8.27.26 Industry Calendar; MIAC’s Dan Libby on MSR Analysis; Caution Abounds

Robbie Chrisman Welcome to the Chrisman Commentary, Daily Mortgage News Podcast. I'm your host, Robbie Chrisman. Topics on today's episode include the upcoming mortgage conference schedule, there's headlines, and there's reality, and my interview with my ex Dan Libby on the long-term drivers of MSR value, portfolio construction, and hedging effectiveness. Here, take a listen, do a little preview. Robbie Chrisman Do you see lessons from your analysis being kind of the new norm? Or as you alluded, everything's kind of thrown out the window here and and uh we're in completely new times, and you expect that to uh you know be unpredictable going forward? Dan Libby It's definitely we are in different times. As I said, the traditional factors have really have a low explanatory power in terms of the level of pricing. But you know, the absolute price levels for bulk MSRs do not seem to be out of line for where they have been historically relative to where primary mortgage market rates are. More of the price, the price invariability seems to be driven in terms of what's happening in the primary market for MSRs, the SRP market. But you know, again, the the price movements still seem to be following the same type of relationships and transmission mechanisms or factors uh driving those price movements. What could be the change, the change in the regimes that bring about a shift in this relationship? Every regime that we've looked at in the past, when we look at you know, what are the factors that have caused those, what are the catalysts that cause those regimes to change, it's always something different. So in the current regime, I think that you know the thing to think about is large shifts in rates, large shifts in participant involvement in the markets. And it's always something to be on the lookout for, is what brings about the end of one regime and the beginning of another. Oftentimes it is, you know, the Fed's behavior uh in terms of a movement and rates, but it's it's something that's clearly more easily identifiable in the rearview mirror than looking forward. But you know, it's it's certainly something to be on the lookout for. Robbie Chrisman Yeah, we've certainly seen a shift in the Fed's behavior here recently. So it'll be interesting for you and me to talk in a year or two and look back and see if that changed much. When we look at this analysis that you've done in the MSR market, the SRP market, how do you take all the observations and actually use them in portfolio construction, stress analysis, and hedging? Dan Libby The practical application is not to replace good forward-looking valuation models with historical regressions. In fact, it's it's possibly the opposite. How do we use the um historical regressions in the model work that we do is really to tell us what similar, you know, what similar assets actually did uh when markets became stressed or those assumptions broke down. A forward-looking framework, an OES framework or other forward-looking valuation framework tells us what you think the asset may do in the future under a particular set of assumptions. But it's really looking at history that gives you a sense of what markets actually did do when they were confronted with uh different challenges. And there's really never two historical sets of challenges that are the same. It's typically a combination of uh factors to look at and to be aware of. But it's only through historical work and appreciating history that you can be prepared for environments that are upcoming that may not unwind exactly as your forward view anticipates. So for portfolio construction, the historical work gives us a much broader range of realized valuation and risk relationships that we could observe from the relatively short history of any individual portfolio. For stress testing, instead of simply asking only what happens if rates rally 100 basis points, we can ask what happened historically when rates rallied and the curve changed, the mortgage basis moved, primary mortgage spreads changed, and liquidity conditions deteriorated. So a natural outgrowth of this is that as you're looking forward in terms of uh modeling your portfolio, typically you want to include some type of stress analysis or scenario analysis. It's through that type of historical uh viewpoint that you can inform your way that you create your stresses and look at you look at your portfolio from different points of view. So for hedging, I think the lesson is more about humility and less about precision. I don't need to know the MSR's true duration to two decimal places. I need a hedge that is directionally right, robust across plausible environments, sized within a risk tolerance that recognizes model uncertainty. So basically a hedging approach that is fair, consistent, and robust. And it's history, looking at various regimes where that becomes useful. It gives us that historical bound, that historical, those historical boundaries around what is plausible and what plausibility actually means. And ultimately, I want to compare those long-dated realized risk sensitivities with the forward-looking durations coming out of our OES models. And where they agree, that gives me confidence, and where they disagree, that's exactly where I want to ask more questions. So I view the historical work as more as another lens on the same asset, not a replacement for evaluation and forward-looking models, but a way of challenging them. Robbie Chrisman Thanks to Experian for sponsoring this week's podcasts. From lenders and landlords to employers and consumers, Experian helps connect the lending ecosystem with data and insights needed to make faster, confident decisions. Lead a smarter housing journey with Experian, and you can learn more at Experian.com/slash mortgage. You know that on Crispincommentary.com we have a complete industry events calendar. It's pretty easy to go check out. Book those flights in advance. They're not cheap and have been going up given the war in the Middle East. Lenders and vendors are casting a critical eye on ROI, given how much it costs to send an individual thousands of miles away week after week. In September, we have in Hood River, Oregon, the Penny Mac Yearly Conference from 913 to 915 and Dallas from 915 to 16. The Lone Vision Innovation Conference takes place. MBAMW is on September 17th in Washington, D.C. There's Acuma September 20th in Las Vegas. From 921 to 923, the New York MBA conference is in Schenectady, New York. I will be speaking there. Compliance and Risk is from September 27th to 29th, hosted by the MBA. In Yipps Planning near Detroit, the Michigan Mortgage Lenders Association is having its annual conference October 4th to 6th. October 11th to 14th is MBA Annual in Chicago. That also includes MBA's Tech Exchange on October 14th in Chicago. On November 18th, there's the Mortgage Bankers Association of St. Louis annual luncheon. I think my dad's going to that one. And if you really want to plan ahead, there's the 2027 IMB conference set for January 25th to 27th in Tampa. Optimal Blues Summit, February 1st to 3rd in Scottsdale, Lenders One Summit, March 7th to 10th in Texas, IC, March 15th to 17th, Great Rivers. Anyways, we could we could go on and on and on. Meanwhile, investors are voicing concern over U.S. Treasury Secretary Scott Bessent's recent decision to increase long-term government debt purchases, fearing it could undermine the Federal Reserve's efforts to manage inflation. The move, aimed at lowering borrowing costs, has received criticism since it is price management rather than liquidity management. Besson's expanded Treasury buyback plan is narrowing long-end swap spreads and boosting demand for bullish U.S. bond futures options as traders price in a policy backstop. The move has eased yields and squeezed bearish positions, though investors warn it does not address the deficits in debt supply, driving structural pressure on long-term borrowing costs. Perbescent, the U.S. Treasury will maintain its scheduled debt auctions despite doubling planned buybacks of longer dated securities. Expanded repurchases of 10 to 30 year securities are intended to support market liquidity after yields approach two decade highs. The first expanded buyback is scheduled for September 10th, while the Treasury has not disclosed how it will fund the purchases. The Fed already owns $1.6 trillion, leaving $4.1 trillion with a maturity of $10 plus years that can be bought. In a market that big, buying only $2 billion more does not matter. In 2011, Bernanke did his own operation twist, which was initially worth $400 billion, but wound up being $700 billion. For today's interview, I wanted to welcome back to the show MIAC's Dan Libby to talk about the long-term drivers of MSR value, portfolio construction, and hedging effectiveness. He's head of hedging and funding services at MIAC and has spent his career across mortgage research, structured transactions, fixed income investing, and risk management. Robbie Chrisman We are here today to talk mortgage servicing rights. Always a topic of high intrigue, at least for me and in a certain sector of the mortgage industry. And it should matter to most everybody in the mortgage industry because uh servicing and recapture, those are those are huge deals right now and uh big plays being made by some of the largest players. And so I'm very excited to have you on to talk about the MSR market. Anything particularly interesting going on in the MSR market? And uh I know there is, so I'm sure you'll have something to say there, but uh kind of can you bring in the historical side of things and and kind of talk about how there's there's been a long history of MSR pricing and how that uh compares to where we are today? Dan Libby Yeah, thanks, Robbie. So, you know, as in in starting off this um this talk, which is really the result of some research that I have been doing at MIAC, um, it really all began from, you know, sometimes the most interesting questions come from the simplest observations based on the most straightforward premises. So at MIAC, we have the luxury of having some proprietary price histories, both for current coupon, MSR, new issue, and current coupon MSR secondary pricing, going all the way back to pre-crisis 2008. And when you put that history next to primary mortgage market rates, you can see the interesting observations. There are periods when we recycle back to the same level of rates, but MSR prices don't recycle back to the same price levels. Or even more fundamentally, you some, you know, some could ask when you consider that we're looking at current coupon pricing, why shouldn't it be that the prices are invariant to rates altogether, or at least you know, nearly so, uh, either in the primary market or in the secondary market? So the the basic question to start with is all of this price history really reflecting just technical or random factors, or is there something more rare that we can conclude in looking at this data? And then when you start to get into that, what's even more interesting is that what relationships you do find, they're not stable. You know, there are periods when mortgage rates explain MSR valuations extremely well and when they explain very little. And so to begin this process, to begin this investigation, I started by dividing the history, this long-dated history, into eight distinct market regimes and asked what else is driving valuations within each one. And in today's regime, what I call the sort of higher for longer regime, it's particularly interesting. MSR multiples are high in absolute terms, and people frequently describe bulk MSR prices as expensive. But we note that mortgage rates are also extremely, extraordinarily high relative to most of this history. And so bulk pricing doesn't necessarily look as rich as the absolute price multiple suggests. And by doing this type of rigorous analysis, we can quantify that. In fact, the more striking observation might be that new issue, MSR pricing, what we call servicing release pricing or SRP, relative to the enormous increase in mortgage rates, SRP multiples have appreciated surprisingly very little. So the question that I was interested in wasn't simply are MSR prices rich or cheap, it was what has historically driven MSR prices and how have those relationships changed across regimes? And what does that tell us about today's market? Robbie Chrisman I do have one more question on the historical nature of things because you've obviously studied MSR pricing across different historical market environments. What can we learn, particularly for portfolio construction, stress analysis, and hedge effectiveness? Dan Libby First lesson is that context matters enormously. As I said, the same level of mortgage rates has produced very different MSR valuations at different points in history. So here we looked back at historic at history going again, going back to 2008, across eight different regimes, beginning with the great financial crisis, QE1, QE2 slash HARP, the taper tantrum, all the way up to today's what I call higher for longer environment. And these regimes weren't chosen simply just to improve the statistical fit. They represent genuinely different economic and market environments in which different drivers of mortgage valuation and risk changed. What are called various market or financial transfer mechanisms occurred to impact the markets and ultimately to impact MSR prices. So sometimes the mortgage rate itself dominated in these transfer mechanisms. And at other times, Fed intervention, refinance behavior, liquidity, the mortgage basis, the shape of the curve, many other factors come into impacting these various regimes and being the predominant factor, economic factor, across these various regimes. All of that is really important for portfolio construction because historical relationships aren't necessarily stable. A hedge or stress relationship calibrated during one regime may behave very differently when the market transitions into another. And I have personal experience having lived through that period. The initial shock came through the long end after Ben Bernanke's taper comments. And the Treasury and Curve movements became extremely correlated. And if you had the regime driver right, you could manage your way through it. But if you didn't, the conventional relationships you were relying on become could become very painful very quickly. So the broader relationship isn't that history repeats itself, it's that history gives us a set of very different real-world experiments. We can reserve which relationships remained stable, which broke down, and which additional factors mattered when they did. So that gives us a better framework for asking today's practical questions. What is really driving MSR value? What risks are we actually exposed to? What are the hedging the hedges protecting us against? And what risks matter in the current regime? Robbie Chrisman Well, now that we have a better framework, let's talk about the research a little bit. How did you structure the analysis and what did you learn about the factors that have historically driven MSR valuations? Dan Libby Yes. So we started really simply. As I said, we have long-dated hit price histories going back to 2008. So the first question was how much of MSR valuation can we ex can we explain just with the primary mortgage market rate? So that became our simple model, MSR prices versus the PMR, primary mortgage market rate. And the surprising result was that even such a simple relationship explains a meaningful amount of MSR valuations over time. So, like I said, this became our baseline for which to evaluate the efficacy of other model constructs. It was very useful for looking at the intrinsic valuations and outlier prices over time. But the primary mortgage market rate is really the end result of a transmission mechanism itself. It can be decomposed into the treasury rate in the primary secondary spread in the curve shape, all of which matter. So we progressively decomposed that relationship rather than simply throwing in more variables into the regression and performed the next level of regression by using what I call a traditional model, using PMR, the treasury curve in the mortgage basis. After that, we took a different construct of the model, what I call the structural or attribution-based model. We took a one step further and divided the mortgage basis into primary, secondary spread, and secondary swap spreads to get more explanatory power. So what we ended up was three basic models across eight different regimes versus an unvarying relationship across time as another control element. So we ran those relationships separately across eight regimes and also those regimes for specific relationships with a single global relationship across history. One of the useful things we learned was that a model can fit MSR prices reasonably well within individual regimes, while a single global relationship across the entire spectrum, across the entire history, performs much worse. So it isn't just that volatility or model error changes across regimes, the valuation relationship itself changes. But what's important is that even if the explanatory power is low, it isn't necessarily a failed result. It in and of itself can have informational value. It tells us that the observable rate and spread factors that we were measuring aren't explaining the price level as well as they may in some other regimes. So other extraneous factors may be influencing valuation. So this became particularly interesting for today's higher for longer regime. Our historical valuation relationships have weakened dramatically in the current regime. Some of those market factors still retain meaningful explanatory power for month-to-month price changes, however. So the exercise evolved from building the best regression into something much more useful, identifying when historical valuation relationships work, when they break, and what those breaks may tell us about the market. Robbie Chrisman So let me bring this back to the practical question that we started with today. And that is what did the analysis actually tell you about MSR valuation and risk, particularly in today's market? Dan Libby I think there are four practical conclusions, and they're somewhat related. The same mortgage rate does not necessarily mean the same MSR value. PMR is clearly the dominant long-run valuation factor, but its relationship to MSR price changes materially across regimes. So, like a 6.5% mortgage rate in one market environment that does not necessarily imply the same MSR multiple as a six and a half mortgage rate would in another, even though they're still talking about a current coupon MSR valuation. So that's where the structural attribution model became most useful. But treasury rates, the shape of the curve, secondary mortgage spread, and the primary secondary spread help explain why apparently similar mortgage rate environments can produce different MSR valuations. And when those factors stop explaining the valuation, that's important too. It tells us that looking outside the traditional rates fact, looking at the factors like liquidity, capital availability, financing, supply and demand, recapture economics, and other market technicals may become dominant. So the second thing that we learned is that the relationship between bulk pricing in the secondary market and new issue price in the SRP market don't always tell the same story. Sometimes they diverge. So one of the things that surprised me in the long history is how differently secondary and new issue pricing can behave, even though both represent the economics of mortgage servicing. And that distinction looks especially interesting today. People frequently describe conventional 30-year bulk MSRs as lofty because the absolute product price multiples are high, but primary mortgage market rates are also high. When I compare the bulk multiples with the market environment, I'm not convinced that the data says that bulk MSRs are obviously expensive relative to their primary fundamental valuation drivers. It may be that the servicing release premium or the price of MSRs in the primary market may actually be more of the interesting observation. SRP multiples have appreciated surprisingly very little relative to the enormous rise in the primary mortgage market rate. So the relative value question may not simply be why is bulk so rich? It may also be why hasn't SRP prices appreciated more in the primary market. Third, today's higher for longer environment appears to be an unusual valuation regime. This is probably the most interesting result. In today's higher for longer environment, the traditional rate and spread vectors seem to explain remarkably very little of the level of conventional bulk valuation. The explanatory power of the models is close to zero. But yet, when we change the question from valuation levels to month-to-month price changes, those same underlying factors still explain a very high percentage of the price movements. So, in other words, factors still seem to matter very, very much for how prices move, in other words, for risk or for duration, but they don't explain as well where the absolute price level should be. I think that's an important distinction. The risk transmission mechanism still appears to be functioning better than the historical valuation relationship. So that suggests that there's something else embedded in today's price setting mechanism that our traditional rate and spread factors aren't capturing. So I don't want to overstate it, but the obvious candidates are the availability of capital, financing and liquidity, the evolving buyer composition, supply and demand technicals, economics for the servicers, and recapture value all seem to be a com all seem to be a combination of factors that are influencing price value. But again, the hedging effectiveness seems to be unaffected. Fourth is just that point. Valuation risk and hedge risk are related, but they're not the same question. So the last question was whether we could use the same long-dated price histories to learn something about realized MSR risk and hedge behavior. So we turned the regression around. Instead of explaining the level of the MSR multiple, we looked to explain the month-to-month changes in the multiple with changes in treasury rates, slope of the curve, secondary spread, and primary secondary spreads. That gives us historically realized partial price sensitivities, effectively a way of observing how MSRs actually responded to those risk factors across different regimes. The risk relationships appear materially more stable in today's market than the valuation relationships. And so that's important for hedging. A model can have a difficult time telling you exactly what an MSR's price should be, but still do a reasonably good job of describing how that MSR will respond to changes in rates or spreads. And ultimately, that's one of the distinctions I care most about as a risk manager. Valuation uncertainty does not necessarily mean hedge ineffectiveness. So it does mean that we need to understand what our hedge is protecting us against and what risk may sit outside the model. Robbie Chrisman All right, Dan, let's bring this all together and put it in a bow. So after looking at almost two decades of MSR pricing and risk, what's the main takeaway? And I guess maybe more importantly, what should market participants be watching from here? Dan Libby Well, so the biggest takeaway is that MSRs are extraordinarily regime dependent. We tend to think of them about them in terms of a multiple, a duration, or a convexity number, but none of those numbers exist independently of the market environment that they're produced in. The history since 2008 includes all these various regimes, the great financial crisis, Fed extraordinarily Fed intervention in the recent regime, higher for longer rates. And through all of that, one relationship has been persistent. Mortgage rates matter enormously to MSR values, but the transition mechanism changes. Sometimes treasure rates dominate, sometimes the mortgage basis, the curve liquidity factors dominate. What appears to be happening with valuations in today's market, the variables that have historically have explained asset valuation have simply stopped explaining very much. However, there's comfort in the fact that risk metrics and risk parameters are still performing extremely well in terms of evaluating the uh month-to-month price movements of this asset, which should be comforting for the for those of us that are managing MSR MSR risk. I'm reminded that there's an expression from history that says all models are wrong, the good ones are useful, but I would add, but especially in the proper hands with an experienced person uh at the helm. I don't see that as a reason to distrust models. I view it as a reason to understand what the models know and more apparently what it doesn't know. History doesn't tell us what the next regime will be, but it gives us sort of a laboratory containing a remarkable range of things that have already gone wrong and occasionally gone right. So if we understand how MSRs actually behaved through those environments, we can construct portfolios, stress them, and hedge them with a little more humility about what we know and much better appreciation for what can change. And I think that's the simplest conclusion of them all. Models tell us what we can expect, what we expect should happen. Markets tell us what did happen, and good risk management requires paying close attention to both. Robbie Chrisman I definitely heard some optimism in there, which is good. Love having you on the podcast. And this is very technical for a lot of people out there, but it's certainly valuable and certainly matters to the mortgage industry. So thank you very much for your work and thank you for joining me today. Dan Libby Thank you, Robbie. I really enjoyed it. Look forward to speaking with you again. Robbie Chrisman Caution is a good word to describe what I'm currently seeing in the bond markets. Agency mortgage-backed securities and U.S. treasuries gave back some of their recent gains yesterday as economic data showed a resilient economy without meaningful downside surprises. Persistent inflation remained a concern with headline PCE at 3.7% and core PCE at 3.3% year over year. Q2 GDP held at 1.5%, but upward revisions to consumer spending and core PCE, combined with stronger than expected durable goods orders, reinforce the view that economic momentum remains firm. The sell-off extended into the afternoon yesterday as the $70 billion five-year auction drew weaker demand than Tuesday's two-year sale, pushing 10-year and shorter-term yields to fresh highs. Cautiously Constructive also describes energy prices, which reflect optimism over Middle East diplomacy despite no definitive breakthrough. Markets are pricing in a more cautious, there's that motif again, Fed, pushing the 10-year treasury yield back above 4.65%. Together with other signs of a July economic slowdown, recent data gives the Fed greater justification to remain on hold rather than raise rates, despite inflation remaining above target. Fed chair Walsh heads to Jackson Hole with an opportunity to rebuild bond market confidence by emphasizing the Fed's commitment to restoring 2% inflation while avoiding explicit forward guidance. Weekly initial claims in at 203,000 about as expected, continuing claims 1.778 million, July advanced international trade deficit in at 1.118 billion, kicked off today's economic calendar. Later today brings a treasury auction of $44 billion of seven-year notes, and the Fed's economic symposium in Jacksonville begins. After this volume of news, we find agency MBS prices a little change in Wednesday's close, the two-year yielding 4.22, and the 10-year yielding 4.66, unchanged from yesterday's close. Let's wrap up with a joke and some housekeeping. Hippos can swim and run faster than humans. What does this mean? Well, obviously the bicycle is the only way to beat them in a triathlon. Thanks to Experian for sponsoring this week's podcast. From lenders and landlords to employers and consumers, Experian helps connect the lending ecosystem with data and insights needed to make faster, confident decisions. Lead a smarter housing journey with Experian, and you can learn more at Experian.com/slash mortgage.