Do you own your tech, or does your tech own you?
I remember back in 1975 when fax machines started being used. You would wait patiently for several minutes for the machine to give you a piece of paper that you couldn’t even read. And yet, we were impressed. As long as I can remember, tech played some part in mortgage lending. Before computers, we used typewriters with a screen. I know I’m dating myself, but in order to see how far technology has come, we have to look at where it began. Take, for example, this Nov. 1984 piece in The New York Times:
Electronic typewriters come equipped with circuit boards, which contain a host of semiconductor chips, eliminating much of the machinery that drives manual and electric typewriters and providing advanced features such as memory capacity. Because electronic devices have fewer moving parts, they are generally considered more reliable than their electric counterparts.
We have seen a significant change in our technology, the amount of it, the ways in which we use it and what it can be used for. It is transforming the way we live, interact and even do business. But it’s not the technology development that moves gradually, it’s the adoption. How well are you using your tech, and are you using it in the most efficient manner possible?
Where there is a potential for change, there is resistance. People are really resistant to change of any kind at any point in their life – that is just a human characteristic. And it still happens in our industry today through the unwillingness to adopt new technology. It is hard to get people to use a new system, even if it is better. You have to take time and effort to convince people that they’re going to personally gain from it, you have to make sure they’ve trained adequately and then you have to tell them to do it.
Lagging technology is not a technology issue – it’s a people issue. There’s tremendous technology out there. In fact, I’ve invested in great technology many times and couldn’t get it used. People like to have control and ownership, so they’re reluctant to use what others suggest, or let go of what they use.
And at times, even the fi nance industry’s regulators stand in the way of change as they fear the consequences consumers may face as fintech’s usage increases in the mortgage lending process. But that is all beginning to change.
Back in 2018, the Office of the Comptroller of the Currency and the Federal Deposit Insurance Corp. announced they were exploring granting federal bank-like licenses to tech-driven fi rms that off er financial services. This was part of a broader push by the federal government to boost small businesses.
Oftentimes, lenders are willing to invest in technology, but don’t want to switch to newer technology in order to keep up with the “latest and greatest” because they have already invested significantly in their present technology. Switching would also require them to retrain. The digital mortgage is evolving, but do I think I’m going to see a widespread fully digital mortgage before the end of my career? No, I don’t.
Digital mortgages are still discussed and approached, but we continue to have problems getting online notarization. We still have a long way to go before we see uniformity in the industry due to regulatory concerns and the vast amount of technologies being used.
But even amid the struggle to get technology integrated into the mortgage process, there is still hope, and increasing potential.
A recent survey from Fannie Mae, which surveyed 184 lending fi rms on their interest in AI and machine learning, showed about 66% are familiar with AI, but only 27% are using it in their business now. And of those, only half are using AI on a consumer-facing front.
Other lenders, however, are making plans to begin using it. Looking ahead two years, 58% of lenders expect to be using AI and machine learning in their mortgage process. Another 22% predict they’ll be investigating AI and 19% are on a wait-and-see plan.
Only 2% of lenders stated they have no intention of using AI in their mortgage business.
The greatest opportunity for technology and AI use in the mortgage industry is data transfer. Right now, lenders have a heavy data-entry process. At any point in this process, they could make keying errors or insert human errors that could cause problems later in the origination process. Using AI not only ensures the accuracy of the information, it cuts down on the mundane data-entry work and shortens the time to close.
As it turns out, humans aren’t that great at data entry. A study from the National Center for Biotechnology Information evaluated more than 20,000 individual pieces of data to examine the number of errors. The study showed that out of every 10,000 entries, there were 650 errors – a 6.5% error rate.
How much money are lenders losing due to that? Experts use the 1-10-100 Rule to determine what businesses are losing. If a typo is made, it could cost the lender $1 at first. But if it is not corrected, the second time it would cost them $10, and if not corrected the third time, $100.
For mortgage lenders, however, the stakes could be even higher. Data points such as Social Security numbers and dates of birth are necessary to document identity verification to comply with the Bank Secrecy Act. And data entry errors could also lead to mistakes in loan amounts. A $10,000 loan, for example, has different implications in the compliance world for reporting, documentation and pricing than a $100,000 loan. Even if the loan is funded correctly, a single zero incorrectly entered in a bank’s loan management system can lead to costly oversights.
It’s somewhat surprising to some people, but our sufficiency levels and the time it takes to close are worse now than they were before the housing crash in 2007, with days to close running at 40 to 45 days and the fallout rate is running anywhere between 25% to 35%.
The latest Ellie Mae insights report showed that the time to close all loans hovers near 45 days, in the upper 40s for purchase originations and the lower 40s for refinances. This is ridiculous for a time when, with the right technology, loans could close in 10 to 25 days, or less.
The key is focusing on AI technology where it can be the most beneficial. AI is good for refi nance transactions where you have sufficient data to issue an automated report, and that can account for over 20% of refi originations.
Also, if consumers are educated and informed throughout the origination process, the closing will be a breeze; but if they aren’t given sufficient information throughout the process, they will want an explanation of every document they sign, creating long, cumbersome closings.
AI can help lenders by pushing out notifications that give buyers and sellers automatic updates, letting them know exactly what is going on each step of the way. This allows consumers to stay in the know, and ask their questions along the way.
NOT JUST ANOTHER BUZZWORD
At the risk of alienating many technology experts, I don’t see blockchain as having any application in our industry. One reason for its low acceptance rate is that it is more susceptible to fraud than any other system out there. I think the future of the digital mortgage cycle will be more about the collaboration between the industry’s participants.
Blockchain, AI and machine learning – we have heard a number of various buzzwords come and go through the mortgage industry, each time wondering if this will be the next best thing. Artificial intelligence seems to be here to stay, but it has a very specific usage.
AI is useful for specific markets and products. It is not for everything. And, because of the different regulatory environments, it’s rare that anything developed for the housing industry can be used nationally. For example, in some states, you have to use an attorney just to close a transaction. Moreover, real estate data access varies by market, with some having automated title plants and some depending on courthouse searches.
The bottom line is, there is no technology that’s going to come along that will change the regulatory environment, and so we must continue to navigate around that as best as we can. Because of this, the housing industry can’t take a “one-solution-fits-all” approach. Each time we begin to adopt new technology, we have to determine if it works best in the space, and where it can best be used.
Artificial intelligence seems to be here to stay, but it has a very specific usage.
AI can’t completely take over the mortgage process, but it can help us shorten the closing process; we have to focus on the areas where AI can create the most of these efficiencies. Attempts to eliminate re-keying data are already underway and will be a game-changer for mortgage lending.
But right now, it doesn’t seem as though the technology and acceptance are there to do it. Right now, optical character recognition has improved significantly. In fact, accuracy is almost 100% (much better than the previously discussed human accuracy rates.) But there is one keyword there – almost.
Until we can trust a complete, 100% accuracy rate, how will lenders trust handing over their livelihood to the hands of a machine, without human eyes ever seeing it? It is a big step, and while it can happen, especially through the new strides we are seeing in AI and machine learning, it is one I’m not sure we’re quite ready for yet.
LET’S WORK TOGETHER
Technology today is driving housing industry participants to work together and collaborate on a number of different fronts. Take iBuyers, for example, as many companies broaden their spectrum to include lending and real estate services. This kind of collaboration could only be brought on by technology and will continue to increase.
As it stands, there is a tremendous amount of regulatory oversight over housing that has caused the industry to evolve in the different verticals we see today, such as real estate agents, lenders, title companies, settlement agents, appraisers, inspectors and many more. And each of these verticals has worked to evolve their technology separately according to each one’s regulatory and practical needs.
But this means many different industry participants are often working to gather the exact same information multiple times. That presents one of the greatest opportunities for AI, which still hasn’t been totally realized – effective data transfer.
How do we, as separate industries, work to eliminate re-keying data, prolonging the process and risking more human errors by working together? That will be the next problem the housing industry will look to solve, and it will be able to solve it through more integrated processes fueled by AI, machine learning and other tools that continue to evolve as we speak.
The role AI will play in the digital mortgage process is still being determined. While it will fix many problems, it cannot be applied to every part of the process. Lenders will increasingly trust AI as its accuracy improves and technology continues to take over every part of the digital mortgage process. But lenders will soon learn that the investment is worth it as they allow AI to do the more mundane tasks and move their employees to parts of the mortgage origination process where their talents can be better utilized.
“THE TRUE ADVANTAGE OF USING AI IS TO REDUCE COSTS FOR LENDERS, DECREASE TIME TO CLOSE AND MAKE THE ORIGINATION PROCESS FASTER AND EASIER TO UNDERSTAND FOR EVERYONE.”
People love to talk about a mortgage that can be originated without a single human hand touching it. To be sure, today’s technology, as it’s being developed, puts us at a point where we might be reaching close to that, but we’ll never completely get there. The true advantage of using AI is to reduce costs for lenders, decrease time to close and make the origination process faster and easier to understand for everyone. From here, we just have to focus on learning when and where to use it.
About the Author
Patrick Stone serves as founder and executive chairman of Williston Financial Group. Previously, Stone served as Vice-Chairman of Metrocities Mortgage, a 2005 top twenty mortgage lender, and as chairman of The Stone Group. Stone also serves on the boards of Fidelity National Financial, First American Corp., FNIS, MicroGeneral, SKLD, World Minerals, Green Street Advisors, DigitalMap, Homegain, RedVision, and Wystein Capital.
Reprinted with permission from the author‹ Back to Blog