- 3 months ago
From Austin, Texas, HousingWire’s Allison LaForgia sat down with a panel of mortgage and credit leaders to unpack the rapidly evolving landscape of credit modernization, touching on everything from trigger leads and tri-merge credit to VantageScore adoption and the role AI may ultimately play in reshaping the industry.
Joining the conversation were Andrew Davidson, Jennifer McGuiness-Lubbert, Gregory Sher, Robert Zimmer and Shelley Leonard.
The discussion began with trigger leads and how lenders are responding to recent regulatory changes. Gregory Sher, Managing Director at NFM Lending, said it is “a little early to know exactly the impacts,” adding that lenders are now “focusing more on front of the funnel to make sure that if consumers are contacted by the servicers there, we are top of mind.”
Joining the conversation were Andrew Davidson, Jennifer McGuiness-Lubbert, Gregory Sher, Robert Zimmer and Shelley Leonard.
The discussion began with trigger leads and how lenders are responding to recent regulatory changes. Gregory Sher, Managing Director at NFM Lending, said it is “a little early to know exactly the impacts,” adding that lenders are now “focusing more on front of the funnel to make sure that if consumers are contacted by the servicers there, we are top of mind.”
Category
📚
LearningTranscript
00:06From Austin, Texas, I'm Allison LaForgia, and I am sitting down with a panel to go over all
00:12things credit modernization. We have Andy Davidson, we have Jennifer McGinnis, we have Greg Scherr,
00:17and we have Robert Simmer, and last but very much not least, my co-captain for this conversation,
00:23Shelly Leonard. Shelly. Hi. Thank you for joining me. We're going to start with trigger leads.
00:30Have there been large changes in the industry post-regulation rollout, and how have things
00:37changed from a strategy perspective? Greg, let's start with you. A little early to know exactly
00:44the impacts. As a lender, we're focusing more on front of the funnel to make sure that if
00:52consumers are contacted by the servicers, we are top of mind, and there are certain strategies
00:57behind that that we can and have implemented. But I think it's too early to tell just what the
01:04impact is going to be. For us as an organization, it's going to be a loss unless we mitigate it
01:08through these measures that I'm talking about. The other thing is there still are some people
01:13violating the code of conduct rules, and I'm getting lots of word from our loan officers
01:19that borrowers are being contacted by various lenders that somehow get a hold of their data,
01:25even though they're not one of the two or three entities that's allowed to have the trigger data.
01:30I don't know how that's happening, but it's happening.
01:32So have you changed your upfront pull strategy and moved away from soft pulls, or you're still
01:37using them?
01:38No, we're still using soft pulls.
01:39Okay.
01:39We think it's smart to take the borrower as deep into the process as we can and then pull
01:45what triggers the alert there at the end when it's almost too late to do anything about it.
01:50So we're still doing that.
01:51Sure. I think that we've heard a lot of similar things from the lenders that work with Zactus
01:58that they're in a wait and see mode is what I would say. They haven't really made changes
02:05to their strategy yet. They're still hearing that consumers are getting a lot of calls and
02:10they're concerned, confused about who's policing the new legislation.
02:16So similar feedback. We really haven't seen a shift away from soft pulls at all. We did hear
02:21there's some people that are starting to consider it, but I think that there's also the consumer
02:26impact that we have to think about, right? Consumers know that pulling your credit does
02:31affect your score. And so consumers are smarter about their credit these days. And so I think
02:37consumers are also potentially pushing back on, no, I don't want you to do a hard pull.
02:42I want to talk about tri-merge versus single bureau approaches for a second. Very specifically,
02:51what does the data tell us about tri-merge versus single bureau approach?
02:58Andy?
02:59Some insight on that. So we did a study that we actually got data from all three credit bureaus
03:05of the Vantage score. And by using Vantage score, there's no difference in the model between the
03:11three scores. And what we found is that about 25 to 30% of the borrowers have exactly the same
03:18score
03:20from all three bureaus. And think, oh, that's good. It's not going to be too much risk.
03:24Well, on the flip side, about one and a half percent of borrowers actually have score differences
03:29of over 100 points. Right? You think, oh, it's most of the same that it would be like a normal
03:36bell-shaped distribution, but it's not. There's a lot of risk out on the edge. And then if you look
03:41sort of more closely or sort of more aggregate, in most cases, if you pulled a single score and then
03:48you looked at the median that you would have gotten if you had pulled the median on that, about 10
03:53% of
03:53the time, the median falls into the next 20-point higher bucket. And about 10% of it falls into
03:59the
03:59bucket that's 20 points lower. And so the real question is, like, who's taking the risk on that
04:05difference? Is it the borrower is not getting the right loan and they're paying too much? Or is it
04:10the investor taking on more risk than they were hoping to take on if they had had the correct
04:15information? So dealing with both this sort of very wide dispersion and just sort of the middle level,
04:21you know, misstatement of risk makes going to one bureau very difficult relative to the tri-moge.
04:29Yeah. And I mean, I think, you know, the fact that, you know, credit reporting by creditors and
04:35lenders in general is voluntary is part of the problem. And, you know, you've even seen a pullback
04:41by certain banks and credit card companies of reporting it all now that there are certain
04:46things that could be compensated to report. So I think that's important as well. I also think,
04:52you know, to, you know, Andy's point, the two different models work differently. So for example,
04:57FICO does have a model per credit bureau, whereas Vantage uses one model, pulls all of the trade
05:03lines out and then puts back what they consider in their algorithm. But that put back is not
05:09necessarily everything that's on each of the bureau's reports, because there are certain
05:13things that are excluded in their model. And also the application of the points are very different
05:22between the two models and where the hits lie. You know, one of those big ones is always credit
05:28utilization. Now with the newer models, you know, having a trended data, I think is going to be a
05:35big deal for borrowers on that. Single versus tri-merge, don't undo it when you're also talking
05:41about introducing new scoring models. Leave well enough alone. We don't need the extra friction points.
05:46Um, and, uh, like I said, make credit reporting mandatory. And then we could talk about what we
05:52could do there. Yeah. I think that, um, in our analysis of the data that we have across about
05:5730% of credit polls in the industry, um, I think we have to remember it's not just about the
06:04score,
06:05but it's also about the underlying credit data, right. And the trade lines. And when we did our
06:11analysis at Zactus, we found that when we analyze trade line level data, um, that there's 26% of the
06:21consumers that we looked at that had differences across the three bureaus. Um, and so meaning that
06:27those trade lines were not reported on all three bureaus in 26% of the time. That's a lot, right?
06:34That's a lot, um, that you potentially could miss out on if you're just choosing one bureau. Uh, I think
06:41that that is the other really valid point that, that you gave Jen that just really resonates with
06:47me is about how much change can the industry take at one time and truly be able to decipher
06:55where the risk or benefit is coming from. Right. And so I do think if we need to think about
07:01it in a
07:02measured way, not only from an implementation perspective, because there's a lot of changes
07:07that have to take place in the LOS and the POS and the MI companies, you know, in the credit
07:11verification providers, you know, in the lenders, they're in their downstream operational reporting
07:16and credit risk models, as well as in the GSEs and all the other lenders and the bondholders and
07:21everyone in the, in the ecosystem, um, to do that, to go with scores first, which seems like
07:28seems to be approaching much quickly, much more quickly, um, makes sense. Then let's see where we
07:34are. Let's see how we're performing. Let's see what the results are before we introduce any
07:39additional change. I think just that measured approach, given it's taken us this long to get
07:45to this point, just makes sense to me. I do think there is sort of an upside even to the
07:50disruption
07:50side, which is, you know, for 30 years, the mortgage market has just viewed the FICO classic score
07:56is just the score without really focusing on what's underneath that. And one, now we have
08:02two scores that actually like score people differently. And then on top of that, this
08:06one bureau versus three bureau is really getting more people to focus on, well, what is a credit
08:11score? And then also like which data is in and which data isn't in the credit scores. And,
08:17uh, so I think that's a positive, um, let's say whether or not it leads to cost savings in
08:21the end, isn't that clear, but I think there's a much greater appreciation as to what borrower
08:27credit data is than there was before all of this controversy. Let's look at recent announcements
08:35from the FHFA and FHA. What do we think those recent announcements mean for the industry concerning
08:41the new models and what industry, what information does the industry need to move forward?
08:48That's easy, but at least the FICO 10T data for the 10 years. Yeah. It's kind of like, um,
08:54so the Vantage data for a 10 year period of time has been out for some time. The FICO 10T
08:59data from
09:002013 to 2023 is complete. According to what Fannie and Freddie have, um, in their, um, pilot information,
09:09they're going to release a supplemental file file for 23 to 25 for Vantage and then 2013 to 2025
09:16for FICO 10T in the summer. My thing is, if you can release a supplemental file for Vantage,
09:23you can do that for FICO 10T. Let's stop wasting time, give the market the data and let's analyze
09:29the two models. That's my view. So I can say that since the announcement last week, we've gotten more
09:35calls from clients about Vantage score than we had for probably like the four years before that
09:42you and me both brother. And, um, you know, and I, so part of the problem is sort of data,
09:48but one of the problems is that the data we need can't exist until the Vantage score is actually used
09:53to originate loans. So we have historical data and the historical data says that, you know, these two
10:00models are different. First order, they're roughly the same second order. They're very, very different.
10:05Very different. Um, and then if you allow lender choice, then we're not really using either model
10:13anymore, right? Cause the higher of two scores is not FICO score. It's not Vantage score. If it's a
10:19higher three scores, 10, 10 T, right? That's not any of those models. And so that means that there is
10:25no actual distribution of borrower behavior associated with those scores. So we estimate
10:32if you just between classic FICO and Vantage score could choose the higher score, that that increases
10:39delinquencies by about 40% at the same score level. And that translates into like a half a point to
10:47a
10:47point in price in terms of either LLPAs or credit risk transfer or other credit valuation.
10:55at that score level. And when you change something like that, that means all of the intuition that
11:01people have developed over the years, no longer really works. And so investors tend to back away
11:07from markets where they're not sure what's going on. So, you know, we need more data.
11:13Andy makes the most salient point. You make a great point about not having the most important data
11:17that's needed. How do these loans perform once they have a Vantage score? And that's all the more
11:23reason why you've got to be pragmatic and dip your toe in here. You can't go all the way in
11:29because
11:29if it goes sideways, just the littlest bit, Rob, you can speak to this. No one's going to want to
11:35pay
11:36up for these loans. It's going to change the complete dynamic of the market. Don't you think?
11:39I do. My candid sense, though, is debate and discussion are always good. But politically
11:46speaking, again, Singapore is not going anywhere. And until there's substantial industry consensus,
11:53which is probably going to take years, there's just no way this can possibly move.
11:58So I enjoy the conversation. But when you go to Capitol Hill, they're like, now,
12:02when you talk to people in the agencies, they say we have our hands full. I don't want to spoil
12:10the
12:10party. But I have to say that on the concern of will we be able to restrain costs in general
12:17for
12:17mortgage credit scores, the CHLA public position is we're asking the FHFA director to direct the GSEs
12:25to examine the viability of creating their own credit score subsidiaries. And I think that exercise
12:31alone would also teach us a lot. And it doesn't mean, oh, you're going to form them, but ask them
12:38to assess the viability. And the answer might come back from the GSEs, we may not need the credit score
12:44the way we thought we did. So we just want to hear more about this. And we want to hear
12:48about it from
12:48the GSEs. It's not a foregone conclusion, but let's at least examine this. They have the data.
12:54They have plenty of smart people. I realize that's not happening anytime soon, but it is a way to look
12:58ahead and see if we can all be smarter. Shelly, I want to ask you what's going on with FICO
13:02Direct
13:02and what should lenders be paying attention to? So FICO Direct is progressing, albeit slowly.
13:11There are a number of verification providers such as Zactus who are in build, test and validate mode
13:17with FICO. And so, you know, it's slowly moving. We're dependent, obviously, on the bureaus to provide
13:26test data and updated agreements to allow providers like Zactus to calculate the score on the data that
13:35we get from the bureaus. So we've got to go through some of that legal work as well. But it's
13:39progressing.
13:40So we're getting closer. I think as close as a couple of us, the providers are, we potentially
13:48could see it at the end of second quarter, beginning of third quarter. Then the next question
13:55becomes adoption, right? What are lenders thinking from an adoption perspective between the direct
14:01model, which is purchasing through someone like the credit score itself, still reliant on the credit
14:06data from the bureaus. But then we would run the same FICO algorithms. And as Jen said, again,
14:12separate algorithms by bureau, but the same data is going in the same models that the bureaus run,
14:18that same algorithms is exactly will be running the direct model. The output is the same. So there
14:24should be no downstream impacts to all the integration. So we think that adoption won't be
14:29held up by technology. But the question is, which model is right for a lender, right? So I would
14:36encourage lenders to talk to their provider to help them analyze different scenarios, different ways
14:43that they pull credit, different ways they can optimize their credit pulls, and look at that in
14:49both the direct and indirect model to determine what's best for them, so that they're prepared to
14:53determine, do they want to adopt? Do they want to consider adopting? How could it benefit their ultimate
14:58consumers as well as their internal businesses?
15:01Now, you mentioned one piece in there that I'm going to apply to this whole conversation is that
15:07progress is slow moving. And we've talked about many different changes and different positions on
15:14different changes. But what does this type of modernization change mean for the entire loan
15:22cycle? You just wrote an article about this for HousingWire.
15:26Let's start with you. And let's start with what the change means for the whole long cycle.
15:31Well, I mean, I think first and foremost, it's if the change ever truly gets effectuated
15:36through the full cycle. So, you know, you know, you want to think about it as, you know,
15:42borrower to broker slash lender, lender to warehouse lender, warehouse lender to aggregator or GSE,
15:48they then have their own warehouse lender. If you're the GSE, you're then, you know,
15:54issuing you MBS. If you are the aggregator, you're issuing generally RMBS. Once we get there,
16:00you know, this isn't just a lender pricing thing. The whole loan aggregator or the GSE
16:05is controlling that LLPA to buy these loans. I will tell you, I had a call the day that this
16:11announcement came out with 35 aggregators. Nobody has had advantage to their seller guide yet.
16:16And that was on agency and non-agency. It'll be interesting to see which aggregators decide to
16:24allow vantage first. I think that's going to be the first gate to get passed for them.
16:29Will the aggregators allow it? And then away from that, once you get passed there,
16:34the question is, what's the bond buyer going to demand for it? I think Andy said it very well.
16:39You don't have 30 years of data on performance, on these loans, et cetera. You know,
16:44how is the bond buyer going to feel about that? They do need to run two separate models.
16:48We slightly disagree respectfully on if the investor really understands the credit score
16:54and the credit trade line data. As an aggregator, and I've also been an issuer of securities for 25
16:59years of my career, we know what it is. We know how it moves. We have, you know, models to
17:05actually
17:05look at it now regressively and project it forward to understand it. And that's actually how we price
17:12loans. That's how we price bonds, right? So I think that's important. You know, in the initial
17:18whole loan trades, we've seen, we've seen 50 to 60 basis points have hit for the vantage only pools.
17:25We have not, I know New Res said that they securitized $10 million worth of loans
17:30into a securitization. But I think what nobody's paid attention to is seller guides were not changed
17:37back then. So they may have priced those loans off vantage. In order to put them in the UMBS,
17:41they had to run a FICO score. And nobody's talking about that. And I think that's important
17:46as well. And, you know, $10 million worth of loans versus a $2 trillion market is not going
17:53to get whole loan buyers and bond buyers, you know, on that train quickly.
18:00Me, I like to be progressive. We have done a ton of work on all three of the scoring models
18:06and what
18:06the differences are. In theory, we could price to every one of them if we wanted to. But I still
18:12want the FICO 10T data to check even more loans. But at the end of the day, it's all about
18:17getting
18:17the work done and getting all the different counterparties to agree. And that's not just
18:22lenders. That's not just aggregators. And that's not just bond buyers. It's warehouse lenders. It's
18:26bond repo providers. It's the rating agencies and how they're going to handle the structuring and
18:31subordination for bonds. It's if the CRT guys are going to want to buy risk share deals using,
18:37you know, scores that they haven't had significant data for. It is a full life cycle impact.
18:43100% agree.
18:45Yeah, I think that summarizes pretty well. But, you know, I think over time, the markets will
18:51adjust. To me, like the example I've given to people in the past is that Freddie Mac used to be
18:58the
18:58most liquid TBA. But they had a delay, a longer delay, so that their securities traded at a lower
19:05price. So Freddie Mac created gold. And the idea was to shorten it so that they would price higher
19:11than the Fannie Mae securities because they priced at a lower yield. During that transition,
19:17there weren't enough Freddie Mac gold or new, old or gold. And so Fannie Mae TBAs actually became more
19:26liquid. And that then thus dominated from there forward. And then the Fannie Mae dominated. And so
19:34that was something no one really predicted at the time. And then it took the creation of the UMBS
19:39to bring them back in line. And I just think we're subject to these sort of unknown effects when you
19:45introduce change to a market.
19:47Rob, this one's for you first. While we all try to parse through the noise surrounding this
19:52conversation, what else might influence the whole credit conversation?
19:59We're going to see how the performance stacks up, especially as discussed in the MBS investor
20:04community. The aggregators are potentially an issue if they don't move. It would force more of our members
20:10to go direct, which is not necessarily a good thing because, again, more choices are better than fewer.
20:16If I had to make a prediction now, single pull and buy merge are not viable for the reasons we
20:22discussed.
20:22We're going to be spending a lot of time watching and measuring how VantageScore does, how 10T does.
20:29I'm often asked, like, oh, how fast will VantageScore? Because everyone knows CHLE is for it.
20:33We're very excited about last week. But I'm often asked, well, how quickly will the market
20:37adopt it? And I said, it's going to be slow. Not in a political way, but in an operational way.
20:43The
20:44mortgage industry is tremendously conservative and change takes a lot of time. So I had someone
20:49tell me that last week wasn't a big deal because nothing will change in the short run. And I'm like,
20:53but that's always the mortgage market and it's probably a good thing. So I think we'll start
20:59learning a lot more in about 12 months time. That's my prediction.
21:06You're going to be being more effective.
21:07I'm sure you'll watch that pretty quick.
Comments