Digital India is great initiative with vision to transform society in digitally empowered society. Keeping up with the current trends in technology and economic scenario in India, the decade of 2015 to 2025 would probably be termed as the “digital revolution” in India. Everything is getting Digital from individual validation, communication channels to transactions in physical currency. There has been major shift in the way even the currency is being perceived today. It would be really exhilarating to see how the physical currency would be replaced by the digital transaction. With the advent of Unified Payment Interface System, payments banks, mobile banking, etc, physical currency would to some extent be replaced by digital transaction.
While there is probability that fraud involving physical cash may reduce, there is equal likelihood of vulnerabilities to increase in the digital world. However the same comes with certain advantages in the form of Digital Footprints, which the analyst and investigating professional can leverage upon.
There would be significant rise in the amount of data that will get generated over the next decade in all formats. These frenetic pace and voracious streams of information in a highly mobile environment have created dangerous pitfalls and thus increasing the need for analytics for better insights.
AIM Daily XO
Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
As per Global Fraud Study survey 2016 conducted by Association of Certified Fraud Examiners (ACFE); victim organizations that lacked anti-fraud controls suffered greater median losses. Analytics would play a pivotal role in fraud management process.
Understanding Fraud Triangle through data analytics:
Download our Mobile App
Application of analytics in areas of Fraud Risk management framework:
Detection of Fraud:
Analyzing heap of data and deriving meaningful insights out of it has become crucial in present digital world. Use of Benford’s law may be evaluated and applied to the population. Other fraud detection techniques may range from deep diving ad-hoc analysis of existing information, pattern analysis, network analysis, outlier monitoring, and behavior analysis to developing some of the complex algorithms and models.
There are three critical aspects in any fraud investigation; understanding of the issue, forensic analysis and impact assessment. Data analytics can provide vital support in the investigation process. Investigating a fraud requires identifying root cause and evidences to prove the case. With analytics, traces left by fraudsters in the digital environment can be identified and adequate evidences can be collected.
- Building analytics solution by developing real time triggers for exception scenario. This would again require developing of model for behavior analysis of patterns followed by necessary action on the report.
- Early warning system – Providing indicators of probable vulnerabilities which may lead to Fraud and may need further analysis before conclusion.
Fraud Analytics Maturity Model:
It is very important for any organization to continuously assess them in line with fast paced changing business landscape. Below is the ‘fraud analytics maturity model’ which can provide some insights on how far is organizations maturity level in acceptance of data analytics in its fraud risk management process.
Stage 1: Receptive: Known from others.
Fraud Response based on complaint either through internal or external. No use of analytics for concluding the event
Stage 2: Receptive and Scalable: Known from others but replicated for exposure.
Based on Response received, data analytics carried out to see how the same is impacting overall exposure and taking corrective action.
Stage 3: Pre-emptive: Self-identified.
Self-identification of the fraud based on seeding/assessment of controls through analytics. This would include deployment of analytics for continuous monitoring of controls and focused sampling.
Stage 4: Preventing and predicting:
This would involve developing an automated system and methods to detect fraud using a predictive model. This may involve analysis of various variables and its correlation with the desired output. An early warning system can be put in place to identify potential frauds.
Five step approach to fraud analytics:
- Hypothesis building: Understanding fraud vulnerabilities and developing comprehensive fraud risk register is crucial before one embarks on the analytics journey. This can be broadly classified in three categories Corruption, Financial loss and financial misstatement.
- Mapping available data: Another important aspect in analytics is understanding the data structure and linking the same with the underlying hypothesis.
- Designing algorithm: This refers to exploration of suitable analytics techniques which can provide desired results. This includes developing support tools for monitoring.
- Testing and validation: It is inherent for any model or system to be validated before deploying the same.
- Deployment and continuous monitoring: For a successful Fraud analytics framework there has to be proper decision making and monitoring mechanism.
Challenges in Analytics:
- Data availability and hygiene issues: This is one of the by far the biggest challenge in fraud analytics. Without availability of data, it would be difficult to have good analytics framework.
- Handling big data: Big data comes with its own set of characteristics in the form of volume, velocity and variety.
- Skill set: Fraud analytics would typically require blend of three skill set – understanding of fraud vulnerabilities, knowledge of data structure and strong analytic techniques. Acquisition of such skill set may sometimes become a challenge.
Frauds will increase as transactions volumes would increase with continued digitization. Analytics not just supports marketing and sales but in growing digital scenario it has bigger role to play in risk and forensic analytics as well. The world is changing and with everything becoming automated and digitized, it is inevitable to embrace analytics in forensics and fraud detection.