Data automation in mortgage lending was a frequent topic that was discussed at many recent mortgage technology and business conferences.
Mortgage lenders collect tons of data on every loan application from every borrower. The insights gathered from such data make loan processing more efficient. Every player involved in the home ownership journey – be it a lender or an insurer – should visualize the role of data. Though the pros of collecting and leveraging data can strategically outweigh any risks or disadvantages, lenders need to be aware of regulatory constraints and technical challenges that arise when dealing with data.
Data-driven decision-making (DDDM), apart from addressing customer-specific requirements, also drives initiatives to foster innovation and efficiency at scale. This white paper discusses the key characteristics of mortgage data and the various use cases of how DDDM creates a frictionless home ownership journey for all stakeholders through a human-centric approach.
Data challenges in the lending business
Every year, lenders invest trillions of dollars to modernize their business
While there is an increased focus on digitalization and automation of processes, a prevalent lack of data culture to support such investments impacts business metrics such as cycle time, approval rates, and more.
Lenders have access to the complete financial DNA of every borrower applying for a loan and as well as comprehensive details of the properties intended for purchase or refinancing. This data reveals insights, trends, and patterns which can be efficiently and effectively utilized to manage their business and meet ever-changing customer expectations. However, the collected data comes with its own challenges. The top three challenges with mortgage data are:
Large quantities of siloed data: A typical mortgage loan application has at least 250 financial, personal, and subject property data points. With millions of borrowers applying for loans every year, lenders have access to huge data sets on borrowers and properties. This data is stored in different systems and formats, which makes it difficult to visualize.
Multiple formats from different channels: Data collected by lenders can be both structured and unstructured. For instance, documents such as pay stubs, bank statements, and lease agreements may come in as scanned copies, images, or PDF documents through various channels. These documents then need to be read through optical character recognition (OCR) systems to get digitized.
Continuous ingestion of data throughout the life cycle: Lenders receive data continuously both in real time and in batches from multiple players in the ecosystem. Similarly, they receive data from different types of services from various providers and vendors, and data flows in many directions at every stage of a loan application.
Can data-driven decision-making transform lending?
Data is a significantly important factor in understanding the lending business.
Lenders can get a 360-degree view of customers which can be used to personalize offers and tailor-fit prices based on unique individual needs, within regulatory constraints. With data, lenders and servicers can analyze the loss risk on loan portfolios, including prepayment and delinquency probabilities and plan proactive actions accordingly. To that end, lenders can implement data-driven decision-making (DDDM) in a multi-step approach to find the right insights or the who, what, where, when, and why in order to make the best decisions. The steps are as follows:
Identify the business problems to be solved or objectives to be achieved
Find out the valid, authenticated, trusted, and qualitative sources of data
Collect, prepare, prioritize based on complexity and impact
View and explore through various visualization market tools
Develop insights through visual analytics and an intuitive approach
Decision on insights by creating models and processes
Let us analyze some use cases for lenders and servicers, both from origination and servicing perspectives:
New product innovation for focus groups: Many loan applications fall out during the loan life cycle either due to borrower attrition or rejection by lenders due to policies and procedures. However, these dropped applications have already taken the time and productivity of the lenders leading to non-value adding expenses. Also, each loan that is not closed leads to revenue loss for a lender.
By leveraging the data processed through past approvals and denials based on segments, lenders can identify new products to make origination more inclusive with options like higher debt-to-income ratio, lower closing fees, lesser interest rates, or low down payments. Recently, a leading US-based bank introduced a new mortgage solution for first-time home buyers, which offers bank-provided down payments and no closing costs. This is targeted at designated markets as well as specific customer segments in regions where some ethnic populations are concentrated. This is a special-purpose credit program which requires no minimum credit score. Such initiatives lead to higher conversions, higher growth, revenue, and profit.
Productivity improvement and personalized journey: Low productivity increases cycle time and cost of origination significantly and at the same time, reduces customer satisfaction indirectly. Recently, the productivity of mortgage loan employees, including sales, fulfilment, and production support functions, has reduced drastically.
Traditionally, lenders create the processes, workflows, and rules over time and continue to use them as standard operating procedures. Through analysis of past data, lenders can identify and eliminate redundant processes, workflows, or rules that cause delays for processors, underwriters, and closers. This data can also be used to personalize processes and procedures based on the borrower profile or persona, which can lead to faster processing and higher conversion. With the help of decision intelligence engines, lenders can review process performances periodically and arrive at optimal, effective, and efficient procedures and policies. Investors and regulators have also started exploring more inclusive credit models in place of decades-old classic credit score models.
Fraud detection: Based on our experience, most loan applications lead to wire and title fraud risk at the time of closing and many of the associated transactions are not registered or valid in title insurer systems at the time of closing. Another key area where data can be used is fraud detection. Various analytical tools available in the market detect anomalous patterns in data and identify signs of potential fraud, which are proactive approaches to preventing fraud. Some vendors provide audit services, while others provide rule engines that can be configured by lenders based on their lending policies. However, most of these tools or services do not provide accurate results due to the lack of comprehensive, high-integrity data. Government-sponsored enterprise investors who buy loans from primary markets have access to collected mortgage data. These investors have provided lenders with fraud tips and identify common red flags and key areas of fraud.
Cost savings and customer retention: Servicers can proactively identify, inform customers, and help them switch between products or apply for home equity loans with themselves, thus retaining customers. Else, customers can be enticed by brokers and other lenders to switch their servicer, leading to attrition resulting in loss of portfolio, revenue, and profit for servicers.