HOUSINGNEWS REPORT
HOUSING PRECOGS: BIG DATA PREDICTIONS BEYOND HEURISTIC HUNCHES
case with mortgages. Nobody has a photo of their mortgage. A VA loan is a VA loan. Lenders can originate more of them tomorrow. What counts most for borrowers is low cost and convenience, the ease of application. Increasingly, what counts most for lenders are digital platforms, the use of artificial intelligence, and an ability to overcome clutter, to be the first and only lender with whom borrowers interact. Lower costs for borrowers Borrowers want both convenience and lower rates. The coming competition will not be for the fastest loan application but the one which produces the lowest cost, the app that allows borrowers to pick from competing mortgage offers. “For a typical $250,000 loan,” says Freddie Mac, “the average expected savings from only one additional quote is $1,435.” It adds that “80 percent of borrowers who obtain one additional rate offer will save between $966 and $2,086. The average expected benefit increases to $2,914 if the borrower receives five rate quotes. Eighty percent of borrowers who obtain five offers will save between $2,089 and $3,904.” You can see the disruption opportunity. Well-funded fintech firms will spend $1 million a week promoting platforms where borrowers save and lenders compete. Lower costs for lenders Given origination costs of better than $8,000 per loan lenders have every
reason to use AI not only to generate more business but also to reduce production expenses. A major target will no doubt be personnel expenses, which according to the MBA averaged $5,346 per loan in 2017. Speech recognition software will play an important part in the cost-cutting process. Once clunky and mechanical, such systems have improved greatly in recent years. In India it’s estimated that the human-like Google Assistant program has received 450,000 marriage proposals. For loan officers virtual speech programs and other automated services represent both promise and peril. The promise is that a single loan officer can be more productive because the application process is increasingly automated. The peril is that there are only so many loan
applications to be written. If some loan officers — or fintech systems — are more productive it means large numbers of the nation’s 306,000 loan officers are not. Instead, they now represent an expense of almost $19.5 billion a year (306,000 loan officers x an average salary of $63,650 according to the Bureau of Labor Statistics). The idea that there will be so many loan officers in five or 10 years is improbable. Ditto for large numbers of mortgage underwriters. “I don’t think targeting marketing using predictive AI will lower a marketer’s expense,” said Brad McDaniel with Likely.ai. “It will rather have a huge impact on their return on investment. Think about the difference between bombing that occurred in World War II and that happens today. Do you think today
7
MAY 2018 | ATTOM DATA SOLUTIONS
Made with FlippingBook Online newsletter