Speaker 1Elevated household debt levels, high inflation, and economic uncertainty. As recession fears mount, analysts are digesting an array of data for signals to gauge the financial health of the American consumer. DVO1's In the Trotches of Structured Finance podcast dives into consumer data to explore macroeconomic trends, loan performance, and monetary policies. Here's your host, Vadim Verhoglad.
Speaker 2Welcome everyone to another episode of the In the Trotches of Structured Finance podcast. I'm your host, Vadim Verhoglad, and I'm glad to be back on a more normal cadence with you guys. We're expecting a more normal one episode a month, or in some cases, more than one episode a month cadence, so we're hoping you stick with us and listen in. So today's episode will be broken out into two parts. In the first, we're going to be taking some excess questions from DVO1's recent consumer credit webinar. We weren't able to answer all the questions in a lot of time, but I did want to address them as some of them are very pertinent and fairly relevant. In the second part, we'll be doing our periodic update on consumer credit performance across most markets. We're not going to touch on auto because we just did an auto-dedicated episode last month and we largely covered performance trends then. In this podcast, we'll be leaning heavily on several sources. First, for those of you who listened to the webinar, this is follow-up questions from there, but most of the data is going to be coming from our recent consumer credit performance reports. With that out of the way, let's kick off. So we had a lot of performance questions in the webinar, and I think a lot of it has to do with the fact that we've constantly been touting that consumer credit performance has generally been very strong. But of course, when you say things like that, people will naturally lead off and they'll naturally ask, what does strong look like and what does strong look like across recent villages? So one of the questions we got, and the one I'll lead off with, is someone asked whether we saw a similar trend to what TransUnion was reporting, where TransUnion's quarterly origination rate was down by 0.5%. TransUnion report said that originations on prime and super prime from Q2 and Q3 of 2024 was trending worse than 23 and 22, but that near prime and subprime were trending better, and if that matches some of the trends that we're observing in our own data. So there's a lot of nuance in this question, mainly because when you're thinking about a vintage as recent as Q3, 24, you're still talking about a vintage that is very much in its ramp up period. What do I mean by that? When a new vintage comes about, the first six. Months pretty much don't have any default rates. Why? Because it takes a loan four months to go from first delinquency to default, and it's very rare that a borrower defaults in the first two months of taking consumer-antiqued loan. In large cases, because issuers will programmatically kind of weed out fraud, and they in some cases will actually go and buy out what they call early pay default loans. So default rates for the first six months are fairly de minimis. Delinquencies will start to rise in months three. Basically, they'll start increasing on three, and they'll continue to increase pretty much for the next year or two, and at that point, the vintage gets very small and behavior gets very cuspy. So when you're talking about a vintage like Q3, 2012, 2024 rather, you only have at this point 13 full months of total performance trends. So you're very much still in your delinquency ramp up period, but more importantly, you're only in the early stages if you charge off ramp up period. I mean, you're Q1. You're on the loss ramp. So it's still fairly early to say, hey, this is how the vintage is definitely performing. But with that out of the way, let's tackle the question anyway. We spotlighted in the last Q consumer-unsecured reports that the overall, the overall, we're talking combined, 2023, 2024, Q3, Q4 vintages are seeing modestly steeper impairment increases in kind of coming out of the gate as compared to Q1, Q2 of 24. Now, that's a tough comp because Q1, Q2, 24. There is some really, really good performing vintages. On an ROI basis, these are record high vintages, pretty much across them. And with that said, while it is, these vintages are certainly weaker than Q1, Q2, 24. They are better than 23, particularly early 2023 trends. But within that overall framework, it is possible that you're seeing a little bit more underperformance in top grades offset by basically stable or some better performance in the middle and bottom. That's caveat number one. And caveat number two. When we say performance, a lot of people will immediately say performance. They'll go to delinquencies because that's the most kind of pertinent indicator. The thing to keep in mind is delinquencies in isolation are certainly bad. But when you're thinking of delinquencies, you're comparing on current balance and current balance. That means there's also an impact from prepayments if vintages are paying particularly fast, which has certainly been the case for some of the 2024 vintages. So instead, what I suggest. I suggest to people that are thinking about total performance is to look at just return on investment. Really simple metric. It tries to encompass all of these factors with the downside being that our return on investment or ROI is very much an original, like kind of is very much based on original balance. So it can be skewed and it can be capped by the fact that if a vintage is paying particularly fast, you may have lower ROI, not because of bad performance, but just because of the increase in return on investment, because of really rapid payback. And that's kind of some of what we've been talking, what we're seeing in the 24 vintages on an ROI basis. Therefore, taking all that into account, we are seeing a slowdown on ROI, specifically on some of the top grade ROIs, actually, where the top grade ROIs for the 20 later 24 vintages are actually running a bit slower than they are for early 24 and even some 23 vintages. That is certainly true. The story you compare. The story you compare to early 24 is performance. Absolutely. Late 24 top grades are worse than early 24. But as compared to 2023, I think the answer there is more prepayment driven rather than outright bad performance. However, there is one thing that I want to very, very, very importantly keep in mind. We personally have some questions about the TransUnion reports. Fundamentally, for our questions, we find the reporting for TransUnion to be, in some cases, kind of spotty. And for some of that spottiness, look no further. Then there are monthly insights report, which we used to publish on this regular monthly. It was part of our performance report. But when we look deeper, when we looked at just the volume of restatements that those reports were suffering from and the inconsistencies those reports are suffering from, mainly around geographic data, it gave a significant pause on really looking to TransUnion as a source of truth for their kind of insights. Again, we're not saying the reports are bad. Not saying all their reports are bad. We're not saying all their reports are bad. We're not saying the quarterly report is as bad as the monthly and has problems. But we are pointing out that it can be difficult to just interpret TransUnion data solely because ultimately that is very much revolved around data that has to be reported to the bureaus. It has to be normalized and therefore kind of regurgitated back to see clear performance trends. There may be some cases where a borrower goes delinquent and that delinquency is not reported to the bureaus because maybe the borrower is in a workout situation. Or in a modification situation. That can happen. How often does that happen? We don't know. And clarify, I don't even know if TransUnion knows. But I'll give you an illustration of how pernicious this trend can be. Their monthly insights report posted, you know, post delinquency trends by state across products, across autos, mortgages, bank cards, and consumer unsecured loans. Black Knight with its or ICE with its excellent mortgage monitor also posts delinquency trends by state for mortgages. And theirs is based on their robust kind of loan level data set, which covers something like 90 something percent of all mortgages in the U.S. And that data set has some significant dispersion and delinquency behavior by state. On the high end, you see Georgia, which has kind of seen about three, four percent higher delinquencies on a year over year basis versus the low end Maine is seeing something like a 12 or 13 percent decline on a year over year basis. Those same two states in the TransUnion. Data have the exact same year over year change in mortgage logancies all up something like one base or one percent, all exactly the same. And if you look at the dispersion at the on mortgage level, there is almost no dispersion in terms of delinquency increases by state. It's almost constantly like one to two percent across almost every state. And some states are seeing that kind of behavior is also true on their auto data, on their credit card data. And a lot of that has to do with the fact that they frequently revise their prior year estimates. In the report. So that gives us some real concerns about using the TransUnion report. Nevertheless, in this case, it does seem to be kind of accurate where you are seeing some weakening in performance in Q3, Q4, 24 among the prime universe. But in some cases, that is driven by prepayments. If you compare it to 2023 and in terms of comparing to 2022, it's really not comfortable. The performance of late 24 vintages are still much, much better in prime, near prime, any kind of great, great segment compared to 2022 be a next question. This is very related to the prior one is whether elevated prepayment rates are impacting investor returns and factoring into pricing. Absolutely is absolutely impacting investor returns. Why? Because when you're getting back, just as an illustration, in some cases, you're getting back money within a very, very rapid period. As an example here, let's use the 24 Q, 2024 Q1 vintage. That vintage reached 20%. 20% cumulative prepayments prepayments. by month 13, meaning about a year out, 20% of the loans you originally bought would have prepaid. And that's about six months faster than historical trends would suggest. This divergence, coupled with the fact that some of the recent loans have seen very faster amortization schedules, means the best performing borrowers are either amortizing down or leaving the pool early instead of remaining current. That is obviously going to decrease the amount of overall interest you're expecting to see, and that's also going to create a denominator effect where if the best borrowers are leaving the pool, you're leaving more and more of the worst, which therefore increases delinquencies and charge-up rates. It doesn't ultimately increase or decrease that segment, the denominator effect, doesn't decrease returns for investors, but it does mean that you're getting less interest from the best loans that you've bought. And nothing illustrates that better than the return of capital question. So using that same vintage, 20% of the interest you're expecting to see, and that's also going to create a denominator effect for Q1, that vintage would have returned 50% of initial investment capital by month 13. Meaning by month 13, through all the interest you received, all the prepayments, all the amortization, you get back 50% of your initial investment by month 13. That's not quite the six months earlier than historical trends. That's more like two, three months because amortization isn't as variable, but that is still a significantly faster pace. However, ROI is not the metric that most investors will use for calculating their own returns because of the limitations of the fact that it's based on original bets. And because of the factor of time-weighting things, and because of the factor of, again, prepayment rates. Instead, what investors frequently use is ROI. That is a bit of a more bespoke calculation. That is something that DVO1 has done for clients. So if you are a client and you're interested, more prospective client, please reach out to us. We have a formulaic solution to do this. But when you're thinking about it, you're going to have to think about it. You're going to have to think about it. That ROI, that faster prepayment is going to kind of wash itself out because ultimately ROI is not based on original balance and will instead be adjusted based on time-weighted of cash flows. The last two questions we want to address both actually touch upon the same thing, which is the private credit. And the questions are whether the private credit entrants in this space make the market more opaque for DVO1 or more opaque for investors in general. Pretty definitive answer here. No, it does not. That's your authorities first. In our data sets, we work, especially in the consumer and secured space, we work with all the major fintech lenders who provide us not just their securitized data, but their entire platform data set to DVO1. That covers every single loan that they've ever originated or that they've originated recently. Some of them had changed their program several years ago where maybe the program was making a certain category of loans, which are definitely not being made anymore. They readjusted the program and some of those loans may be excluded, but we even have data sets that do include those loans. So very much the case that we have a platform data set that covers all loans, not just the ones that get securitized, but the ones that go into pass-through securities and the ones that go into private credit. There are some exceptions here. There are an occasional set of investors or an occasional part of the issuer platform that they don't put in that platform data set because of a specific request from a client. That does happen, but it's very much one-off. It's not loans that are done through the private credit channel ultimately find their way into the platform data sets anyway. But more importantly, private credit gets a lot of attention recently, but we've got to let go of the notion that private credit is a recent phenomenon. It's not. Consumer unsecured space, the fintech lending, has been around since 2006. That's when the loans were first starting to be made. And as an institutional asset class, it emerged in 2012. The first securitization in consumer unsecured was 2015. And because of the kind of sector's self-immolation in 2015, 2016, there really wasn't much securitization activity until 2017 and 2018. It's a six-year gap between when institutional loans were starting to be made and securitizations picked back up. Were no loans made at that time? Absolutely not. Lots of loans were made at that time. Lots of loans were sold to investors, which ultimately were either held on a balance sheet or went into warehouses or etc. A huge amount of that segment was private credit. Huge. So private credit is not new. It doesn't make the market more opaque. It doesn't alter the dynamics. It's just one another funding channel in the sector. It's just another form of capital. It's been here for forever. And there's no reason to say that it wouldn't stay. Nor that it kind of prejudices certain investor classes or excludes certain others or makes the market less transparent. No, it's a funding channel. It's a set of investors, just like any other set of investors in this space. So. So with that, we conclude our questions from the webinar and are going to shift to a recap of performance and originations. Here, we'll focus on two sectors, just to consumer unsecured and non-QA. As I mentioned, auto we covered in a previous podcast. Since our last update on kind of consumer trends, which is about six months ago, trends have largely been fairly simple with a couple of notable sections, which we'll get into specifically in the data. Starting with consumer unsecured, this is obviously one of the sectors that DVL1 is best known for. This market has continued to perform very well in most months. In fact, over the past two years, in 23 out of the past 24 months, the market has beaten seasonally adjusted trends. Seasonal adjustment is very, very important in consumer unsecured because there is absolutely a pattern of delinquencies because this is a market that trends on trades on day count. If a month has 30 days versus 31, that has some expected behavior. Certain months also have expected behavior like months around the holiday shopping season. In the aftermath, performance tends to be kind of weaker in the fourth quarter of the year, and it always improves in the month of March. March is the best performance month of the year. So you have to look at everything on a seasonally adjusted basis. And in that regard, as I mentioned, 23 of the past 24 months, performance has beaten seasonally adjusted trends. And in fact, it was 23 out of 23 prior to October 25. October 25 was the first month since late 23 when the sector didn't beat seasonal trends. In fact, now prior to. Prior to October, the magnitude of improvement when you compare to how much better performance was in seasonally adjusted had started to slow down. In Q1 25, we were talking about just record setting improvements. In some cases, the improvement was like two, three, four times better than you would have expected on a seasonal basis. That slowed to maybe one and a half, in some cases, 1.2 times better. But the largest reason there is the fact that so much improvement had already occurred. You were already looking at improvement on improvement. You've already seen delinquency rates or impairment rates go down more than 200 basis points in aggregate. Charge-offs are down kind of as a similar degree. We're already talking about charge-offs back to pre-COVID levels at a seasonally adjusted basis. So you're not really talking about a market that has a humongous amount of ability to keep improving by these magnitudes on the seasonal basis because the only time performance had really gotten that much better was in 2021, which of course was at 50%. So that was a very different timer for consumers. That said, it's not all sunshine and rainbows in the sector. We have been talking about some performance deterioration, which we talked about earlier in late '24 vintages, and we've also covered that because one of our charts shows performance by loan age, and we've constantly pointed out that impairments on loan ages three to six months, which represent the most recent vintages, have been consistently climbing over the course of the past year. Now, let's touch on October. October was the first month in the past year. It was the second month in the past two years where consumer unsecured underperformed relative to seasonal trends. We suspect a very large part of this was driven by the government shutdown, and we have empirical data for this. We actually covered it in our past report, which showed that a government shutdown is not as easy to track in performance as something like a hurricane, which is very much localized. We constantly write in our performance reports, certainly did a lot of that in 2024, where certain months we just simply don't talk about seasonal trends, and we say, "Hey, impairments are higher because of a hurricane." We don't talk about seasonal trends, we just say, "Hey, it's because of a hurricane or natural disaster." That was true in the early '25 for California for wildfires, and of course in late '24 in Florida and regions around the Gulf because of Hurricane Selina and Milton. In government shutdown, that's a little bit more difficult to confer. Now, we did use a proxy. We looked at the Maryland, Virginia, and D.C. metro areas, which of course will have more federal employees than other areas, as a proxy of the impact of the government shutdown. It wasn't an impact. We saw about a 60 to 70 basis point pickup in impairments relative to what you would expect in those areas. And there's positive news. By December, we already have data through early December, we've seen all of that impairment increase corrected. In fact, it's actually overcorrected, which we'll cover in our upcoming performance report. But yeah, we're already seeing impairments in those areas back to, in fact, even a little bit lower relative to national average than they're expected to be historically. So, it seems very much, as with natural disasters, very much a one-off situation where issuers were very proactive, stepping up, providing modifications. So, largely, it was a non-event. Going back previously, the way that we think about performance, of course, is ROI and return. We touched upon the different vintages, vintage behavior before. But what we wanted to bring into this report is one of the newer segments in our publications, which is looking at ROI differences, not just by vintage, but by vintage and attribute that is easily comparable across all issuers. Really that boils down to two attributes, FICO and income. We don't use grade here, mainly because. We've constantly spotlighted the grade as, of course, the individually most important attribute in differentiating credit performance across issuers, but it's also subjective. Not only does each issuer have its own grading system, but their internal grading system can change over time, which is why DV01's benchmark does periodically review our kind of high-level categories of top, middle, and bottom grades to see if those grades are still warranted. But things that don't really change are income and FICO. FICO score is, generally speaking, similar across issuers as is income. These are numeric figures. They're not subject to change. Somebody can't have a $200K income borrower and say, yeah, this borrower's actually only earning $50K. So that is something that we can observe over time, and we can observe the behavior and the ROI kind of behavior across those attributes, which we spotlight. In all of our performance reports, the main things that have changed is there's a fundamental reshape of the dynamic of FICO and its impact on returns over the past three years, and a much more subtle but nevertheless important change in behavior across income. So FICO, in a period of low performance volatility, or generally kind of strong performance, you expect to see higher FICO loans deliver lower ROIs relative to lower FICO. I mean, returns on lower FICO. FICO loans tend to be higher, and that's commensurate with higher risk, because the performance volatility tends to be a lot higher. You've got a lot more delinquencies. You've got a lot more charge-offs. And kind of selecting the wrong vintage long issuer or the wrong grade can really kind of punish investors in a good time, even a good time. And in a bad time, this is also the segment that tends to underperform. That's exactly what we saw in FICO. So historically, pre-COVID, lower FICO constantly delivered higher ROI. Completely false. During the 22 and 2023, with higher FICO across the board delivering higher returns. And that pendulum has swung all the way back, where, once again, we're seeing lower FICO scores delivering higher ROIs. In fact, the 2022-4 second-half vintages, those earlier aforementioned Q3, Q4s, are seeing the widest difference in ROI between lower FICO and higher FICO than any vintage has been historically. That's really very true. It's not to say that this will definitively stay true going forward, but at least so far, no vintage has seen a kind of bigger shift to lower FICO outperformance than that particular vintage. Income, on the other hand, has been a completely different story. Simply put, since 2021, investors have been paid to be in higher income loans. Paid, period. The higher up in income you've gone, the higher ROI has been. And that's been especially true, actually, at the highest two income buckets. The highest we can get. Categorized for statistical significance as borrowers with above 200K incomes versus borrowers. The next category is 135K to 200K. And consistently, the above 200K category outperforms 135. And going further and further down, higher incomes outperform on an ROI basis relative to lower. There's been a mild correction in the 24 vintages compared to 23 behavior, but very mild. These vintages are still showing that higher incomes are outperforming. Just not. Not by the same magnitude. This seems like it should be kind of counterintuitive because buyer FICO borrowers tend to also have higher incomes, but it's really not. And the reason is that FICO is the largest driver of pricing differentiation after grade. And again, grade we talked about is subjective. And also grade is highly influenced by FICO. And we'll talk about the difference in pricing in a moment. Our income is not. Income is not a pricing variable. So the GWAC differences across income ranges have largely converged over the course of the past three years, especially relative to pre-COVID levels. So shifting gears into originations rather than performance, again, comes sticking to consumer unsecured and something that we didn't cover in our prior report. 2025 has been a very, very strong year for origination growth in the consumer unsecured space. Prior to October, the sector was registering year over year growth rates of 30 plus percent and some months as high as 50 to 60 percent over the course of 2025. And that was true in each of the prior 12 months. Moreover, June and July of 2020. We're the only months in the sector's history where monthly origination volumes topped $5 billion. Since those peaks, originations have slowed and they've been below $5 billion every single month since then. But they should be kept in context. Prior to 2025, the record origination month was $4.5 billion. And that was in May 2022. Every single month since May 2025 has eclipsed that figure. So while it's not crossing the $5 billion threshold, each of the months is still above that $4.5 billion. More of the origination growth is coming at a time when the overall growth in credit card balances has slowed considerably, slowed to its slowest growth rate in years. And if you account for new account openings, credit card growth may actually be downright negative, meaning that most of the growth now is being driven by new and new credit card accounts rather than existing borrowers drawing more on their balances. Now, what should give investors a lot of comfort is that even though origination volumes are growing, credit standards remain the same. As a proxy of this, top-grade originations have continuously made up about 50-55% of total originations over the past two and a half years. It has never deviated out of that space in that time period. At the high end, it's maybe 55-56%, low end, 50-51%, always in that range. Historically, that range is around 40-45%, pre-COVID to today, 50-55%. With that, we've also had a general upward shift in FICO and a somewhat upward shift in income, but incomes have been relatively flat over the past year. The more interesting behavior, and because you weren't secured, it's actually around, this comes in two flavors, both of which are related. First, as we talked about, the most important variable in pricing is grade. Grade, of course, FICO is a very important component of grade, but grade itself is more important. In that regard, top grades have seen by far the largest decline in growth wax over the past 18 months. Top-grade GWACs, after peaking in Q2 2023, are now more than 100 bps and actually more than 110 bps below those peaks. Middle grades are also about 100 bps below, but that's much less important on a relative basis because GWACs on top grades are close to 10% and middle grades are close to 16-17%. So, obviously, a 110 basis point move from 11-10 is more important than a 100 basis point from 17.5-16. But the more interesting behavior is how GWAC has interacted across the two variables we discussed earlier. FICO and income. Across FICO, we've seen consistent widening in GWAC differences between ranges, especially over the past three. This is especially true when you look at pricing differences at the extreme range. The extreme being above 760 FICO versus below 670. That we call the extreme range where you have a significant amount of loans, enough where it's statistically significant, but also the higher end of production. In our origination reports, we spelt light that this is the highest end of production this difference is wide. This difference is more than 1,000 basis points or more than 10%. And the difference in 2025 has risen substantially and is now at the highest end that it's ever been. In fact, back to the highs that we saw pre-COVID. And the market was quite a bit different than pre-COVID, so perhaps the comp here isn't necessarily as important. But nevertheless, very, very wide difference. Certainly wider than anything we've seen over the prior three years. But that's not the only area that's wider. We've seen a similar widening. In the next tier down, meaning 720 to 760 versus 670 to 695 FICO. Same thing, widening. And actually, in that case, it's actually wider than pre-COVID loans. The only area where the difference, the GYF difference is not, or is still tighter than pre-COVID and hasn't risen quite as much, is in that top, top tier, which is 760 plus versus 720 to 760. And we suspect the answer there is, the answer overall here is we're seeing a lot of growth and we're seeing a market that is increasingly competing for the top borrower. And they're increasingly competing there by offering price, lowering their coupons, and really trying to attract the top, the best quality borrowers. There isn't the same competition for the lower end and for that middle grade. But when you're talking about the borrowers that are looking for high single digits, low double digits, APRs, this is probably a prime borrower that's refinancing their credit card. That's where people are really, really competing. Now, conversely, when you look at GWAC differences across incomes, as we talked about earlier, higher income, you've been just paid to have higher income because they've outperformed. I think the largest reason for that is the fact that GWAC differences across incomes have narrowed substantially. For every single income range, every single one, the GWAC difference in pricing is materially lower than pre-COVID lows. In some cases, the most pernicious examples here being 200K plus income versus 135 to 85K. That we call the first, we call the top two, and then the second we call like the second highest versus the second lowest. Those differences in GWAC are less than half of what they were prior to COVID. Less than half. Again, the reason here is that income is not a variable that is used in loan pricing. I think that's actually a requirement. I don't believe that is. I think that might be a requirement for originations. I don't know if that's a deliberate choice or a requirement, but that is the fact. It's not a variable that is. highly used in pricing, it's actually a variable that is used more in verification. So in that regard, income has not, the GWAC differences in income have narrowed. And of course, that's led to higher ROIs and a higher, we're running up on time. So the next sector we'll cover is non-QM, and that will take quite a bit less time. The non-QM mortgage market has, on the exact opposite track of consumer unsecured, has seen its worst two years in the sector's history. Now, the sector is not particularly old. It's less than a decade old in size, but nevertheless, it has been a very, very tough Alongside FHA, it has exhibited significant performance pressures, even as all other segments of the mortgage market, namely the GSEs and portfolio loans, are still seeing delinquency rates that are near record lows. Not really the case for FHA, certainly not the case for non-QM. In fact, if you look at these two categories combined, they represent the majority of the increase in delinquencies of the overall mortgage market over the past two years, something close to about 60 or 70%. The delinquency started rising. In non-QM, in the summer of 2022, it's been continuing to rise over the past three years, and 2025 has been more of the same story. Continued upwards, the pressure on delinquencies is mainly driven by weaker attributes. So think lower FICO, in non-QM, that means below 700, and higher LTV, really, than above 80 LTV category. But within that, I think the most important variable is vintage. All of the delinquency increases in the non-QM space have come from the 23 and 2024 vintages. There is a little bit of a positive momentum, there's a little bit of improvement. In the late 2024 vintage versus 2023, but nevertheless, these are significantly underperforming vintages, especially compared to 2021 and 2022. But the other factor that matters here is loan purpose. Loan purpose is always a factor in mortgage performance, because historically, purchased loans tend to be weaker than refiled loans. Why? Because in a lot of cases, it's a first-time homebuyer, and in some cases, you just don't know if it's a first-time homebuyer or not. Versus when you're looking at refinancing, even if it's a cash-out loan. You are talking about a borrower that's already had a loan on their house. So they've been in the house for a period of time, and they've already had at least one loan, which they've obviously refinanced. When you isolate performance to just purchase loans, which again is the vast majority of production for 2023 and 2024, delinquency trends look much, much less ominous. They look, the 23, 24 vintages still underperform, but by a much narrower margin. This is true for not just impairments and delinquencies. But especially for cure rates. On the other hand, the second largest portion of production in 23 and 24 has been cash-out refis, and those have underperformed materially. Not just in impairments, but especially in cure rates and curing behavior, where curing behavior is just materially worse on charge-offs. And that's in some cases to be expected, because you are talking about a borrower that took cash out of their home to, you know, consolidate debt, do an improvement, who knows. But this doesn't matter, you know, for 90. Something percent of cash-out loans, this is not really a big deal, but for the borrowers that do go delinquent, it may be the case that they're, you know, they were in financial troubles, and they return to financial troubles, and they're finding it much harder to cure than purchase loans. But the most important thing that we can consider for non-QM, and which continues to be true for the whole mortgage market, is that higher delinquencies have yet to translate to material losses. It is still the main, or still remains the case that not a single rated non-QM bond has ever taken. A dollar of loss. Not a single bond. Not a single B-rated one. Not the triple Bs. No segment. No issuer. Losses on the DVO1 non-QM benchmark can be measured in the single-digit basis points. And across the entire market, all non-QM loans ever originated that were securitized. And though we don't really know the private side, but in non-QM, securitization is a very substantial portion of all volume. The total number of loans that have liquidated with loss over the past decade is less than 700, you know. So food. Food for thought. That number will go up with the higher delinquency rate you're expecting it to. But on the other hand, non-QM is a fairly low LTV product. The average LTV is in the 60s. Home price adjusted LTV is going to be in the 50s and 40s. So you're not really seeing a lot of borrowers that are taking losses. Because of that, the primary focus for investors remain prepayments. And also, and actually in some cases, call activity. Starting with the call activity first, non-QM as a segment. As the market has seen, actually saw a significant rewrite of docs compared to, let's say, pre-GFC securitizations where calls were typically at 10% or in some cases even 5% of collateral. And non-QM, that number is actually 30%. So calls happen a lot earlier. There's also incentives for the issuers to call because a lot of the bonds in the coupon stack have an escalator at year five where the interest rate on those bonds steps up substantially. And we have seen a huge amount of call activity in 2025. Largely stemming from some of the securitizations done in 2022, materially wider spreads, as those have come due because of the significant tightening in funding spreads on AAAs and because of the pay down nature of the collateral, most likely the AAAs are gone from those deals and now you're in the single AAA stack. And of course, refinancing that collateral is fairly favorable today. So we've seen a lot. As for prepayments, of course, they've been responding. They've been responding to lower rates, but ultimately what investors care about is S-curves. And DVL1 is one of the only places where you can actually look at dynamic behavior on S-curves. We can look at prepayments on both our benchmark, on all matters of securities, on a security versus benchmark basis by loan rate incentive over a period of time, whatever you define. And you can do that right in ARL. And what we've seen, especially in October, we saw a significant prepayment surge in October that was driven by substantial S-curve steepening of the 50 BIFs plus rate incentive. The S-curve shifted parallel up, in some cases up as much as 1, 2, or 3 CPR. In some cases, actually even higher. And a lot of that was concentrated in 12-month bank statements. Prepayments on 12-month bank statement loans specifically increased something like almost 8 or 9 CPR in October. And with that, we conclude today's episode of the In the Charges of Structural Finance podcast. I know we've thrown a lot of information at you guys today. This is more of a kind of technical episode for investors that do follow this market. But if you are curious about any of this data or anything that we've covered, all of our research remains free. It will remain free. Continue to do so. And all you need to do to access it is reach out to our sales team, reach out to our marketing team. And if you have any questions about the content that you've heard here, again, reach out to those same channels and we'll be sure to respond. For the interim, this has been another episode of the In the Charges of Structured Finance podcast. I'm your host, Vadim Verhoglad. And once again, please, if you're listening, leave us a rating and review on your platform. We would love to get our ratings up. It gets our podcast out to more investors. It'll give us more impetus to produce content and kind of significantly increase our footprint. It takes two to three seconds and helps us out tremendously. So with that, I bid you guys adieu. I don't think we'll have another episode in December. So wishing you guys a happy holidays and we'll be back in January with a lot more content. There's a lot of things to discuss based on some of the research we've been recently publishing. With that, thank you. And until next time, take care.
Speaker 1Thank you for listening to DV01's In the Trunches of Structured Finance podcast. Stay up to date on the latest trends by subscribing to our podcast and our market research. If you have any questions about this episode or want to learn how DV01 data can help you understand what's going on in the market, contact us at sales at dv01.co. You can also learn more about DV01 products and market research by visiting our website at www.dv01.com. zero one dot co.
PodcastTranscript AI