The danger is always there–more so now for enterprises in banking and e-commerce. But are automation and Machine Learning helping or failing here? Sumit Chadha CIO–India Mortgage Guarantee Corporation (IMGC) explains how technology is turning into an asset worth investing in. Interestingly, IMGC has been rolling out a lot of changes and making a good return on technology. For instance, automation in areas for under-writers, for claims management, and for DMS. Also, it introduced RPA technology for data entry, which led to faster and accurate data entry and saving of 15 mins per file (Processing time per file). Let’s see how deeper all this goes and is there room for Blockchain, hyper-personalisation, and exception-based human role already?
Can you elaborate on what gaps/problems did initiatives like RPA and automation solved? And how well are they performing?
Before introducing the bots, a significant amount of time was used to obtain data from the lender’s system/Credit Appraisal Memo and transfer it to our onboarding system. We have efficiently utilized our workforce for decision-making rather than monotonous data input since the deployment of bots. Also, when the number of bots deployed grows, the efficiency of the deployed bots grows, and we can exploit the potential of automation fully. In the future, we want to install more bots or link with Lender systems via APIs (based on agreement with lender partner).
What has technology done to elevate claims management and underwriting?
TAT (turnaround time) becomes an essential element for us because we also underwrite the loan application once the Lender partner has underwritten the case. We could significantly automate our underwriting and claims administration processes by using a Strategy Management System/Business rule engine. The strategy management system recommends the underwriter if the loan application complies with the credit policy or claims policy norms. The underwriters are just reviewing the exceptions, which saves them a lot of time.
AI/ML has begun to play an essential role in the mortgage sector in areas like risk assessment, document management, fraud management, detecting earlyindicators of delinquencies.
Can AI or Blockchain help fix critical issues in the mortgage industry—like transparency, fraud, risks, financial inclusion, customer safety, NPAs, etc.?
AI/ML has already begun to play an essential role in the mortgage sector in areas like risk assessment, document management, fraud management, and detecting early indicators of delinquencies. The usage of blockchain in the mortgage sector is yet to be established, but there is a great deal of discussion going on about it right now.
What has changed fundamentally about industries like mortgage and actuary in the pandemic phase?
Many firms have moved away from the old methods of verifying a client through physical or phygital mode and towards digital verification and digit/paperless consumer onboarding. The pandemic has accelerated the speed of automation, propelling IT to the forefront of business. Organizations that focus less on technology or automation have been pushed to accept new technologies and move quickly to remain competitive in the market, which is ultimately beneficial to the organization or the industry and the country as a whole.
Has the job of risk analysis become easier or more challenging now? How much can AI and analytics help make underwriting and risk analysis stronger?
The use of AI has undoubtedly made the job of risk managers simpler, and AI plays a vital role in strengthening risk analysis. For example, our in-house bureau score analysis methodology (built in-house) assists underwriters in making decisions on the overall bureau report after assessing each of the tradelines in the report and their individual monthly performance.
Have algorithms got better in forecasting and precision? What gaps—like poor or inadequate data, fraud, wrong binning approach, modeling errors, etc—confront them?
Organizations have been able to foresee better the outcomes of different circumstances such as part-prepayments, foreclosures, and NPAs, among others, with higher precision than ever before because of the use of big data and data science. However, there are still obstacles that data scientists encounter, such as data preparation and purification, managing data from many systems, data security, and so on, which offer a lot of challenges for data scientists before they can come up with a forecast or comprehensive insight. Even though the industry has developed answers to these issues, the concerns have not been fully resolved.
What are your views on decentralization and hyper-personalization? Would these forces become staple factors soon?
Gone are the days when you could simply categorize your customers and create generic marketing campaigns. The use of mass marketing strategies is no longer acceptable. Instead, retail shoppers demand personalized communications and products tailored to their specific needs. A customized experience fosters goodwill, boosts loyalty, and puts a smile on the customer’s face. Most retail firms have begun to use personalization (if not hyper-personalization), and I am confident that in this era of technology, where the world is scurrying, hyper-personalization will gain popularity in the near future and will be crucial for most retail organizations.
What impact, and how soon, can Machine Learning make in reserving, fraud detection, and risk prediction?
The Cambridge Dictionary defines fraud as “the crime of obtaining money by deceiving people.” It predates mankind. There has always been a danger of one party defrauding the other since people began exchanging commodities and services. And there has always been the possibility that a third party would defraud both the vendor and the customer. Because of the growth and proliferation of e-commerce, fraud has taken on new forms and grown more potent than ever. As the popularity of e-commerce, online banking, and online insurance grows, fraudsters exploit every flaw in every system they can uncover. Often, critical data is stolen, and millions of dollars are withdrawn before professionals can patch up a system. Fraud has become a serious concern and an uncontrollable cost for e-commerce firms on a worldwide scale.’
Preventing, detecting, and eliminating fraud is currently among the top priorities for e-commerce and banking businesses. Machine Learning development services are amongst the most promising techniques for achieving them.
There are still obstacles that data scientists encounter, such as data preparation and purification, managing data from many systems, data security, and so on, which offer a lot of challenges for data scientists before they can come up with a forecast or comprehensive insight.
ML has already been used to detect email spam with success. It also makes focused product recommendations to millions of online shoppers. The availability of Big Data allows Machine Learning to develop at a grand scale and improve significantly over a short period. ML is making inroads into the e-commerce and banking sectors thanks to advances in statistical modeling and rapid processing power. These industries have high aspirations for better fraud detection utilizing machine learning as a tool to prevent cybercrime.
What next are you eyeing to deploy? What lessons from your BFSI stints at ICICI and PNB Housing Finance would you bring in for the next roadmap for IMGC?
As a further step in automating underwriting operations, we will consider introducing an Intelligent Document Processing (IDP) tool to reduce time in extracting data from the properties’ legal and technical verification reports. This move will be the final stage in the end-to-end automation of our underwriting operations, leading to the total automation of IMGC’s underwriting processes. I have a techno-functional background and want to push technology to the forefront of business by automating most operations through technological interventions. I’d also like to deploy the finest of the market’s solutions, taking a cue from what I’ve learned at PNB Housing Finance Limited.
By Pratima Harigunani