All right. We'll get started. I'm very pleased to have PennyMac Financial Services join me on stage. With me today, I have David Spector, Chairman and Chief Executive Officer. Welcome, David.
Thank you, Terry. Great to be here, and thank you all for joining us.
All right. We'll get right into it. You guys filed a quarter-to-date update yesterday. Maybe just give us some color around that. What are you seeing around third quarter trends?
Yeah. Clearly, as we sit here today, we're seeing rates at a much higher level than we saw at the beginning of the quarter. Mortgage rates are up about 50 basis points, and so like you would expect in a rising rate environment, we're seeing production taper off as a result of that. That's leading to some, obviously, reductions in our forecast for originations in all three channels. In our consumer direct channel, we're seeing margins hold in, which is good news. Obviously, with the increase in rates, we'll see a higher percentage of closed-end seconds getting originated. So the weighted average margin at a consumer could go up in the second quarter, but with originations down, the overall revenues will be down.
Similarly, in our broker direct channel, we are seeing a little bit of margin pressure there with production going down, and we will be talking a little bit more about broker in a few minutes. In the correspondent channel, we are seeing, again, production down a bit there. Margins are holding in really nicely there. There is good news there. On the servicing side, look, we continue to see a great story out of servicing. Our servicing portfolio continues to run at very low delinquency levels. It is a portfolio that has close to $530 million of revenue or servicing fees every quarter, and we are seeing good results there. On the corporate side, we had some headwinds in Q2 from credits from AWS and some revaluation of incentive compensation. We should run flat to a year ago.
It is a story that, given the increase in rates, we are again going to be at a, what I call a post-COVID trough level in operating ROE. Again, it will be in that single mid to high single digits, probably mid in Q3.
Got it. That is helpful color. Forgot to mention, we are going to be mostly discussing PFSI on this chat, and then we will address PMT a little bit at the end.
Right.
Again, helpful color. Lots to unpack there. Maybe just taking a step back, the macro has been tough this year. Mortgage activity has remained pressured for most of this year. As you look across the balance of 2026 and 2027, what are your expectations for industry volumes, and what do you see as the biggest drivers of upside or downside from here?
Yeah. As I mentioned earlier, with the higher rates, I think volumes are going to come in on the lower end of forecasts. From an internal point of view, I think there's tremendous opportunity within the company. We're seeing a lot of good results coming out of our consumer direct channel on the deployment of our new technology. More exciting about that is the fact that we're really beginning to see the benefit of having legacy-free technology and being able to build AI agents on top of it. On the mortgage fulfillment side, which is our processing group, we're seeing processing costs down approximately 50% on a marginal loan basis. On the loan origination side, we're seeing loan officer efficiencies up about 20%, and that's only going to grow.
Today, as we sit here at the higher rates, you have the benefit on what you're originating. In many ways, what's more exciting to me is the fact that we've been carrying a lot of excess capacity over the last few years to be able to really seize on an opportunity of lower rates. What we're seeing with the technology is the need to maintain the levels of excess capacity that we've been holding have been greatly diminished, if not eliminated. That's what led to our reduction of production expenses of $60 million that took place in July. We really reduced the capacity, and that should reduce the volatility of the earnings in our consumer direct channel.
In our broker direct channel, we're doing the work rather quickly to use the technology that we deployed in our consumer direct channel to deploy it in our broker direct channel. Now, there's a little bit more work on top of that that we need to do in the fact that when you run a broker direct business, there's a broker portal that has to go along with that that's being built. That's something that I think is going to be a real exciting opportunity for our brokers. The brokers that we deal with typically deal with the number one in the market or the number two in the market. I think as we continue to be competitive with price, while being able to offer a better product, that's going to lead to share growth within our broker direct channel.
In our correspondent channel, we've been the leading correspondent aggregator for many years. That lead has shrunk a bit, and we're going to get into the reasons why momentarily. But we're going to continue to be the leading correspondent aggregator. There, it's just a matter of when you're in the correspondent business, you're in it by servicing to really support the flywheel that we've created. As we see opportunities to participate in that market, and we like the servicing values that we're seeing in that market and how those are being constructed, we're going to continue to see growth on the correspondent side. On the servicing side, it's with great pride I say there's not a lower cost servicer than us. I think that's with our servicing portfolio, it's $700 billion, and the cost to service that we have for that.
With the technology that we have in place and the work we're doing with our technology called Plaisse, we're going to continue to drive down the cost to service. That's something that I'm really excited about. With the Senstar acquisition, look, we're going to add scale to our portfolio, and that's only going to help us drive down costs further, while equally as important, grow the capital-light part of our business model. That's something we're very focused on as we look out into the future. Finally, across what I call our shared services groups, technology, finance, and capital markets, the benefits we're getting from AI and AI tools like Claude and Cursor and others are meaningful.
The technology investments that we're making within the company, which I expect to crest in the third quarter, are really going to start coming down as we finish a lot of the work that we undertook at the beginning of the year. For all things considered, if you think about the beginning of the year, we walked in expecting three interest rate cuts. As we sit here today, we're expecting three interest rate increases. By and large, I'm really happy and excited about what I'm seeing in the organization.
Got it. Super helpful. I want to talk about returns. Last quarter, you revised your operating ROE target to the mid-teens by year-end 2027. Can you maybe just walk us through the glide path from here and the major building blocks required to achieve that outcome? Maybe just more specifically, how should investors think about the relative contributions from technology-driven cost saves, just overall operating leverage and the market recovery piece?
Yeah. Look, the glide path is, I think, one that we believe is very much achievable. It is going to be achieved. With the increase in rates, that could get pushed out a quarter, but let's not get into the quarter-by-quarter gyrations. From where we started, if you look at the Q2 results, number one, you had this issue with the mortgage pipeline that has about a $20 million-$25 million loss that's not going to be repeated. Because as you start to see changes we've made, along with some normalization that's taken place in the MBS market, we'll eliminate that. More importantly, in mortgage banking, you have this phenomenon when you take an interest rate lock, you recognize the revenue, but the expense associated with that lock gets recognized at the time that loans close.
As you see rates generally increasing, what happens is, number one, as I mentioned, you recognized your lock earlier, and so you have your expense later. But more importantly, as production slows down, your revenues are slowing down while you are incurring the expense. And furthermore, your amortization is running higher than the current run rate that you are seeing from your locks. And as you get a normalization or more of a true-up of revenue and expenses happening roughly at the same time, you are going to see an increase in ROE of a couple points. And right off the bat, you are going to get the benefit of the increase there. Then as we see the continued improvement of efficiencies in our consumer direct channel and our broker direct channel, our broker direct channel we forecast to continue to see growth.
Our market share is going to continue to grow there. Then you can start to see a couple points of ROE coming out of that. Then you layer in the fact that our technology expense right now is running at elevated levels because of the work we are doing in our production channel, in our servicing channel, in our drive to $55, as we will be talking about shortly, working on the Senstar acquisition, and really taking advantage of all of the AI tools that that cost is going to come down meaningfully. And so you can see a couple points taking place there in improvement in ROE. And then finally, as you start to see a normalization of the mortgage market, that will get it from the mid-teens to 20% and above just based on the size of the market.
I believe that you are going to continue to see this company get more and more efficient, and you will continue to see our costs, whether it is in production or servicing or everything that an organization of the complexity of ours is required to run as efficiently as it runs, you will see the costs come down in a very meaningful way.
Got it. That is helpful. Maybe just drilling down to the channel aspect. You have spoken in the past about being more selective and correspondent. Just given the competitive pressures and elevated GSE activity, can you provide an update on how you are approaching the channel today? And what the optimal channel mix looks like just given the current macro backdrop.
Yeah. Look, I think as you all know, correspondent is the key strategic initiative we have to really load the flywheel up for when rates decline, we have loans that we can refinance into our portfolio. The issue around correspondent is that there are certain parts of the market that we believe are a little bit too aggressive in how they're viewing recapture and the recapture opportunity in MSRs. We saw this especially take place at the end of last year. We've made the strategic decision that we want to be very precise as we have a reputation for being in terms of how we look at loans in the correspondent market. Okay, are these loans coming from correspondents that have slow prepayment speeds? Or maybe perhaps we recapture better out of brokers or loan characteristics surrounding the loan.
We've just been a little bit more particular, and been much more focused in our capital allocation of views in terms of bringing on new MSRs. That's what's led to the market share dip that you've seen. Having said that, we see in correspondent, certain market participants following our lead in terms of how we view servicing and how we view, in particular, the value of recapture of certain loans. I expect the market to follow suit, clearly by the end of the year. All it's going to take is one little rally and people see how quickly certain loans run off, that they're going to wake up and they're going to understand how they should be viewing the loans. Having said that, we are a leading government correspondent aggregator. We're growing our non-QM aggregation business. We're growing our jumbo business.
Even on the GSE front, we're making a lot of inroads in terms of the loans that the GSEs don't want to buy. We are an active buyer of loans that are typically executing better outside the GSE footprint. That's business that we're continuing to grow and accelerate in our correspondent channel.
Got it. That's helpful. Turning to broker, that continues to be a focus and the eventual transition onto Vesta remains an important milestone.
Absolutely.
How should investors think about the market share opportunities and competitive positioning in that channel?
Yeah. Look, I think that in the broker channel, we all understand what's happening in the channel today. But it's very early to tell what effect that's going to have come three and six months from now. I do think that we've established ourselves as a clear number two to the two market leaders in the broker channel. That channel's an interesting one because as a broker, you have to choose which one of those leaders you want to deliver to, or you want to originate your loans. We're in a very good, unique position to be able to really have an opportunity to do business with all of our brokers.
We have great technology that will be fully deployed by the middle of next year to the broker channel that is going to give brokers the opportunity to achieve the efficiencies that we ourselves are seeing in our consumer direct channel. That's something that's going to be a meaningful effect. As I talked about on the glide path, we expect to continue to grow share over the next two to three years in a very, very meaningful way. I think the broker direct channel is one that has been under a little bit of stress at the moment, but I think that it's too early to tell in terms of what that's going to mean. I believe that margins are going to have to come up in broker, and that's something that we'll see over time.
I also think that some of the phenomenon that we saw in broker, that we saw in correspondent with how people were thinking about MSR values and recapture, were starting to find their ways into the broker channel, and I think that that's getting mitigated rather quickly. By and large, we're really bullish on the broker channel, and I think it's something that you're going to see us play a more meaningful role in as time goes on.
Got it. To be clear, 10% market share is still the goalpost.
Well, it is. I think suffice it to say we have four months left of this year, so we are not going to get there this year. Look, market share from my perspective is a guidepost for all of you to think about as you are building your models, where we are going to be. I will tell you internally, for me, the most important and only driving factor is return on equity and the amount of return we are achieving. I think that we are going to get to 10% share. I know that. We got to get through this little blip here.
Got it. You have been reporting improving recapture rates across conventional and also government. How much of that improvement is being driven by tech? Are you seeing any early impacts from the trigger lead bill?
Yeah. The recapture numbers are really good.
They are the best I have ever seen in my career. Under the leadership team that we have in our consumer direct channel, they are hyper-focused on recapping the portfolio. We are seeing conventional recapture rates approaching 30%, which I have never seen in my career. On the government side, we are north of 50%. The recapture numbers are really strong. I think it starts with leadership, and then it starts with the workflows that we have in place and the marketing initiatives you have in place to get the recapture. Then you layer in what is happening with the technology. As I pointed out earlier, you just being able to be more efficient as a loan officer allows you to be able to go after more loans. That is the benefit of Vesta and the loan origination technology that we have in place.
Then, we've introduced our natural language virtual assistant, or NLVA, that's allowing borrowers to self-serve, and that's leading to increased activity in nights and weekends and holidays. That's something that we introduced in our consumer direct channel just 60 days ago. That's something that we're already seeing some exciting results. Look, the goal, and we're going to get there, is to be able to have what we call the driverless mortgage, where borrowers can just self-serve, and be able to originate a loan without dealing with a loan officer. That's something that is going to be more and more prevalent throughout the industry. But look, I think there is something that we have in the fact that our technology that we're using is legacy-free, and it was built to allow for quick adoption of AI.
Most importantly, that AI is being built and driven by the business. That's something that, throughout my close to 20 years at PennyMac, I've always felt that the business should drive technology, and they should own the building of it. AI is allowing us to do it. So there we're seeing real good results. Look, the trigger lead bill has helped. Okay, we shouldn't ignore that. It's given us a little bit more value in terms of owning an MSR as opposed to when you pull credit, you have 60 different lenders calling the borrowers and hassling them. So we're seeing good results coming out of that. Look, that's contributed to these historically high recapture rates.
Got it. Just turning to technology, you've highlighted significant improvements in cycle times, also targeted 80% automation by year-end 2027. What are the most important benefits investors should expect as these initiatives scale, both from cost and a growth perspective?
Look, I think from your position out there, the exciting part should be the volatility associated with the staffing up and laying off of human capital is getting greatly reduced and very quickly. That should lead to more stable earnings in the company, okay, and reduce some of the volatility around the move in interest rates. As I talked about earlier, we were keeping excess capacity for potential rallies, and the need for holding that has been greatly diminished. So as we look at what our needs are going to be for a 50 basis point rally and 100 basis points rally, it's come down a big number. So right off the bat, I think that to me is one of the most exciting parts about the technology that we're building in the organization.
I think that, look, from a technology expense standpoint, as I talked about, we walked into the year with a lot of initiatives. I think as we see those initiatives complete, we're going to get real benefits as a company. Obviously, we'll be spending less on technology, but more importantly, we're going to be getting the benefits of the technology that's being built that will lead to greater profitability in the company.
Got it. So you targeted a reduction in servicing costs to $55 per loan from
Correct
north of $80 today. Where are you today relative to that goal? What are the most important drivers remaining?
Yeah. We started the year at $89 a loan. We expect to finish the year at $80. I expect to be at $70 by the end of 2027. Sometime by the, call it end of 2028, beginning of 2029, I'm really hopeful that we'll be at the $55 a loan. Look, when you have a portfolio of 4 million loans, you can do the simple math. For every dollar we save, it's really meaningful to the company. The drivers of the $80 - $55 are really around a few things. Number one, there's a heavy focus on default. There's a lot of work being done on the AI front to give the distressed borrower the opportunity to get solutions, to get questions answered, and to deal with their default in a much faster, more efficient way.
I believe that there's many cases where a borrower in distress would rather deal with a natural language virtual assistant or deal with something AI-related as opposed to dealing with a human. I think that there are opportunities. We saw during COVID, 90% of our borrowers took forbearances without speaking to a customer service representative. So building technology solutions and building the alternatives is something that we're very focused on. The natural language virtual assistant is in many ways more powerful in servicing because it's very prescriptive what you need to do in servicing. So if you get into a period of high distress, the need to add more people or the distress it puts on our servicing people are going to be greatly diminished.
Furthermore, I expect the technology that the AI that we're deploying on our own technology is going to benefit our customer service representatives. So whether it's the inline QC when they're on the phone with the borrower, or whether it's just the ability for the borrower to self-serve, is going to allow the more simple issues to be dealt with through automated means and allow our customer service representatives to deal with the more complex issues in a faster, more efficient way. Then finally, there's a lot of work that goes into the investor accounting function, the compliance function, the complaint functions, that have a lot of people decked against it, that we're quickly deploying AI that will help drive down cost as well.
Got it. That's helpful. Just maybe switching to credit. There's been some headlines regarding government loan delinquencies trending upward this year. What's your outlook on government loans and the broader mortgage credit? Anything PFSI is doing on the servicing side to help performance?
Yeah. Look, our delinquency numbers are continuing to hold in really nicely. On the government side, the government servicing is naturally going to print higher delinquency numbers. We have always taken the position that we want to price servicing at the loan level. What that means is that we've had a tendency to lean in correspondent more in the direction of higher FICO, little lower DTIs, little lower LTVs, and really try to focus on giving attribution on what we perceive is going to be the better-performing servicing. As a result, we have a servicing portfolio that is a little bit less credit sensitive than the market as a whole. So that leads to better delinquency numbers coming out of our government servicing portfolio. On the conventional side, that portfolio continues to operate very strongly. We're seeing delinquency numbers pretty low there.
And it's something we're very focused on, but I'm not, as I sit here today, really concerned about a credit event as we sit here today. Now, if you see unemployment increase or you see other things take place in the market, that could change. But as I sit here today, the portfolio continues to perform strongly. And as I said, at $530 million a quarter of servicing fees, it really speaks to the value of the balanced business model that from when we started this company in 2008, we've set out to create what I think is today the gold standard of the balanced business model.
Got it. So at this point, we'll switch to PMT for a few minutes. Can you just talk in general around the strategy there and what excites you about what PMT is doing?
Yeah. So PMT, as many of you know, is a really unique REIT that has been set up to really take advantage of the synergistic relationship it has with PFSI. And so over the years, we've been able to use that synergistic relationship to create really unique mortgage-related investments. As we started the year, we looked at PMT and we said, okay, a couple of things. Number one, PMT has a lot of mortgage servicing rights. And the returns that they were achieving on the mortgage servicing rights when you looked at it versus returns of securitizations of owner-occupied loans or securitizations of agency-eligible investor loans and second home loans, or even securitization of jumbo loans, those latter investments look to provide better returns than investments in mortgage servicing rights. And so what we've done in PMT is de-emphasize the creation of new MSRs, and we've actually sold some MSRs.
We had a $13 billion sale that closed this quarter, and we are taking the capital and we're redeploying it into credit investments. And the credit investments that we're creating in PFSI are mid-teens returns in the base case. In a stress case, they're still high single digits. And we believe that as we repurpose capital out of what was going to go into MSRs into credit-related investments, we'll continue to see the returns in PMT go up. And that's something that we're really excited about. And look, there are things we can do to speed that along. There are things we can do to continue to look to grow the returns in PMT. But suffice it to say, we are moving with great urgency to get the returns in PMT back up to double digits.
Got it. How do you view the current dividend of $0.40, and how should investors think about dividend coverage GAAP over the past few quarters?
Yeah. So look, as a REIT, we have to pay out taxable income. A lot of times you'll have a mismatch of sorts between taxable income and GAAP income. It has been our dividend policy since we started the REIT to try to hold the dividend as close to GAAP income as we can giving consideration to the requirement to pay out taxable income. We've had this issue over the past few quarters where the dividend has been higher than GAAP income. We continue to look at this and it is my stated goal to get the GAAP income and the dividend closer to one another. As we work through this period where the taxable income gets paid out as a dividend, the dividend itself should migrate closer to GAAP income. Now, that doesn't necessarily mean the dividend is coming down. Okay.
Or if it does, it's not going to come down to where the GAAP income has been, to the last quarter or two, because at the same time as we redeploy our capital into mortgage servicing rights, we're starting to see the GAAP income go back up. I think that the final thing that I'm really mindful of is we like to maintain a stable dividend. As we look out over the next four quarters, we're going to look to really establish a dividend that we, as I said, is reflective of GAAP income, that we believe can be a stable dividend, and it's something that's reflective of the performance of the company.
Got it. That's helpful. At this point, we'll just pivot back to PFSI. Just want to get your thoughts on just capital allocation priorities, how you're thinking about capital allocation, whether or not you can do a buyback.
Yeah. Look, I think that we have a robust capital allocation framework within the company. Buybacks are on the menu of capital allocation. Having said that, we're very focused on our non-funding debt leverage, and we're sitting today at 1.8 x, and I'd like to get that number down a bit. There's natural ways, obviously, that will come down with increased profitability or in a rally it'll naturally come down because of the way we run the servicing hedge. I don't want to see the leverage go up. By the way, I would expect that leverage number to come down over the next 12 months just based on what I'm seeing from our financial forecast. But leverage is at the top of my list.
I then say, okay, from a capital allocation standpoint, and we're seeing it in how we think about correspondent, do we want to invest in the MSRs? Do we believe the MSRs can give us the mid-teens returns that we've been very vocal as saying is our cost to capital? Even if it's at the risk of foregoing a little bit of gain on sale, we're going to be very focused on making sure that we run this company with a culture and a desire and a need to deliver mid-teens returns to investors. That's something that is a part of our capital allocation methodology. Then finally, the investments in technology plays into it as well. As we look at where we started the year, as I talked about, we had a lot of technology initiatives.
Just from a return standpoint, made a lot of sense that we opted to sign up for, and we did so with a clear view that we felt production was going to stay strong. We expected, if you think about it, an element of rate cuts to come into play, and we felt that we could get through the year in 2027 and be able to afford the technology while reaping the benefits. As we sit here today, we're already seeing the benefits of it. As that technology investment comes down, that should give us even more capital to invest in the company.
That's super helpful. We have a few minutes left. I'll open it up for any questions. Shy crowd. Okay.
There we go. We got one right here.
I will just repeat the question. What is the primary method for increasing recapture rates?
There is a few things. Number one, it is how do you market to the consumer, okay? You need to be really careful that you are not inundating them with emails and phone calls, and you are creating a negative experience for that borrower. Secondly, you have to create a user experience for the borrower that comes to you that meets their needs. What is that? Do they want to speak to a LO? Do they want to originate the refinance through chat? Do they want to come online and be able to do it without a human in the loop? Do you offer them a kind of a single-thread user experience, whether they are in the servicing, whether they are coming in through servicing or through consumer direct.
If you call in to find out if your taxes have been paid and you can refinance your loan and save $300 a month, I want to be able to make it a seamless experience for you. Then if you want to do it with a human out of the loop or with a human, then you have that opportunity. Mortgage banking historically has been very siloed between production and servicing. Further to that point, we run two call centers. There is no reason for that. As we blur the lines between are you a servicing customer or are you a production customer, that is going to lead to higher recapture rates.
Finally, I think that you want to market the company to your servicing portfolio to help them understand that although they may have originated their loan with a correspondent or a broker, that you are a top producer, that you offer refinance capabilities, and you can close a close-end second within seven days, or you can close a rate and term refinance within 10 -1 4 days. So the marketing of that is vitally important. And those are the marketing initiatives that we speak about.
Got one question over here.
Yeah. Look, I think that historically, mortgage banking, in particular, has struggled with the variable cost structure. In particular, when you see rates come down, there's a lot of time and effort that goes into hiring people and getting more space and having to deal with the infrastructure you need to meet the demands of the market. The really exciting part of what we have done in our consumer direct channel has created a cost structure that we can staff up very quickly through just increasing the amount of AI agents that we need to meet the demands of the borrower. Okay? And those AI agents both are on the front end and on the fulfillment side. So just to give you some raw numbers, we on a 100 basis point rally, we were forecasting we would need 450 LOs. That number is down to 280 right now.
That's not because we're going to do less recapture or less loans, we're just going to become that much more efficient. And that's what we're seeing out of our technology. Okay? Similarly, on the mortgage fulfillment side, the amount of people we needed to process loans is coming down materially, okay, in a 100 basis point rally because of the AI agents that are getting developed on the technology. So what that means is you just have just a lot more efficient operation that is going to reduce the earnings volatility of the company.
Okay. I think we're at time, and we'll just end it there. Thank you very much.
Thank you, Terry. Thank you all for your time today.