This piece originally appeared in the November 2023 edition of MortgagePoint magazine, online now [1].
“In God we trust; all others must bring data.” This quote, widely attributed to statistician and business management theorist W. Edwards Deming, serves as a stark reminder of the importance of using data to improve manufacturing quality—which is what Deming became known for in the automobile industry.
Of course, quality is as critical to the mortgage industry as it is to car manufacturers. And yet, efforts to ensure high-quality data are often at odds with another industry need—speed, whether that applies to speed in mortgage production, transferring MSRs, or countless other business needs.
For this reason, lenders and servicers alike have gravitated toward data extraction technologies in recent years to streamline workflows and reduce costs. But unfortunately, not all solutions deliver the same results.
Accurate data is essential for the smooth flow of operations, meeting regulatory requirements, and profitable trading in the secondary market. Yet methods of checking loan data for completeness and accuracy have traditionally been time-consuming and error-prone tasks that not only drive up costs but also place organizations at risk of repurchases and noncompliance with the GSEs, investors, and state and federal agencies.
For lenders, ensuring data accuracy is essential to streamlining workflows, compliance testing and audits, and producing high-quality loans that can be confidently sold on the secondary market. The ability to extract accurate data promptly also empowers lenders to make informed decisions, improve underwriting processes, and reduce the risk of costly loan file errors and defects.
Such defects are growing more costly as well, particularly in today’s purchase-oriented market. Earlier this year, Fannie Mae and Freddie Mac introduced new repurchase strategies to deal with the increase in defects for loans originated in 2020 and 2021, when lenders were dealing with capacity issues. A recent analysis by Freddie Mac found that purchase loans had a 36% higher rate of defects compared to mortgage refinances.
Similarly, servicers rely on accurate data to streamline the loan onboarding process, enhance servicing efficiency, and automate critical servicing functions, such as payment processing and escrow management. Ideally, these efforts result in improved borrower experiences and increased operational efficiency, too.
For these reasons, both lenders and servicers have increasingly turned to third-party data extraction technologies.
However, the bulk of these offerings not only fail to improve loan data quality but often add extra time and cost to the equation.
The root of the problem is most of these providers rely on legacy technology and optical character recognition (OCR) tools that have not evolved over the years to keep up with the increasing diversity of loan document types or new data extraction methods. This leaves organizations with the burden of spending additional time manually reviewing and fixing errors by hand.
Of course, third-party loan QC providers have emerged as an option for ensuring higher quality loan data. However, there is a trade-off in terms of turnaround time. In fact, one of the chief complaints often heard in the business is that lenders and servicers can rely on their third-party providers to get perfected data back, but they must wait a day or longer to get it—which often results in bottlenecks in mortgage workflows and hampers timely decision-making.
Thankfully, lenders and servicers can achieve the best of both worlds—fast and accurate data extraction—by harnessing the power of AI.
By leveraging sophisticated algorithms and machine learning capabilities, AI-powered document automation technologies can accurately extract and analyze data from a wide range of mortgage documents, including pay stubs, tax forms, and bank statements.
This automated process eliminates the manual effort involved in data verification, reduces errors, and ensures a higher level of accuracy. Ultimately, it enables lenders and servicers to generate purified data to fuel their downstream automation initiatives, resulting in cost savings, improved customer service, and significant SLA improvements.
With AI-based document automation and extraction tools, lenders and servicers improve the quality of their data without sacrificing the speed they need to impress their customers and business partners.
The new enhancements include enriched audit trails for extracted data, providing users with more detail about the source of each data field, which can support improved levels of automation in their workflows.
Recent advancements in AI have ushered in a new era of data quality improvements without compromising speed. In a challenging market like today’s, when lenders and servicers need all the business they can get—yet also need to stay compliant and keep costs down—the time to embrace AI is now.