10 Best Practices for AML Models in Financial Institutions

Who doesn’t get excited about Anti-Money Laundering (AML) models? We understand, only a few of us do. However, there is something for everyone to be excited about. If the model is performing well, then the thousands of dollars spent to buy and maintain it are being put to good use!

Why do institutions use automated models? Automated models are intended to help take a significant amount of the manual burden to find potential suspicious activity from the human eye and brain. Unfortunately, sometimes the model isn’t performing well and isn’t taking much of the burden off BSA departments.

Here are some best practices to help improve that situation.

1. Review parameters and features each year.
At a minimum, review parameters and features each year to determine if any can be altered to provide more meaningful results. Numerous alerts could mean that the settings are too liberal and flagging too much data. However, that is not always the case.

2. Determine how suspicious activity is being detected for SARs filed.
If the activity is not detected by the AML model, then assess if changes can be made to the model to help detect that type of activity in the future.

3. Take advantage of new system enhancements.
Typically, these are intended to help provide more meaningful information and/or to provide more operational efficiencies for the BSA Department.

4. Make sure that all customer-initiated transaction codes in the core are mapped to the model.
Sometimes we find issues, where a previously retired transaction (tran) code has been resurrected and used for a new purpose or a new tran code has been created, but the BSA department did not know to update the model. A good practice is to make sure that the Accounting and Operations Departments know to inform the BSA Officer about these kinds of changes.

5. Map new products to the model.
If the institution creates a new product, map it to the model to help ensure that the activity is captured.

6. Pay attention to changes to the model, including enhancements.
Whenever changes are made to the model, including model enhancements, test the results for a period of time to help ensure that the output is meaningful and makes sense.

7. Validate the model periodically.
A validation should help the institution assess the model’s performance along with verifying data integrity. We believe this should be done every 2 to 3 years and every time a significant change occurs at the institution or with the model itself. Significant changes include changes in how transactions are processed and placed in the model, buying a new branch (and related customer accounts), merging with another institution, and changing the core processor or model to name a few.

8. Consider having the third party that supports the model perform an efficiency review of the model every 2 to 3 years.
This is different than a model validation. During an efficiency review, the third-party service provider should share common issues being experienced by other institutions, ideas on how to more effectively set parameters and features, and suggestions on how to best use their newest enhancements to the system. Negotiate the efficiency review into the contract terms to help keep the costs down.

9. Allow the model to help risk rate customers
If the institution is not using the model to risk rate its customers, then determine why not, and make changes accordingly. It could be that it is just habit. Or, the BSA department could be concerned about the model’s risk rating ability.

10. Assess Notifications/Alerts

When updating the BSA risk assessment each year, it is also a good time to assess whether the model has produced notifications (sometimes called “alerts”) to help detect activity assessed as medium and high-risk. Consider the best practices mentioned above to make the changes needed in the model to support the Bank’s risk assessment. If this still does not work, consider reaching out to the third party that maintains the model and ask for their guidance.

Institutions spend thousands of dollars obtaining and maintaining an automated AML model to support their BSA program but the results don't always outweigh the cost. Brown Edwards assists numerous financial institutions with BSA model validations, which have resulted in better data for the BSA department so their time can be spent investigating the activity, not just trying to find it.

Contact our financial institutions team today to more effectively identify and manage risk.