Data analytics can add value across your business - find out how.

What’s the issue here?

Money laundering can often take the form of complex webs of transactions, designed to obscure the criminal source of funds. Internal audit functions have a broad remit and often lack the specialist AML skills required to identify these patterns of activity, limiting the amount of value they can add through expert reviews. Data analytics can be applied to identify anomalies and patterns of activity, which may constitute money laundering, helping you to make the most out of scarce subject matter expert resources.

Suspicious transactions – does your internal audit team know what to look for?

Exploring available techniques

There are four main ways to use data analytics to identify financial crime, including new options such as machine learning and network analytics:

120x120-document-check.pngData mining

Examining a database for specific information or to identify patterns

What’s sort of problem can it be applied to?

High profile cases, such as the Panama Papers scandal, prompt banks to review their records and transaction histories to identify whether their clients have banked, or made payments to, particular entities or individuals or have been otherwise involved.

What’s the solution?

You can use robotic tools to scrape publicly available data and can then search customer transaction records to identify where relevant names appear in your records.

This use of data mining gives you greater control and reduces reliance on commercially available screening lists such as Worldcheck or Dow Jones, which may not carry all of the data you need.

120x120-compliance.png Network analysis

Identifying a network of individuals or entities, which may be linked to common suspicious activity

What’s sort of problem can it be applied to?

Financial crime investigators often rely on manual methods to identify linked parties and build a clear view of money laundering networks.    Automated analysis can supplement and, in some cases, outperform skilled investigators in this task.

What’s the solution?

Network analysis can identify deep webs of linked activity within your customer and transaction records which may not be immediately visible including:

  • Tracing the flow of funds through multiple ‘hops’
  • Identifying entities with similar transactional patterns to check whether they are related
  • Identifying communities of entities or individuals with strong and/or exclusive relationships

120x120-risk.png Rules testing

Testing the logic and assumptions applied to your screening and transaction monitoring processes

What’s sort of problem can it be applied to?

Transaction monitoring and screening solutions need to apply outright exclusions to some entities and individuals (such as those on sanctions lists) and also to allow more sensitive risk-based decisions around certain transactions and counterparties.  Effectively testing that these complex rules are working properly can be difficult.

What’s the solution?

You can create test files containing pre-determined customer and transaction records, run these through a test environment and review the results against your risk appetite and against actual operational results.

120x120-machine.png Machine learning for rule optimisation

Using artificial intelligence to improve the rules used to screen and monitor transactions

What’s sort of problem can it be applied to?

Transaction monitoring and filtering systems generate a significant amount of false positives that require costly manual resource to review and discount.

What’s the solution?

You can use machine learning and statistics to identify the key data points to refine your rule sets to reduce false positives.

What to do now?

If you aren’t already using the techniques above, you should be looking at how to adjust your existing approach to make better use of analytics to improve your financial crime controls. Initially, this is likely to require a detailed review of your data – assessing both availability and quality and taking steps to improve its integrity where necessary – before considering what additional skills and tools to bring to bear.

An optimal approach will deploy these techniques in sequence: leveraging data mining to identify anomalies and possible breaches; using network analytics to follow funds across networks and to build up a clearer picture of patterns of activity; using these patterns to improve your testing of monitoring and screening; and then using machine learning to adapt your approach in line with emerging AML risks.

Finally, using these tools, whether for first line screening or audit, presents an organisational challenge as well as an opportunity.  As well as having a material impact on the skill set of staff involved in financial crime monitoring and the numbers of resources required, there are questions of regulatory compliance and alignment to policy to consider. When will you know that new solutions are performing well enough to decommission old ones? What sort of governance will you need to apply to new tools such as artificial intelligence? It makes sense to think about these issues whilst you are designing any new approaches, to make sure that you can deploy them successfully.

For further information on how best to deploy these tools and for use them in a way that will be satisfactory to your regulator, please contact Daniel Evans or Lucinda Hallan.

Our experts' thoughts on financial services
Cryptoassets and 5MLD - what's in scope? Find out more