With technology and innovation advancing at such a rapid pace, it is often difficult and onerous for financial and banking institutions to keep up, let alone stay ahead of the game, which can substantially increase the threat posed by fraud. The risk, according to the Kroll Global Fraud and Risk Report 2019/20, can be attributed to both internal and external contributors. Thus, tactics adopted and implemented to proactively utilize fraud protection and early detection methods are needed by the banking industry more now than ever.
The Scope of Banking Fraud
Financial fraud is costing the world upwards of $5.127 trillion each year according to the 2019 Financial Cost of Fraud report produced by Crowe and the Centre for Counter Fraud Studies at the University of Portsmouth. Jim Gee, Partner and National Head of Forensic Services at Crowe stated, “The figures quoted in the 2019 report are stark. Globally, fraud losses equate to a shocking US$5.127 trillion each year, which represents almost 70% of the $7.442 trillion which world spends on healthcare each year [according to WHO figures].”
And, the report’s author makes note of the fact that because no crime has a 100% detection rate, the cost associated with detected fraud significantly underestimates the true nature of the problem. The types of financial loss the report identifies, and measures includes:
- Payroll
- Procurement
- Housing
- Education
- Healthcare
- Social Security
- Insurance
- Tax Credits
- Pensions
- Mining
- Agriculture
- Construction
- Compensation
Figure 1: Courtesy The Financial Cost of Fraud 2019 report (Available at http://www.crowe.ie/wp-content/uploads/2019/08/The-Financial-Cost-of-Fraud-2019.pdf)
A few of the contributing factors are believed to be greater individualization, more complex processes and systems, more online and digital transactions, the increasing pace of change, and the perception of a lack of transparency among government and business leadership. But ultimately, the banking sector can no longer rest on its laurels.
Payment Card Fraud Worldwide
Fraud losses attributed only to payment and credit cards are projected to rise to $35.67 billion in five years and $40.63 billion in 10 years according to The Nilson Report, a leading global card and mobile payments trade publication. This kind of financial loss is felt by payment card issuers, merchants, acquirers of card transactions from merchants, and acquirers of card transactions at ATMs on all credit, debit, and prepaid general purpose and private label payment cards issued around the globe. According to the report, card issuer losses occurred when, “criminals took over valid accounts, cards were lost or stolen or counterfeited, new accounts were opened with the intent to commit fraud, accounts were opened using a mix of valid and bogus information (synthetic fraud), cardholders or their family members made purchases and then disputed the charges (friendly fraud), and a few smaller categories.” This report drives home the point that credit card fraud detection is no longer a luxury, it’s a necessary staple of any financial institution.
Figure 2: Payment Card Fraud Loss as reported by The Nilson Report (Available at https://nilsonreport.com/mention/407/1link/)
“Sadly, too many organizations adopt a reactive approach to fraud and only look to tackle it once it has taken place, and losses have already occurred. A change of perspective is needed. Fraud is an ever present, high volume, low value problem and only a small proportion is detected. The question is not if it is taking place, but at what level. We need to view fraud as a business cost – by understanding the nature and scale of the cost, we can reduce its extent – enhancing the profitability of companies and ensuring better funded public sector and charitable organizations,” Jim Gee reiterated.
The Proactive Approach to Fraud Prevention and Detection in the Banking Industry
A proactive approach is warranted but not always widely instituted among financial and banking organizations. The first step is identifying the problem. The Kroll Global Fraud and Risk Report 2019/20 classifies the types of incidents that have significantly impacted organizations this past year and they are varied.
Figure 3: Courtesy Kroll 2019/20 Global Fraud and Risk Report (Available at https://www.kroll.com/en/insights/publications/global-fraud-and-risk-report-2019)
The next step that is integral to finding success in fraud prevention and detection is, being strategic while utilizing artificial intelligence and specifically machine learning tools. There are multiple types of machine learning that can aid banks in fighting fraud as detailed in the table below, which may be intimidating at first sight. However, each method of learning has benefits when it comes to technology being able to synthesize and acquire the necessary skills to combat fraud.
| Type of Machine Learning | Category of Learning | Description |
| Supervised Learning | Learning Problems | A class of problem that involves using a model to learn a mapping between input examples and the target variable. |
| Unsupervised Learning | Learning Problems | A class of problems that involves using a model to describe or extract relationships in data. |
| Reinforcement Learning | Learning Problems | A class of problems where an agent operates in an environment and must learn to operate using feedback. |
| Semi-Supervised Learning | Hybrid Learning Problems | Learning where the training data contains very few labeled examples and many unlabeled examples. |
| Self-Supervised Learning | Hybrid Learning Problems | Learning problem that is framed as a supervised learning problem to apply supervised learning algorithms to solve it. |
| Multi-Instance Learning | Hybrid Learning Problems | Learning problem where individual examples are unlabeled; instead, bags or groups of samples are labeled. |
| Inductive Learning | Statistical Inference | Learning that involves using evidence to determine the outcome. |
| Deductive Inference | Statistical Inference | Using general rules to determine specific outcomes. |
| Transductive Learning | Statistical Inference | Used in the field of statistical learning theory to refer to predicting specific examples given specific examples from a domain. |
| Multi-Task Learning | Learning Techniques | Learning that involves fitting a model on one dataset that addresses multiple related problems. |
| Active Learning | Learning Techniques | A technique where the model is able to query a human user operator during the learning process in order to resolve ambiguity during the learning process. |
| Online Learning | Learning Techniques | Learning that involves using the data available and updating the model directly before a prediction is required or after the last observation was made. |
| Transfer Learning | Learning Techniques | A type of learning where a model is first trained on one task, then some or all of the model is used as the starting point for a related task. |
| Ensemble Learning | Learning Techniques | An approach where two or more modes are fit on the same data and the predictions from each model are combined. |
Table 1: Different Types of Learning in Machine Learning (Courtesy Machine Learning Mastery)
Machine learning, in any of these formats, can help to improve fraud protection and detection process efficiency while reducing the cost associated with managing fraud and financial crimes operations. SPD Group, a research, technology consulting, and software product development firm specializes in machine learning and artificial intelligence solutions. These can include:
- Virtual Assistants
- Data-Driven Insights
- Predictive Analytics
A proactive approach to fraud detection and prevention can take any number of forms but should pay close attention to rules for detection and alerts, utilization of the most accurate detection models, and automating investigative processes. Implementation of AI and advanced anomaly-detection models, as well as transaction monitoring with machine learning, should be instituted by technology professionals.
Credit Card Fraud Detection with Artificial Intelligence
While the same tactics that apply to general banking and financial fraud are applicable to credit cards, it is important to note the role that the AI, and specifically machine learning, has played on credit card fraud prevention and detection. In the e-commerce environment, financial fraud and credit card fraud is the fastest-growing issue to face financial institutions. In fact, according to Cisco’s annual Visual Networking Index, machine-to-machine connections supporting online consumer applications will account for more than half of the world’s 28.5 billion connected devices by 2022. In addition, worldwide ecommerce sales were reported by Statista to be right around $1.34 trillion in 2014, $2.84 trillion in 2018, and are anticipated to scale to $3.45 trillion this year. By 2021, it’s projected that global ecommerce sales will inch up near the $5 trillion mark.
In response, not only are consumers having to be more educated regarding the potential for fraud, but credit card fraud detection and protection needs to be more proactive. According to a survey from strategy consulting firm Altman Vilandrie & Company, half of American companies using ecommerce tools have been breached, indicating the not enough is being done to level the playing field.
