8.4.26 Moving and Shaking; Vesta’s Mike Yu on Mortgage Technology; Weary Agency MBS

Robbie ChrismanWelcome to the Chrisman Commentary, a daily mortgage news podcast. I'm your host, Robbie Christman. Topics on today's episode include moving and shaking around the mortgage industry, how we're seeing inflation concerns weigh on agency mortgage backed securities. and my interview with Vesta's Mike Yu being on the cutting edge of designing mortgage technology. Here, take a listen to a little preview. Robbie ChrismanWhat does a borrower actually want? Maybe a borrower doesn't even know what they want. Or if I'm a borrower, do I really want an underwriting decision in 30 seconds? Do I feel like I got my money's worth if I'm paying thousands of dollars out of pocket for this mortgage to be originated? And I guess tying that to the way you design technology, it has to be at the intersection of where it's used in a practical sense. Thoughts on what borrowers want or like leading them to water? What like what do you think they they want if they they knew what you knew about mortgage technology? Kind of open-ended, but uh it does sit at the crux of a lot of the design of the future of the mortgage industry here. Mike YuTotally. I I think it varies pretty heavily, of course. Purchase versus refi is kind of like the natural, they're two totally different borrowers. Yeah. I really think a purchase borrower, and it's funny because I'm now, you know, at the age where a lot of my friends are buying their first homes. And I think a lot of what I end up hearing is actually they want to feel like they have certainty that this thing's gonna get done on time. They want someone else to project manage it for them, which tends to be, I think, a very hard part of the home buying process. They don't want to be bothered. I actually was just at drinks with a friend a couple of days ago, and she was like, during my mortgage, the most annoying thing was they always wanted to call me on the phone. And I was like, Why can't you use text me? Um, so I I think they want the convenience and they want the certainty, and they want basically to not have to worry about it. They're already worrying about a lot of things when they buy a home and they want to be sure that it's all gonna be fine. I don't think they actually there is always the interesting uh psychology study of like, okay, people are willing to pay more if they feel like the other person did a lot of work, which I think is a true psychological phenomenon, but is actually probably you know, you can solve for that with how you price the loan, for example. Like you're you're giving them this huge capital markets instrument. I'm sure you can find a way to make it feel like they're not actually paying you much for the manufacturing, they're really paying for the capital. And I think people expect to pay for that. On the refi side, of course, I think people actually just care about they care much more about rate and price and they care about it being done quickly. And I think all consumers basically care about not having a headache, right? I think one of the most annoying things for a lot of mortgage borrowers today is they basically have to like project manage their own loan. Like they're kind of not getting the right productive communication from the lender. They're they feel like they're chasing the lender to make sure everything is going to be okay they haven't heard in a few days. And we have tons and tons as an industry of human processes to solve these problems. But I think the certainty is a big thing that people will say in this industry all the time. And I think that's because that that's real. That's like the lived experience of these lenders that talk to borrowers every day. Robbie ChrismanThanks to Figure for sponsoring this week's podcasts. Figure is shaking up the lending world with their five-day HELOC, offering borrower approvals in as little as five minutes and funding in five days. Figure has hundreds of partners in the banking, credit union, home improvement, and of course, IMB space embedding their technology. To learn more, visit figure.com. After a general summer lull, mortgage conference season shifts back into gear with next week's Western Secondary, with plenty of events in September and October as well. So book those rooms. Sheet psychology is one thing, but for lenders, MA psychology is another. Yesterday, an industry source wrote to me saying, quote, I learned this morning that Mason and Mac entered into an agreement to sell to production group to place slash envoy mortgage. They're announcing that to their employees this morning, and they're also in talks with a group to buy their corporate shell and agency approvals. Not sure if their MSRs are going with the production team or the shell buyer. The deal with Place slash Envoy was a private transaction, as in there was no transaction advisor, and they used Wiener Brodsky Kider for legal. Chuck Iverson plans to stay on with Mason Mac to help Mukesh wind down the business, but will have an active role for a year or two under a TSA agreement to support the Mason Mac employees' transfer and integration into the new company. In other news, Better Home and Finance Holding Co., whose stock is down 21% this year, has appointed board member Daniel Lewis as interim CEO effective immediately, replacing founder V. Shal Gark, the company announced on Monday. And shifting gears, U.S. Treasuries and agency mortgage-backed securities opened August on firmer footing as renewed optimism over a potential U.S. Iran peace agreement sparked a rally in longer dated bonds after last week's sharp sell-off. Oil prices declined after President Trump announced new U.S. Iran talks aimed at reopening the Strait of Hormuz, raising hopes for easing geopolitical tensions despite conflicting statements from Iranian officials. With the 10-year and 30-year treasury yields near multi-year highs, the bond market is increasingly testing the notion that higher rates may need to persist until inflation is convincingly contained, rather than relying on the Federal Reserve to suppress borrowing costs through policy intervention. Following last week's FOMC meeting, expectations for additional rate hikes have been modestly paired as the Fed maintained a data dependent stance rather than signaling a broader policy shift, leaving September's decision largely contingent on upcoming inflation data, despite Chair Warsh's more limited use of forward guidance. U.S. manufacturing momentum gained in July, with the ISM manufacturing index rising to its strongest reading since May of 2022, well above the expansion threshold, as robust demand, stronger production, and renewed hiring reflected growing confidence that tariff uncertainty and war-related supply disruptions are easing. The final S&P Global manufacturing PMI held steady, a sign of continued growth in factory activity. In contrast, June construction spending unexpectedly slipped to 0.1%, as weakness in residential investment more than offset gains elsewhere. Attention was also on Treasury's quarterly refunding announcement, and it expects to borrow $739 billion in the third quarter, $68 billion more than it projected in May. His theory is that mortgage operations run on outdated technology, and the loan origination systems most lenders use are so old that only one person can use them at a time while everyone else waits. Some of the country's largest lenders use Vesta, and those lenders use them to automate almost everything from application through funding, lowering operating costs, cutting processing times, and turning fixed headcount into elastic capacity that flexes with demand. Robbie ChrismanAs it pertains to the qualitative factors surrounding technology and people's interaction with them in the mortgage industry. And I wrote you a note saying it seems like the pendulum is swinging back toward the competitive advantage being not in streamlining loan processing, but in engaging buyers earlier through various tools or relationships. And maybe that's that's technology can be used for that, sure, in identifying opportunities. But the philosophical discussion that I want to have is kind of this this pendulum swing against hey, we're just maximizing technology, maximizing, we're trying to figure out kind of the human side of the equation, where to actually build relationships, where to build trust. Your thoughts on before I go super draconian and say you know we're rehuman humanizing things, just thought thoughts on kind of where we are with technology moving from streamlining loan processing into working on the relationship aspect. Mike YuYeah, I think there are a few reasons people are thinking about this. Honestly, one is I do think that there is a little bit of this, to your point, the pendulum that swings back and forth, which is whether it's rate cycles or whether it's like, you know, lenders start some projects and they're harder or they see someone else having some success. I think the reality is people often are trying to figure out like what is the thing that only a few other people are doing that I can be an early adopter to that I can use to like kind of get an advantage. And so I think you have this very natural pendulum swing because as soon as everyone is doing it, you almost like don't want to do it anymore because all your competitors are doing it. And so you've got to go and figure out the next thing. And I think there is like a cohort of lenders who really do that, who set a lot of the dialogue and the conversation in the industry about what the competitive advantage is and where we're using tech. And so I think that, by the way, and you know, very loud voices on social media, for example, is a big reason that that has become a lot of the dialogue. So we'll we'll we'll put that to the side for a moment. That pencil's always going to swing. I think the other thing that has happened recently is people have really started to internalize how much AI can do. And I think it was like really easy for people to grok, like, oh yeah, AI can read documents, because we've been telling that story in this industry for like, you know, decades, actually. It's like the only thing that people actually used AI for before large language models mostly was starting to read documents. And so when this AI wave comes and hits and everyone's talking about NVIDIA and OpenAI and Anthropic and AI, people are like, well, where am I going to apply this in my business? The first obvious place with almost no understanding of the technology, you would say, Well, I'm going to go put it in the back office in doc processing. From a regulatory perspective, it's safer. It's kind of like an obvious use case, et cetera. And what I've found, for example, when I talk to lenders, is people like when they established that was the thing to do, they really underestimated the capabilities of the technology. And like, even today, we go and implement Vesta and we're like, hey, like, you know, the AI can do all these things. And like, okay, but I have a human do this. Can the AI do that? And the answer is almost always yes. And I think in the last three to six months, we have started to see a lot more actually like really impressive voice demos that have started to make their way around the industry. And so people are starting to actually understand that, like, hey, this AI technology is not just something that can do this back office job. And of course, there's more licensing, more regulatory complexity, et cetera. But the AI also is good enough, good enough technology-wise, that you can have a very realistic sending voice agent. You can have a very realistic teaming conversation. And the technology has improved. I would say more than that, lender's understanding of the technology has improved. And so now it's like everything's back on the table. You know, like this is probably something that a year ago, if you had asked, you know, a bunch of mortgage lenders who would not spend a lot of time with the technology, it doesn't seem reasonable that like an AI voice could do that. There'd be like, no way. Like that's ridiculous. That must be a human thing. And now I think the number of lenders that I've seen who are just like, Did you hear this voice demo? It's like crazy. That's really significant. So I think the technology really being able to do a lot of that stuff has also steered some of the conversation that way. Robbie ChrismanHow much do you actually follow the conversation? Because I've noticed that a lot of the chatter out there starts to sound the same the more you listen to it. And maybe that's because people are using AI, or maybe that's because people that think they're cutting edge are not necessarily cutting edge. Like, where are you pushing the envelope forward when it comes to listening to others versus thinking for yourself? Mike YuYeah, I think it's really important actually to think for yourself. And then to the extent that you are listening to others, it's got to be like a real deep conversation with like people that you trust one-to-one. There's a lot of narrative going on in the industry. Everyone wants to go on their earnings call and tell a story about how much AI they're using because that's what the stocks care about, or that's what you know, Wall Street cares about. And so I think it's actually really important to be careful not to get that into all the things that people are talking about on social media. I think there is a lot of people that just say, like, hey, we adopted this tool, we adopted that tool. And then when you actually go dig deeper and you, you know, go under the covers and you peel back the details, you're like, all right, well, they're, you know, they haven't actually let it off the autopilot leash yet. Like you still have a lot of human in the loop, or they're only doing like a very limited set of you know use cases, or they're only doing HELOCs with it, or something like that. And so I think that it is people have this natural incentive in the dialogue to kind of overstate the things that they're doing. There's lots of reasons for it, you know, from a recruiting perspective and a talent brand perspective for these mortgage lenders. Like that's a big thing that people care about. From a I mentioned the stock price thing and how analysts think about it. There's a lot of things that you don't have to get into the details of, and you kind of just say it at a high level in the way that is kind of like most flashy for you. And I think there's a lot of that. And what does that kind of mean? That means that when people are saying, like, hey, I now have AI voice agents that nurture the barber relationship the entire time, they're probably in the very early stages of trying that kind of stuff and they're kind of announcing it like it's there. And so, what does that mean? That means like probably, by the way, like I tend to be a technology optimist. Like, I think all these things are gonna work, but like they don't know for sure that it works yet. If you kind of pivot your strategy to follow them, they might be talking about something else next week that is like the new pilot that they're doing because they talk about these things at the beginning and not the end. And so I think it's really important to be careful and have like your own point of view on how the technology and the future of the industry are gonna evolve together and pursue that. Of course, you want to be collaborating with others. Of course, you want to understand what other people are doing. But I just think you've you've got to get that in the one-to-one, you know, backroom conversations, not in what someone says on a conference stage or not in what someone says on social media. It's just like it's almost never the nuanced deep picture that you need to make a good decision. Robbie ChrismanAnd maybe I put the cart before the horse. I started this conversation by talking about moving from streamlining loan processing into engaging buyers earlier. And I want your opinion. At some point, we will reach this. Okay, you have an underwriting decision in two minutes or in 30 seconds or whatever. You have a valuation of the property that's done in a minute or two and it's accepted by whatever investor you're ultimately selling it to. How far are we from these things all becoming more and more ubiquitous within the mortgage industry? And I guess maybe more important than that is where does the industry go once all the challenges of streamlining loan processing have been solved? Mike YuYeah, I think that we still are pretty far from it being ubiquitous. And we definitely, for example, are on the journey with some of our customers of saying, hey, AI agents can actually do most of the back office workflow and tasks. You can really compress that time to get to a you know fully underwritten loan. There's still lots of moving pieces, but I would say the hardest thing is just diffusion of this technology is much slower than anyone expects. Almost, almost always. If you just look at AI spend in general, like zoom out from the mortgage industry for a second, all these companies, these enterprises have spent billions of dollars on tokens with anthropic, with open AI, with cursor, with cognition, with Zero, with Dekagon, with like all of the first generation of AI app companies. And the number one question, you know, you go on CNBC and they talk about it. It's like, they're like, where is the ROI? What am I getting for all this spend? Am I actually seeing more efficiency? All the tech podcasts now are like, well, AI was supposed to take a bunch of jobs, and we haven't seen, you know, like it take a bunch of jobs and all of these kinds of things. And I think the answer in many ways is like, you can start using the new technology. But as we've always said in this industry, I think, or at least I've been hearing for the decade I've been in it, have you re-architected your process? Have you like really changed what you expect your people to do and all of these other things that have to happen? And so we've got a bunch of AI agents deployed where honestly, the people still like look at everything that the agent does. Like they read every single agent case and figure out like, well, did this AI agent do what I would have done? It's really hard to say we have like fully streamlined the loan manufacturing process if you still have people, you know, QCing basically every single thing that the technology does. And that is has been our pattern for many years in this industry. I do think that we will get away from that over time. And we are starting to see some early customers get away from that. But then you've got customers that haven't even started that journey. And that's a 12, 18, 24 month journey from starting to roll out those AI tools and that technology across your workflow to actually saying, hey, I have people out of the loop on a majority of my loans. And these things really are delivering that customer experience and that cost efficiency that I wanted. And so I think there's a long way to run on that stuff. I think after we do get there, which, you know, for that to like diffuse throughout the whole industry, you're probably still looking at three to five years. And lots of people then, by the way, will be like, oh yeah, three to five years, that means I'm not in a rush. And the problem is it's three to five years, but it's also gonna take you a long time to do it. So if you don't start soon, when you realize two years from now that everyone else is one year away and you're still three years away, there might be some problems. Uh, but once everyone has kind of like worked their way through that process, which I do think is gonna happen, I think realizing the ROI is gonna show up. I think that we're gonna have to be really thoughtful about how we manage and communicate that stuff and what people do with you know the new time that's freed up and how you repurpose people into talking to customers and customer service and sales and things like that. But I think that that's a journey that we'll go on. Once you get there, I think then you're kind of at the fundamentals question is like, all right, so as a mortgage lender, what are you competing on? And today there is some opportunity to compete on manufacturing because it's kind of like this commodity-ish product. And the commodity products, it's like pretty common that I would say to compete on cost of manufacturing. And so today lots of people think there's like an opportunity to compete on cost of manufacturing. Once everyone's cost of manufacturing is, you know, basically just the cost of technology, then the questions will end up being, okay, how else do you compete in a commodity market? It really tends to be things like brand, it tends to be things like having the customer relationship, having the right partnerships where you can get a bunch of consumers that are already in some other ecosystem that come into your ecosystem. Like meeting the customer where they're at is always gonna be a big part of this industry. And I think that's gonna be a huge opportunity to differentiate. Customer experience, I think is one that people have talked about a long time that I personally am a little bit less bullish on just because I'm like the experience is too infrequent, I think, for it to like really make that big a difference in someone's buying decision. But I could be wrong. And then I think there will be, of course, always the opportunity in this industry because of the financial product shape to compete on cost of capital, which today I think the banks would say their cost of capital is not very competitive given all the regulation. How that evolves over the next decade is anyone's guess. And then, you know, lots of other players have lots of various other sources of costs of capital that they're starting to think about, um, working with hedge funds, insurance funds, et cetera. Figure, of course, doing a bunch of interesting stuff. Spring EQ just sold a Cerberus. Maybe that was like a couple of years ago now, but big, big kind of fund with a very different cost of capital profile. So I think there'll be like the front-end consumer experience and the cost of capital side and product innovation, like financial product innovation where you can compete. But this thing in the middle that they're all competing on of uh cost of manufacturer makes sense until to your point it kind of gets to the floor. Robbie ChrismanYou brought up a very operative word there, and that is thoughtful. And it can it can be applied to a variety of sub-conversations within the the overall one we're having here. But but when it comes to AI, blindly throwing AI at everything, blindly trying to maximize the amount of AI usage in your life so that you're ahead of the learning curve or you're you're an expert in it. What is the thoughtful way to go about trying to bring it in to your workflows, your operations, your development of being thoughtful in development of it? Thoughts on doing it versus just throwing trying to throw AI at everything or have AI as a catch-all, or I think there's people that are that are overutilizing it thinking, well, I'm an early adopter, so I'll shake out a head regardless. Mike YuTotally. Yeah, I think as with all technology, thoughtful deployment of the technology is really about what problem are you solving. And by the way, I will say I think that there are plenty of times I have told people they just need to figure out how to use AI in one or two things because the problem that they're solving actually is that they don't understand what the technology can do. And there's like an education problem to solve. But once you've solved that problem, which generally is like, you know, only the problem for the first like one or two use cases for the first one or two months, uh, once you've solved that problem, you've got to be really specific about what business problem you're trying to solve. Are you trying to solve the problem of like, you know, hey, I find that I'm asking, I'm overconditioning my loans. Like I ask borrowers for an average 20% more documentation than I actually need to. You probably want to deploy a very different AI solution to that problem than you know, the AI problem actually being like, I have quality issues on the back end because I'm missing, you know, all sorts of bank statement fraud, that I'm very unfortunate that there's some ring of borrowers that has figured out how to like do really good bank statement fraud. And you would obviously deploy a very different AI solution to catch that. It really is about understanding what problem you're solving. And then you scaffold the whole program around that. It's like, all right, so if I solve this problem, what metrics in my business would move? And if I solve this problem, like which of my stakeholders are going to be happier and like how will they know that we were successful here? And then you say, all right, I've deployed it, and you go back and you measure yourself against the problem you're trying to solve. You're not measuring these like proxy things like how many tokens am I burning, or you know, like how how many loans are using my new solution. Those are like things that lots of people deploying AI today, I think across the world, are trying to measure, and it's not the right thing. It really is like you've got to be anchored in the problem you're trying to solve. And you measure the solution success by whether you solve the problem you originally set out to solve, not by like whether people use the solution. Robbie ChrismanTon of great insights. You know, I always love talking to you. I waited too long between episodes to have you back on, but absolute pleasure, and uh, I'll see you out there on the road. Yeah, thanks for having me, Robbie. Ciao soon. Robbie ChrismanHigher long-term yields, persistent inflation concerns, and ongoing geopolitical uncertainty weighed heavily on agency mortgage-backed securities in July, producing the sector's weakest monthly performance since late 2024 as rising duration risk, wider volatility, and softer investor demand pressured returns. Although mortgage-backed securities remain marginally positive for the year, investors continue to favor shorter duration, higher coupon securities and defensive positioning. Seasonal headwinds, elevated inflation risks, and uncertainty surrounding the Middle East conflict expected to keep the sector under pressure in the near term. Markets are now focused on whether upcoming inflation reports and payrolls data validate expectations for a September rate hike, with investors increasingly viewing June's softer inflation rating as a temporary reprieve rather than evidence that inflation has been durably brought under control. Today's economic calendar includes the June trade balance, Red Book same store sales, durable goods, orders, and factory orders for June, Jolts job openings, and the Atlanta Fed Q3 GDP estimate. And the 10-year yielding 4.67 after closing yesterday at 4.69%. Let's wrap up with a joke and some housekeeping. I told my financial advisor I wanted a portfolio that could withstand anything. He said, Great, we'll invest entirely in the market's ability to surprise you. Or how about this one? Why did the mortgage banker bring a ladder to work? Because everyone kept talking about a higher rate environment. The yellow curve walked into a bar. The bartender said, Why the long face? Curve replied, I'm just steepening. Thanks to Figure for sponsoring this week's podcast. Figure is shaking up the lending world with their five-day HELOC, offering borrower approvals in as little as five minutes and funding in five days. Figure has hundreds of partners in the banking, credit union, home improvement, and of course IMB space embedding their technology. To learn more, visit figure.com.