Big Data: Uncovering Newer Dimensions in Fraud Preemption

Financial services have been vulnerable to fraud ever since the days of barter trading. As the renowned corporate thinker Alvin Toffler opines, the third wave has proliferated the nature and means of delivering financial services spanning the entire spectrum of services such as mortgages, checking, savings, cards, asset management and international transactions. With these proliferations, one parameter that numerous organizations are exploring to fight fraud is “big data.”

With advancements in both the quality and quantity of data as well as financial services and technology, fraudsters have also kept up with the times, developing more sophisticated techniques over time.

An increasing number of financial services organizations have started discovering the power of new data management technologies to fight fraud. With multiple devices and types of transactions proliferating, the challenges involved in fraud detection have also increased. In order to detect fraud and identify security breaches geospatial data from smartphone apps, customer behavior from social media, weblog data from the organization’s online channels, state-of-the-art fraud detection models, and more, have to be stored and analyzed along with data form core operations. The below diagram provides a high level view of the normal business architecture in an anti-fraud mechanism in financial services organizations.

Recently financial institutions have been developing end-to-end capabilities to leverage data mining, modeling and analytics to define specific patterns in data sets and flag anomalies in real time. With real time detection and remediation, companies are rediscovering their ability to altogether stop or quarantine and re-route suspicious transactions. Financial institutions have a lot more data at their fingertips to track both external frauds (involving customers, account holders, policy holders) and internal (employees) ones.

Credit card companies, which deal with a large number of transactions and enormous data, have been doing fraud detection well before the big data term was coined. But fraud detection is still a new use case because big data democratizes fraud detection and there’s ample room for interoperability of structured and semi-structured data and analytics. With big data, a financial institution doesn’t really need to have the massive scale of financial and IT resources to detect fraud.

An interesting example of interoperability in the area of data and analytics is that of banks and credit unions joining hands to tie data and touch points – in order to elevate the level of customer experience. Both firms can derive meaningful Big Data analytics for their business cross-play in the detection and prevention of fraud. Such interoperability models and diverse channel integration not only strengthen the fraud detection and prevention mechanism; it significantly optimizes the cost of managing large scale data as well. This enables FS organizations to focus on specific analytics and leverage each other’s information for a suitable operational cause based on predefined SLAs.

Most credit bureaus today have the ability to provide customized credit ratings/scores for individual account holders. If the banks religiously follow the end-to-end process – enhance their data collection; use fraud detection modeling strategies; routinely train, validate and test the models; analyze results and iterate to implement results using transactions and customers’ credit bureau ratings/scores – they will be able to significantly improve their fraud detection rates on new and existing accounts. At the same time, banks can track the consumer behavioral patterns, which can be shared with credit bureaus to add a meaningful dimension to customized score cards.

Loans and mortgage companies have started leveraging their big data programs to detect external fraud by mining social networks for suspicion of fraud. For example, a person who has been defaulting on his loans due to non-availability of funds, but keeps posting images of his new purchases (car, consumer durables) etc. on his social media profiles, can be brought within the purview of scrutiny.

By gathering data on withdrawals, transactions and the exact locations of these transactions, financial institutions will be able to thread together specific patterns of suspicious transactions. When one combines transaction data with the social media communication tools of semi-structured text or mobile locations, financial institutions can immediately detect and flag potentially suspicious transactions.

Coming to internal fraud, let’s consider instances such as rogue trading, embezzlement of funds, evading policies etc. In the high profile case of a French rogue trader, a leading French bank lost approximately €4.9 billion closing out positions fraudulently established by this trader. While the trader knew how to hide his deeds, the bank could have got a whiff of it at the opportune time only if reasonable flagging pointers were available then.

The concept of tracking individuals rather than accounts, while appealing, could prove extremely challenging. Most financial institutions today aren't really set up to monitor users; they just track account activity. But as more and more institutions actively start implementing practices to track user behavior, the tracking exercise per se might not end up being a stretch. A lot will depend on how much banks want to invest in solutions and services offered by third-parties, which often revolve around ID theft protection. For most bankers, ID theft monitoring doesn’t seem to be the topmost priority at the moment; account monitoring is.

More than ever before, FS firms are cognizant about the negative impact of fraud – given that their reputation, customer loyalty, shareholder confidence, etc. is at stake. Rapidly growing channels like mobile and online banking are also at risk. Moreover, given that banks rarely monitor customer behavior across multiple channels, fraudsters exploit cross-channels because of the difficulty to discover, track and resolve – as more often than not, fraudsters pose as customers, with their activities closely mirroring that of customers. With the volume of transactions increasing year-on-year and steady expansion of customer base in the new/rapidly growing channels, it’s time financial institutions need to start combating fraud with the best of breed technology available today.

Sandip Sinha is associate vice president, financial services practice, and Deepesh Patel is director of business development, at Collabera, an end-to-end IT consultancy and solutions provider.