Increased digitization has influenced various businesses and brought in a paradigm shift in the way they create business models and approach growth opportunities. Moving away from conventional transaction modes such as cash and checks, banks and FIs across the globe are increasingly adopting digital means of payments. Technology also paved the way for many digital payment methods such as pre-paid cards, mobile payments and internet payment services, commonly referred to as New Payment Methods (NPMs).
In some countries, they have emerged as viable alternatives to the traditional financial system. While these NPMs enhanced the ease and speed at which funds are accessed and transactions are made, they opened up new avenues for fraudsters to abuse the financial system and clean illicit funds. In this article, we will explore the market transformation in digital payments ecosystem and impending AML risks, which can be detected proactively and mitigated through a tech-enabled transformation.
NPM Market Expansion
Over the last 10 years, digital payment methods have seen a significant rise in popularity. The period from 2012 to 2015 saw a surge in the adoption of credit and debit cards. According to the 2016 Federal Reserve Payments Study, the number of debit card payments (including payments with prepaid and non-prepaid cards) grew to 69.5 billion in 2015 with a value of $2.56 trillion, up 13.0 billion or $0.46 trillion since 2012. Meanwhile, the number of credit card payments reached 33.8 billion in 2015 with a value of $3.16 trillion, up 6.9 billion or $0.61 trillion since 2012.
The period from 2016 to 2019 witnessed the emergence and growth of a number of mobile payment, internet payment and P2P payment technologies such as Apple Pay and Zelle. From 2016 to 2022, mobile payment transaction volume in the US is projected to grow 36.6% year over year, according to the State of Mobile Payments in 2019. The COVID-19 pandemic has accelerated the growth momentum of the NPM market. As consumers altered their payment behavior, e-wallets and contact-less payment methods have become the new normal. According to estimates, Apple Pay is on pace to handle 1-in-10 global card payments worth trillions of dollars a year by 2025.
Rising AML Worries Related to NPMs
Today, banks, irrespective of segments (retail, commercial or correspondent), are facing AML risks of varying complexity from NPMs. There have been many instances of the suspected use of open-loop cards and online payment systems to launder drug proceeds, laundering of illegal gambling proceeds through prepaid cards, use of crypto exchange and fake support groups to fund terrorist activities. In 2017, FinCEN fined virtual currency exchange BTC-e $110 million for facilitating ransomware and Dark Net drug sales. In yet another instance, the regulator fined Merchants Bank of California $8 million for violations of the Bank Secrecy Act related to transaction laundering.
Along with the growth in transactions via NPMs, it is natural to expect a rise in the use of these new-age payment techniques to clean illicit money or to channel them to fund crimes. However, institutions and their systems are seemingly finding it difficult to identify the abuse. Our analysis of Suspicious Activity Report (SAR) filings statistics released by FinCEN found that NPM-related SAR filings remained almost flat during 2017-2019 at about 14%. While clear regulations to address these emerging AML risks are still evolving in many jurisdictions, the current situation raises serious question—how can financial institutions ensure compliance and alleviate the “Fear of Unknown”?
Role of Machine Learning in NPM-related Money Laundering Case Detection
Customer behavior is indeed changing, and as financial institutions expand their digital operations, they are bound to encounter new forms of AML risks. A tech-enabled transformation can help financial institutions capture more than twice as much value as organizations are focused on applying tactical technology measures such as upgrading current systems, buying ad hoc tools for a specific use case or applying AI platforms and spending months to build models internally and finally throwing these away as the models are unable to hold performance in the real world with data shifts.
As a RegTech company, Tookitaki has always focused on building an end-to-end machine learning analytics solution that can be rolled out at an enterprise level across any organization. Our AML analytics solution addresses varied needs of CDD/EDD, transaction monitoring, sanctions and payments screening. Our innovative approach has proven to accurately detect new cases and safeguard our clients from reputational and financial risk. We combine the novel concept of a typology repository with automated model management to detect missed true cases and ensure compliance within our client fraternity.
Our Typology Repository Management (TRM) framework uses federated learning, an advanced machine learning technique that promotes learning in a collaborative but decentralized manner, maintaining data privacy. The key features of TRM are explained below.
- It is a growing library of money laundering patterns shared by AML experts, regulators and financial institutions. We define a money laundering pattern as a technique involving four key indicators including customer, counterparty, network and transactions that signal suspicious activity. The pattern captures the multiple touchpoints of the customer, without sharing any personally identifiable information (PII) and threshold values.
- It includes money laundering patterns across geographies, products, customer type, technique and predicate offence.
- It comes packaged with our transaction monitoring module, where our clients can ingest money laundering patterns that suit their business needs in one click.
Our Automated Model Management framework converts money laundering patterns into machine-readable inputs to detect new cases. The key features and benefits of the framework are given below.
- It facilitates continuous learning in an automated manner, retaining the performance of the machine learning models as the data shifts. This way, the compliance teams can ensure the models used are relevant to the business conditions and at a low maintenance cost.
- It helps move away from hand-designed models to automatically optimized pipelines that take care of automatic creation of risk indicators or input variables, model optimization and validation.
- Our approach reduces 500-800 manhours of model building effort followed by an equal amount of model retraining effort per cycle.
Overlaying advanced machine learning on crowdsourced and automatic extraction of AML patterns, we at Tookitaki were able to detect new methods of money laundering accurately. We have seen up to 5% lift in total SAR filings from new case detection within client environments.
To know more about our AML solution and its unique features, please write to us and we will be happy to give you a detailed demo.
It is the right time to drive tech-enabled transformation as financial institutions are already walking the data transformation journey with creating a strong foundation for data hygiene & soundness, data mapping, extraction and pre-processing. These measures will help accelerate the adoption of machine learning models and facilitate data validation and governance to ensure glitch-free and futuristic compliance programs.
Content that might peak your interest
