How AI Could Simplify the Mortgage Market

Morgan Stanley's Jeff Adelson and Jay Bacow discuss how AI agents on smartphones could transform the US mortgage market by making borrowers far more responsive to rate moves. The main implication is faster prepayments, which would make MBS more negatively convex, shorten durations, and widen spreads. They also see upside for low-strike receiver swaptions as investors buy duration, while the impact on mortgage lenders is mixed, and broader housing/consumer effects may take years to materialize. - AI agents could autonomously compare lenders, reduce paperwork, and prompt refinancing when economics work, raising historically low refinancing responsiveness. - Faster refinancing would increase MBS prepayments, make them more negatively convex, shorten durations, and widen spreads by about 10bp in the base case. - A 100bp rate rally could push refi volumes 40% above current expectations, amplifying MBS headwinds. - Scaled mortgage lenders would see higher origination volumes and operating leverage, but competition could compress gain-on-sale margins. - Duration shortening will create demand for low-strike receiver swaptions as investors seek to extend duration, particularly in a rate rally. - Over 3–5 years, AI could expand home ownership, home sales, and use of second liens, HELOCs, and cash-out refis. - Key monitoring signs include easier mortgage-application interfaces, faster closings, and a step-up in refi volumes for a given rate decline.