We are losing the battle against financial crime, and it’s the banks that are facing the consequences. 90% of Europe’s biggest banks have been sanctioned for money laundering, and the number of cases just keep growing.
Reports indicate that the global cost of financial crime has topped $3 trillion, with consumers losing nearly $1.5bn to fraud last year. Big numbers – but even these figures are just estimations. We have no idea how large the scale of the problem really is due to critical limitations in current financial crime detection systems.
Banks are paying the price
- In 2019, ING group was fined $900m for failing to spot money laundering
- Deutsche Bank paid out $630m for failing to prevent Russian money laundering in 2017
- Nordea was fined $6.1 million for breaching laws on money laundering and terrorism financing in 2015
- Back in 2012, HSBC Mexico was hit with $1.9bn in money laundering penalties
The problem of improving financial crime controls
Money laundering and credit card fraud cases are increasing rapidly worldwide, so why are current FinCrime detection systems still unable to prevent and detect real crime when it happens?
Right now, there is no robust measurement to evaluate how effective an institution’s fraud and AML system is at correctly identifying financial crime cases. Here are some of the key reasons.
Historical data is not enough: Current approaches to financial crime analytics, such as false alarms optimization, are limited to classic techniques that use historical data to analyse and improve their performance, meaning that progression is limited to what’s happened in the past; not what might happen in the future.
Complexity to model real world environments: Simulating financial transactions is a real challenge. Poor models lead to serious calibration issues that turn the results into useless data.
Unexpected fraud behaviours and rare events: The most common method today used for preventing illegal transactions consists on flagging different clients according to perceived risk and restricting their transactions using thresholds. But, as with other types of fraud control, fraudsters simply adapt their behavior accordingly.
Data privacy issues: It’s difficult to find enough diverse cases of labelled financial crime, particularly without exposing business or personally sensitive information to financial, privacy, legal and contractual issues (as well as GDPR and other regulations).
The problem with false positives and hidden crime: Current financial crime control methods focus only on two things to measure the improvement:
- Reducing the number of innocent people wrongly tagged (False Positives)
- Increasing the number of criminals flagged (True Positives)
Yet current financial crime detection systems do not consider the False Negatives (non-detected criminals), because this information is unknown. If you can’t measure to what degree your system fails, how can you be effective in driving improvements?
However, there is a novel approach in financial crime detection which involves the use of simulators to produce enough financial data which contains both the normal behavior and the fraud behavior. There are numerous research papers that have been produced in this field.
Next generation FinCrime analytics
Thanks to this research, we now know that with the aid of advanced FinCrime analytics tools such as simulation, we are now able to generate information about the missing crime (False Negatives) and calculate the reduction of financial crime. This significantly changes the way we do things now and gives us a new horizon to improve current systems.
With Simudyne, organizations can move from a reactive to a proactive approach to fighting financial crime. They can reduce the cost of fighting crime, retain profits, build and maintain trust while drying up funding available for criminal and terrorist enterprise.
