How I Improved Threshold Monitoring for BDT 100,000 MFS Transactions with Python

Photo by Sajad Nori on Unsplash

Last quarter, while reviewing a batch of 80,000 MFS transactions, I noticed that approximately 10% of them exceeded the BDT 100,000 threshold, which is a critical limit for reporting suspicious transactions to the Bangladesh Financial Intelligence Unit (BFIU). This experience made me realize the importance of accurate and efficient threshold monitoring in Anti-Money Laundering (AML) compliance.

The core problem most practitioners miss

When I was new to AML compliance, I was wrong about this until I faced a situation where a significant number of transactions were just below the threshold, raising suspicions of structuring. In my experience, many analysts miss the fact that threshold monitoring is not just about identifying transactions above the limit but also about detecting patterns and anomalies that could indicate money laundering or terrorist financing.

Background / why this matters in BD fintech context

The Bangladeshi fintech sector, including mobile financial services (MFS) like bKash and Nagad, has experienced rapid growth in recent years. As a result, the number of transactions has increased significantly, making it challenging for AML analysts to manually monitor and report suspicious transactions. The BFIU guidelines require financial institutions to report all transactions exceeding the BDT 100,000 threshold, as well as any suspicious transactions, to prevent money laundering and terrorist financing.

Technical breakdown

To improve threshold monitoring, I used Python to develop a script that can efficiently process large datasets of MFS transactions. The script uses the following Python code to identify transactions above the BDT 100,000 threshold:

import pandas as pd
# load transaction data into a pandas dataframe
transactions = pd.read_csv('transactions.csv')

# define the threshold amount
threshold = 100000

# identify transactions above the threshold
threshold_transactions = transactions[transactions['amount'] > threshold]

# print the number of transactions above the threshold
print(len(threshold_transactions))

The code loads the transaction data into a pandas dataframe, defines the threshold amount, and then identifies transactions above the threshold using a simple conditional statement. The result is a new dataframe containing only the transactions that exceed the threshold.

Bangladesh-specific application

In the context of Bangladeshi MFS, this script can be used to monitor transactions and identify potential suspicious activity. For example, if a customer makes multiple transactions just below the BDT 100,000 threshold in a short period, it could indicate structuring, which is a common technique used to avoid detection. By analyzing these patterns, AML analysts can identify high-risk transactions and report them to the BFIU.

Common mistakes analysts make

There are several common mistakes that AML analysts make when monitoring MFS transactions:

• Failure to consider the context of the transaction: Analysts should consider the customer's profile, transaction history, and other relevant factors when evaluating the risk of a transaction.
• Inadequate documentation: Analysts should maintain detailed records of their analysis and decision-making process to ensure transparency and accountability.
• Insufficient training: Analysts should receive regular training and updates on AML regulations, trends, and best practices to stay effective in their roles.
• Overreliance on automated systems: While automated systems can help identify suspicious transactions, analysts should not rely solely on these systems and should always review and verify the results.

Counterintuitive insight

One counterintuitive insight I have gained from my experience is that sometimes, transactions that are just below the threshold can be more suspicious than those that exceed it. This is because customers who are trying to avoid detection may deliberately keep their transactions below the threshold to avoid raising suspicions. By analyzing these patterns, AML analysts can identify high-risk transactions that may not have been detected otherwise.

Practical conclusion + next step

In conclusion, threshold monitoring is a critical component of AML compliance in the Bangladeshi fintech sector. By using Python to develop efficient scripts and analyzing patterns and anomalies, AML analysts can improve their ability to detect and report suspicious transactions. Your next step today: review your current threshold monitoring process and consider implementing a Python script to improve efficiency and accuracy.