REQUIRED READING: There are many technologies on the market today that purport to be automated underwriting systems. These systems may be very good at what they do – but what they do is not actual underwriting.
Underwriting is an intricate and detailed process. It is about analysis, scrutiny and evaluation – and today's automated underwriting systems are basically technologies that do a lot of surface-level comparing.Â
The mortgage industry seems to have a lot of mixed emotions about whether or not automated underwriting should be used in the industry, and that is perfectly understandable. The way that these systems were previously used – to make underwriting decisions on piles of loans with enough speed to keep up with the volume surges of the early 2000s – left the door wide open to faulty decisions and fraud.
Automated underwriting systems, by nature, are rules-based and will only consider the factors for which they are programmed. During the boom years of the early 2000s, there were quite a few scenarios that were not considered and, therefore, a lot of decisions that were made in error.
Those decisions were not the fault of the automated underwriting system itself. In fact, automated underwriting systems are safe and reliable when they are used correctly. The trouble begins when they are used beyond their intended capabilities.
There is a strategy for making smart underwriting decisions and demonstrating reliability to secondary market investors in today's still-shaky market. That strategy involves using a combination of manual and automated underwriting. Ideally, lenders and investors should rely primarily on manual underwriting for now – and, likely, into the next three years. But as the industry heals itself and investor confidence increases, it should then begin to incorporate more automated underwriting into the mix.
Tomorrow's solutions
However, yesterday's automated underwriting systems – and today's, for that matter – are going to need a significant upgrading once full recovery takes root in the industry and the wider economy.
In order to improve automated underwriting technology, there will need to be a serious shift of attention. The industry will need to create a deeper automated underwriting system that actually engages in intelligent decisioning. By keeping automation in the process, we will be able to more quickly leverage the newest technology advances as they come to the market.
So what could that new generation of automated underwriting systems look like? Let's consider where the current technology needs improvements by considering the actual work that the technology is designed to enhance. The underwriting process involves much more than mere analysis; it also involves intensive scrutiny.
Today's underwriter needs to be part detective, part psychologist and part psychic. This job involves the need to recognize fraud, understand whether the borrower possesses the character to repay the loan and predict economic factors and their impact on the housing market. While we are not close to the point where technologies can wrap all of those functions into one perfect engine, at least we have a worthwhile goal to achieve.
A major problem with automated underwriting systems is that they did not progress at the same speed as the rest of industry's technology. Anyone looking to commit mortgage fraud today can get a huge boost from widely available technology that can easily support fraud, including websites that provide falsified documents and photo-editing software.
Today's automated underwriting systems, however, can only make decisions based on rules. They are not designed to detect fraud. On the most basic level, technology providers need to address this fundamental issue and begin incorporating functions that look for, detect and prevent fraud.
The next generation of automated underwriting systems should also automatically access 4506t-approved tax returns, bank statements, pay stubs and other verifying documentation. These systems should also be able to analyze a document and determine whether it has been altered. This is not as difficult as it seems, and the industry is pretty close to incorporating this into its automated underwriting solutions.
Why is this important? Consider this: When an original document is printed, the font size and thickness is often different than those used in manipulated documents. An automated underwriting system merely needs to incorporate the technologies that detect any change in style, size or format. This is where technology far exceeds the capability of a human being. A computer can detect discrepancies that most underwriters would either miss or take an extensive amount of time to discern.Â
However, most current automated underwriting systems are set up only to give loans a pass/fail ranking. The system of tomorrow needs to be able to grade each loan. There is a big difference between passing with flying colors and barely scraping by – and that big difference can mean hundreds of thousands of dollars in losses or buybacks.
Chetan Patel is executive vice president at ISGN, based in Bensalem, Pa. He can be reached at (800) 462-5545.
