In recent years, the Banking and Finance Service (BFS) Industry has seen a dramatic change in regulatory conditions which necessitates the use of Predictive Analytics for banking and financial service industry. Application of third party Analytics Service in Banking is growing rapidly but as of now it is still at a nascent stage. Advanced Analytics providing sophisticated solutions have just begun to be implemented. The main areas for Predictive Analysis are Customer and Marketing analysis, Product Optimisation, Risk Management, and Fraud Prevention. Due to some major shifts in the Banking industry it has become imperative to apply predictive analytics for remaining profitable and competitive. Some of these changes are –

  • New regulation policies have emerged such as FATCA, FCPA and FINRA. Compliance to them through conventional means would require more human resources but analytics can give automated solutions to keep the processes manageable.
  • Conventional banking products, such as checking accounts and fixed deposits, are losing their relevance because of other new products available. Predictive Analytics provides the framework for marketing and management of all products.
  • Due to the volatility of markets, such as real estate and international trade and investment, analytics are required to find insight into credit and liquidity risks in real time scenario for immediate implementation. They help in the decision making of how much market exposure should be given.
  • The increase in fraud cases requires a more powerful automated analytics system which can give actionable intelligence to point out fraud cases and provide quick response to genuine claimants.
  • Staying competitive is very important when there is a lack of product differentiation and customers have many choices. Predictive Analytics gives the edge when processing vast amounts of data and giving new models for customer engagement.
  • By opening new access points, such as the internet, mobiles and ATMs, vast amount of data is being generated which is only growing with time. Analytics is the only method to handle this data and has the capability to give real-time solutions.

Risk Analysis and Fraud Prevention

Analytics Models help Banking and Financial Service (BFS) Organisations to assess all portfolios and investments for likely losses and risks. According to latest regulations, Risk Management requires a greater deal of transparency and accountability for using Analytics Models for automated actions. Also, Risk analysis and management have now extended to every stage of the financial lifecycle. Advanced Analytics provides actionable solutions which reduce over compliance on KYC departments. With the help of Analytics, many irrelevant instances can be kept aside and human resources can be used for more serious cases. Risk analysis improves business profitability and reduces legal cases. Analytics is a key for identifying patterns of fraudulent transactions on a wide amount of data. Predictive Analytics can provide solutions and insights in a number of ways –

  • Customer account validation through a watch list, reporting suspicious activity at the geographic level.
  • Calculating patterns for fraud transactions.
  • Testing against set risk parameters to report business result forecast.
  • Regulatory demands such as Dodd-Frank, SCAP, FATCA and Basel III require the use of predictive analytics for risk analysis.

Fraud Prevention in Insurance Companies

Insurance companies have been using sampling methods to detect fraud and often they are unable to detect cases because of their conventionally set watch lists. By using conventional methods many parts of the gathered data remains unanalysed, causing gaps in the investigation. Moreover, with the expanding channels of gathering data on clients, conventional methods are unable to cope with this vast information base. Predictive Analytics give the power to sample every bit of data and adapt to new channels. Some of these techniques are:

  • Filtering of obvious cases and referral of low incidence cases for further analysis. Allowing better utilisation of fraud experts.
  • Widening the perspective throughout the organisation. Since various channels are used for data gathering including mobile and the internet, it helps involve the various access points. Fraud can happen at various stages – premium, application, employee related and third party fraud. Using predictive analytics at these various stages helps in finding frauds at different points.
  • Many times important data is present outside the organisation such as public records and third party enterprises. Predictive Analytics helps in integrating data from all the sources. There may be a criminal case, fraud or bankruptcy to give an idea of a particular portfolio. Also, third party agencies can help in confirming the extent and matching of the damage as filed in claims. In the case of medical bills information is important to find if there has been any fraudulent activity. Considering all the agencies involved including social media, Analytics can make fraud management more powerful.
  • Unstructured textual data from social media sources may not be stored in primary data storage but their analysis is also important. Analytics helps in reviewing all the data and pointing out important factors.

There have been advancements in using new methods of Predictive Analytics for fraud detection. Some of these methods which are now gaining prominence are –

Social Network Data Analysis – There may be instances where there might be a mismatch of information provided for claims such as personal information or claims registered earlier. It is important to analyse all this information and find Nodal points which can raise a detection of fraud. Social Media provides a vast amount of data which can be analysed to see if there is any discrepancy pointing to a fraud. Some parameters which need to be seen beforehand are how soon the data arrives, how clean it is, how much depth needs to be covered for analysis.

Predictive analysis of social media is done by applying different methods together. Some of them are – statistical method, pattern analysis and network analysis all done to create a full profile underlining aspects important for a case. Public records information plays a critical role in categorising the risk of fraud considering previous cases of a criminal offence, foreclosures, bankruptcies and changes in personal information. By assessing all this data Analytics can categorise claims according to their fraud risk probability. Some of the steps for sourcing Social Network for Data Analysis are –

  • The data is extracted from various social network sources and fed into data warehouses. Data might be structured or unstructured it is important to segment the data for further analysis.
  • The analytics team assesses the likelihood of fraud based on multiple parameters. Some of them are checking the relations of a claimant on a wide network and find risk points, check with other agencies for previously rejected claims and checking public records for any discrepancy or modification of personal information.
  • Text analysis and content analysis are also being used to give a microscopic analysis of social network information.
  • Calculating risk probability and raising notifications.
  • Once fraud cases are identified their patterns are used for further scanning of data for similar activity.
    Interacting with customers on different stages of customer lifecycle helps in giving greater transparency. It gives an area for big data analytics to gather all the data across various platforms and come up with better insights. This is most effective when the customers are engaged right from the very beginning through all stages of the customer lifecycle.

Credit Scoring Using Predictive Analytics

Credit Scoring is an important tool in deciding credit cards, loans and credit limits. Customer scorecard is a sum of all the points on the different characteristics of a profile, such as residential status score points and region score points. It has many benefits like objective risk evaluation, lower decision cost, less time gap for extending credit and better portfolio performance. However, conventional credit scoring methods suffer from high cost and time for maintaining a credit scoring system has a limitation on the degree of precision. Predictive Analytics provides improved capabilities for credit lifecycle management giving better customer retention and acquisition. Credit Risk Management now covers the whole credit lifecycle after new accounts are created they are looked after by retention and behavioural models of PA.

  • Launching marketing campaigns on verified targeting options for higher engagement.
  • Optimisation of interest rates charged on loans and adjusting credit limits.
  • Study customer behaviour to find acceptability to retention processes.
  • Better credit card account management after analysis through issuing or declining line increases, reissue of credit card and give better rates.
  • Suggesting actions for customer accounts with real-time analytics.
  • With the increased inventory of financial products, multiple channels and more complexity in risk assessment Predictive Analytics has become an important tool to gain actionable insights.
  • New solutions for collection and recovery strategy.
  • Best suited actions for accounts which have overstepped rules and guidelines. Many different methods are available for recoveries such as legal, call centre communication or doorstep.
  • Predictive Analytics offers insights which are easy to interpret from complex data analysis models, which corresponds to better decisions in lesser time. Decision makers do not require knowledge of statistical studies to understand information.

Some points required to effectively apply Predictive Analysis to Credit Scoring are –

  • Data segmentation and profiling. When handling vast amounts of data it is important to be able to demarcate it into different sections for better analysis.
  • Along with Analysis models, there need to be tools to validate the result as well, which will give more accurate results.
  • Direct action capability, in order to be effective decision makers, must include predictive analysis of credit scoring in all their actions to come up with comprehensive results.
  • Access to multiple channels for all round data gathering.
  • Evaluating different models and calculating which one will provide best results.

Future Advancements in Predictive Analytics

More advanced technology which will be able to self-program new models will be coming up in a few years. This will lead to further streamlining of results with lesser instances of false alarms. Using third party agencies to work alongside the models used will lead to further refinement of processes.

Keywords: Predictive Analytics for BFS, Predictive Analytics, Advanced Analytics, Product Optimisation, Credit Scoring, Customer scorecard, Credit Risk Management, Risk Management, Risk Analysis, Fraud Prevention, Fraud Detection, FATCA, FCPA, FINRA, Dodd-Frank, Social Network, Text analysis

Also Read:

About Quadrant Knowledge Solutions

Quadrant Knowledge Solutions is a global advisory and consulting firm focused on helping clients in achieving business transformation goals with Strategic Business, and Growth advisory services.

At Quadrant Knowledge Solutions, our vision is to become an integral part of our client’s business as strategic knowledge partner. Our research and consulting deliverables are designed to provide comprehensive information and strategic insights for helping clients formulate growth strategies to survive and thrive in ever-changing business environment.