When I began my career, back in the late 90s, I was hyper-focused on rule-based technology and methodology. Rules represented our best chance at using automation to monitor transactions and prevent money laundering and other financial crimes.
Two decades have passed, and it’s easy to see how wrong I was. At its very best, a rules-based approach finds one instance of financial crime for every 99 instances that it stops. A measly 1% of true positives, while 99% of its alerts are false positives that keep analysts busy but do little to thwart money laundering.
In place of rules, we are starting to appreciate the value that artificial intelligence (AI) brings in transaction monitoring. From my front-row seat, I’ve enjoyed the role I’ve played in the rapidly evolving fight against financial crimes.
I’d like to take a fresh look at transaction monitoring, and the way it has changed over the last half-decade.
Understanding Transaction Monitoring
Every time money moves from one account to another, we have a transaction. Monitoring those transactions means looking at where the money came from, and where it’s going.
When you think about me sending you $100, the monitoring is pretty simple, but when you start to think about it at scale, with millions of transactions moving across different banks, borders, and currencies, it starts to get complicated.
Hidden among legitimate transactions are illicit transactions, where criminals move money that have been involved in drug dealing, illegal arms sales, and human trafficking.
Regulators demand that banks monitor these transactions to prevent money laundering. That monitoring is important; banks who fail to do so are at risk of millions of dollars in penalties. To make matters more complicated, monitoring goes beyond the current transaction. It requires a knowledge of historical transactions as well and reviewing a customer’s activity and information to get a full picture of the money trail.
The Complexity of Cross Border Transactions
At the heart of a bank’s responsibility is preventing financial transactions from passing through a bank in an international transfer. Money launderers love moving money across borders. For one, it enables them to pay suppliers, but it also allows them to clean money so that it can be used for other purposes.
Despite it being 2021, or perhaps because of it, sending money from one country to another isn’t easy. If you wanted to send money from New York to California, the two banks most likely have a relationship, and the money can easily pass from one bank to the next. However, if you wanted to send money from New York to Johannesburg, South Africa, your money is in for quite a ride.
To protect themselves from falling afoul of regulators, over the last few years banks have de-risked themselves and limited their relationships with foreign banks. The New York bank uses a series of correspondent banks to send the money from the United States to South Africa. The money might go to London first. From there, it may travel to Germany, followed by Greece, and then Egypt. Once in Egypt, the money may move to a bank in Russia, before finally ending up in South Africa.
Each correspondent bank along the trail needs to verify that the transaction is clean, and as the money trail lengthens, it becomes more difficult for the banks to know anything about the sender and recipient. Without effective transaction monitoring in place, any transaction carries with it the risk of onerous fines and penalties from regulators.
What Have We Been Doing the Last 20 Years?
Over the last two decades, we’ve developed automated rules that were supposed to monitor transactions. Rules look for patterns, and when criminals manage to sidestep rules by changing something up, the rules don’t catch up.
To add complication, many of the rules that were used had their roots in retail banking. However, the financial crimes that involve international transfers are different than retail crime, and their financial paths don’t look the same.
To put it bluntly, we’ve spent twenty years using transaction monitoring tools that are rigidly looking for patterns that don’t exist in international money laundering.
The result has been catastrophic. It enabled the Danske Bank’s Russian Laundromat scheme and the Colombian drug cartel to pour money through HSBC for years. Most recently Australia’s Westpac bank discovered it had helped launder $11B for pedophile rings.
Bringing AI into Transaction Monitoring
AI solutions are a powerful step forward in transaction monitoring. Rather than being limited by set rules which can be sidestepped fairly easily, AI looks at all transactions and searches for anomalies in transactions.
When transactions stand out, analysts and investigators are alerted. The transaction can be investigated, and then, either detained or allowed to pass through. Investigators are tasked with far fewer false positives, helping them to be more efficient as they look through questionable transactions.
For the first time, it looks like we have the tools required to stop illicit cross-border payments. That may not be good news for criminals, but it is a positive step forward and offers hope to those who are leading the fights against illegal arms sales, human trafficking, drug trades, and other similar crimes.