Leaders in Lending
Leaders in Lending

Episode · 9 months ago

How Alternative Credit Data is Driving Financial Inclusion w/ Greg Wright

ABOUT THIS EPISODE

Only 82% of the US population can be scored using conventional credit scores. That unscorable 18% is overwhelmingly comprised of marginalized communities—and that extreme disparity is unacceptable in this day and age. 

However, things are changing. Alternative data sources are opening the door for lenders to offer credit products to marginalized communities in ways that do not create more risk. Lenders have tools at their disposal to drive financial inclusion today—not to mention boost ROI. 

In this episode, Greg Wright, EVP and Chief Product Officer at Experian, shares some of the approaches for broadening the scorable population while increasing lender accuracy. 

We discuss: 

- How new products are driving improvements in credit scorability 

- The consumer permission model in credit data 

- The regulatory landscape around credit data sources 

- Better scoring models built with AI and ML 

To hear more from Leaders in Lending, check us out on Apple Podcasts, Spotify, or on our website. 

Listening on a desktop & can’t see the links? Just search for Leaders in Lending on your favorite podcast player.

How we score, what data we use. The analytics were applying definitely opens up doors that we've never had before or on how we can drive financial inclusion for everyone in the US economy. You're listening to leaders and lending from upstart, a podcast dedicated to helping consumer lenders grow their programs and improve their product offerings. Each week here, decision makers in the finance industry offer insights into the future of the lending industry best practices around digital transformation. In one let's get into the show. Welcome to leaders and lending. I'm your host, Jeff Keltner. This week's episode features my conversation with Greg Right. Greg is the chief product officer for the Experience Credit Bureau and we really dive into a couple of topics, particularly his perspectives around the things that are changing to enhance credit underwriting capability, specifically the use of user permission data, the use of extended FCIRA data and then the use of AI and ml fan standalytics capabilities to make sense of all those data points are saying. I think one really interesting point is the financial inclusion lens that Greg approaches us from and how we can say yes to more people particularly people from traditionally disadvantaged communities, while at the same time keep haying our risk the same or lowering it for for lenders. So or really interesting discussion about some of the trends with the regulatory perspectives on these shifts in the way credit underwriting is done. And to get this directly from, I think, one of the bureaus is a really interesting perspective and something every lender needs to be thinking about how these trends will be impacting their business. So I hope you enjoy this conversation with Greg. Right, Greg, welcome to the podcast and thanks so much for joining us today. I appreciate your make of the time. Yeah, Jeff, happy to be here. You know, I've really been looking forward to this conversation, given how much time and energy we've spent thinking about credit data, credit modeling, and wanted to start off with just asking you, like, what are the recent trends you're seeing in the way people are approaching the question of leveraging credit data for the underwriting of credit risk? It feels like kind of an interesting time in that space to me. And then I also later want to kind of dive into like alternative data and some of the newer things, but just at a high level. What are the big trends? You're saying? Yeah, well, I think this is a really great opportunity, just given where we are in the market, the kind of craziness we had with through the pandemic, with the economy. That really has shown, I think, where credit has worked and also where maybe there's opportunity and, you know, an experience. We want to be the consumers bureau and so we're really serious about addressing the opportunity around financial inclusion and I think exactly what you're talking about, how we score, what data we use, the analytics were applying definitely opens up doors that we've never had before or on how we can drive financial inclusion for every everyone in the US economy, and so I'm excited about talking about a lot of different ways and a lot of different approaches to help drive that. So what are you seeing people do? I think the financial inclusion angle is really interesting because you know our perspective, and I think you share this, is that more sophisticated techniques, more data, can actually increase the world that we can lend to, whether that's for scoring UN score able or or better understanding risk to approve people who were on a unapprovable previously. What are the kind of what are the trends you're seeing? And I'm kind of curious. It seems to me like some of these capabilities have been around for a while, but be the trended data or, you know, much more data in the credit file than just like a credit score, one score or another score that learners have tried to rely on. So you know, what are you seeing people do differently and why do you think this wasn't done somewhat earlier? What helped people back from moving towards a more sophisticated approach to underwriting than maybe as being leverage today? Yeah, well, I think that's a great question. Let me maybe take a step back and look. You know, who is in the lending market today is very different than who was in the lending market, say, five years ago, ten years ago, and I think a lot of that is what's driving the change as well. So you think about traditional large financial institutions, they have the scoring models they have, they can have fairly sophisticated analytics, but at the end of the day they're very comfortable using conventional credit scores and their models, as well as a lot about other attributes. Then you look at credit unions and regional banks, again,...

...less sophistication in their credit models, less sophistication in the analytics and even more reliance on conventional credit scores. And then you have new entrance, whether it's Fintech from a few years ago or now even by now pay later players, or large tech players like apple or Amazon, who's coming in and looking at lending in a very different way, and I be pushing the market in good ways, of US thinking about how we leverage scores, different analytics, different approaches, and so a factor of different lenders coming in and looking at it all fresh where sometimes even thinking about credit as like that's the old way of doing it. Let's think about the new way of doing it, and I think that's really refreshing and you know, as experience, we are always looking forward to how we can help drive financial inclusion with a lot of different types of data sources. If you look at let's just make the case why. If you look at conventional credit scores today, they can score about eighty two percent of the US population, a bunch of that is because people are crediting visible. They don't have a credit report in the bureau today. Some of them also have a credit report but are unscorable. So they may have a thin file, which means they only have one or two trade lines, maybe three at Max, in their credit report, or they have a stale history. They're not active with their credit. Trade lines are their credit vehicles today, and so the conventional scores are just they don't have enough data to score them. Now I think some progress has been made. You know, the credit bureaus, working with lenders and working with a company called vantage, came out with a new score. It looked at the traditional credit history that you would have on your credit report. It also looked at what you already talked about briefly, trended data, which is how you behave with your credit data over time and how you maybe are changing your utilization, maybe you're leveraging up or leveraging down, or the type of average balances you have over time, and that can give more information about who you are as a consumer and your ability to take on more credit will responsibly. And they also applied some advanced analytics, so AI and machine learning techniques to build that vantage score model, especially when they went to vantage for now. That took the scoable population of eighty two percent on conventional scores up to eighty nine percent for the US population. Big Improvement but frankly not good enough. And when you start looking into black communities, marginalized communities, let next Brown communities, those scores go down like the scoreble. Populations reduced and I think we're at a point in time when that's just unacceptable. Like we have to do a better job. We have to as a credit bureau, working with our lenders and looking at how we serve our communities, driving financial inclusion for all types of credit products into those marginalized communities as well. To be able to say yes to more people on in a responsible way that does not create more risk. Is something you can do right now. It's not something that is in the future or something we can do some day in the future. I think we can do it right now and we should be doing it. We've seen how you can improve upon conventional scores moving in the vantage and we've even gone beyond vantage into a new parct we just launch called experience lift. I'm not really here for a product pitch. But what's important about lift is we can now scored ninety six percent of the US population. So there's just a small slice. Where's that? I know lift and you've got a product boost, I think as well. I don't want to we don't think in the products. I think you know what is the core of what those things are doing her providing or how are they augmenting the scorebile population. I mean like just help people understand like what's living that improvement and score ability. Great question and maybe I'll take them in turn. There's really three different things. That I think is the future of how we drive more financial inclusion, more SCO ability and, frankly, just showing the full picture of a consumers financial profile and their financial potential. So first of all, and you mentioned experience boost, this is a unique product. We launched on Experiencom wherefore, free you can go and you can add your bank statement, read only data, and we will then pull in off of that your positive built payment history, so paying your utilities, your water bill, your electric bill, even things like your Internet and your cell...

...phone and even as far as your streaming services. All those we identify, we show it back to the consumer. They verify, yeah, that's my water bill, that's my that's my mobile cell phone bill. We show them the data they added directly to their consumer credit report instantly, re rescoret and show it back to them on their mobile phone and says, Hey, here's your improvement in your phyco score because you permission more data to your credit report directly. I think the consumer permission model is a way for consumers to engage directly in their credit history add more positive data. A lot of consumers think, hey, what do you mean? My water bill and electric bill and cell phone bill are on my credit report. I'm paying what I owe. That should be reflected and a lot of people think it already is reflected in the reality is, in the most part it's not, and when it is, it's usually because you failed to pay and we got a negative collections notice and that's what shows up on your credit report. This is a way for consumers to engage directly with their credit report, add their data instantly to their credit report and get the credit for paying their bills on time. I think that's a no brainer and just something we should be doing. The facto going forward. You go to Experiencom right now you can go do it in a couple minutes. But that's been able to allow, I think, now over ten million consumers at additional data to their credit report and we've boosted phyco scores by over fifty million points an aggregate. That just shows the extent to which you can actually, as a individual, have an impact on your own credit report and your own credit score, which, you know, has always been the promise of something like going and doing credit education and engaging with their credit score and eventually, if you do the right things, it will go up. Well, this is something you can do right now that will instantly impact your score most of the time. Yeah, it's for lender that you're doing this as a as a CIRA because, I mean it is it's something a lender can do with the technical integrations, the reading of the data. And then, of course, if I'm using that data to credit decision, my sources of data need to be cris rather have to have to get it from the users or have to find some CIRA source will provide it. So you guys, minderstanding in this context or kind of serving as a CRI filter so I can get at least the impacts of that data on a credit score through experience and not have to deal with the CR applications, like trying to figure out how to deal with utility companies and phone bill companies and streaming services. Right how at that? That's and even better because it goes directly on your credit report and the existing scores in the market, whether it's conventional scores like Phicho or its vantage or other the automatically accept that data, so you instantly get that appreciation of your in the score, positive bill payment history. But the best part is you're still the consumer. You're in the drive receipt. This is your data. You permissioned it to your credit report. At any time you can take it off, you can disconnect your bank account, you can decide not to continue to furnish certain trade lines or or bill payment history. You're in the drive receipt. They're completely in control, which is great. The two other areas that are less around direct consumer engagement but ways that as a credit bureau, as also working with our financial institutions and clients, we can do a lot more to dry financial inclusion. The second category of that is really looking at extended FCIRA data. So fcira is the umbrella and which we all operate. To do credit decisioning. We have to have data sources that are approved and kind of, you know, abide by, you know, the displayable disputable we can show it and clearly demonstrate to the consumer where the data came from and they can tell if we got it wrong and they can go through the process. So that's all under the FCIRA guidelines. But using more data that goes beyond the core traditional credit report. We already mentioned trended data is being one of the ways to use data that has not traditionally been used for credit scoring. But we can go a lot further than that and that's where, you know, something like experience lift comes in. You know, we bought a specialty credit bureau called clarity services, which is in the alternative financial services space. Think paid a lending rent to own short term installment loans. We combined our traditional financial credit bureau with trended data plus clarity plus a positive public records data source from Lexis nexis called risk of you...

...five Oh, we combine all those together and by doing that again, that's where we went from not just eighty two percent scoreable or with vantage, eighty nine percent score but all the way up to ninety six percent scoable. So now we can score basically any credit active inbound inquiry a lender might have. But on top of that it's more predictive because we're using a broader data set. Not only can you say yes to more people, but you're you're actually not taking on more risk, and so we work with clients and they can actually dial back. You want to say yes more? Do you want to take on slightly more risks or slightly less risk, and you can actually dial that in any way that you want to do that while driving financial inclusion and taking on less risk for your portfolio. But what really makes all that work is the third category, which is advanced analytics, applying AI and machine learning models and techniques to how you build a credit score and attributes across multiple data sources. It'd be much harder to leverage three, four, five, six seven data sources if you weren't also applying these advanced analytical techniques, which is how we built experience lift also how we bill be kind of what we call explainable AI. It's taking the model to its optimum threshold and then dialing it back so we can actually have display explainable disperate or sorry, not disperate impact, but adverse actions, and having the codes that all work so that we can all have a behave like a normal credit score with adverse actions and then have a fall under FCRA. And again I think we're you know, five years ago it was kind of scary to talk about using alternative data sources and using AI and ml, but I think today this has become kind of standard course and if you're not doing it, you're missing out and you're actually leaving both money on the table and not you're choosing not to serve communities that you could be serving right right now. So I think those are the kind of three categories, consumer permission, data, extended FCIRA, going beyond the traditional credit report and then leveraging advanced analytics to use all of that. Well, you can't. I mean I can only imagine what the score card would look like if you were using a traditional approach with all those sources. You're very large, very large table with a lot of dimension. That's right. What ask you're taking me? You mentioned that if you're not the you can be leaving money on the table if you're not doing this today, and I think that's I think is actually worse than that, because you're increasingly being not not only leaving money in the table, but what's left for you is being adversely selected by people who are out there doing this and finding in any given credit you know, and you can score population that one letter might see the people who are better than that m a credit as point of view, and they're making better offers. And so I think the lenders who don't do this will actually see the deterioration in their formants because they're going to start to be adversely selected by smarter lenders. But I am curious, I would the point of disputes. I don't think it's all that common, at least among the financial institutions I talked to. They're aware of it, but whether this is really in production at scale most fies. I question. And I think the big reason is a lot of concern or uncertainty around the regulatory environment. And nobody's gotten in trouble for these things. At the same time, you know, they haven't been around for fifty years with nobody having been in trouble. So there's this question of like, you know, are the regulators going to be okay? Are they going to come after this, after they go after big tech for their ranking algorithms and whatever else like? How do I get comfortable that regulators are going to be okay with these approaches? I'm curious your perspective on that. Xkind know you have a lot of engagement the regulators, as we do on this topic. I Love Your Perspective on where you think they're at and what you see coming in a space that I'm not sure it's so much that people have heard negative things for the regulators. They're just are the positive things either necessarily, and so that lack of clarity, I think often for financial stations just means, Hey, let's wait until there's like a more clear answer coming out of my regulatory body. Yeah, and maybe let me take a couple angles on that one, because I think it's a really critical topic. You know, from the Consumer Permission Data Angle, I think we've heard nothing but positive from the CFBB and other regulator saying you should be doing this. And when you look at extend or when you talk about alternative data, they've also talked about you should be leveraging utility data and rental data and other positive built payment history. That doesn't make it to the credit report.

Today, you can go, you know, go do a Google search right now on CFPB and alternative lending and you'll see all of these. You know, they talk about it. They want lenders to be considering all the other data sources that they could be leveraging. That's supposed to beauty of boost is it combines the two of those things together. So super permission data to get you access to your utility payments, your your TV and Internet and other services, as well as even in the streaming. So we're enabling that for consumers because we can't get the big utility players and the mobile carriers actually provide that data to us directly. And Be Nice of the right latter stepped in and say hey, you should also furnish that data directly to us, because that would be a lot easier. But you know, we're making it happen. I don't think there's going to be any negative repercussions on that from a regulatory standpoint. In fact, we're driving financial inclusion, which I think aligns very much with the current administrations focus. And then on the AIML which is that third leg of it, the kind of advanced analytics. Again, a lot of work has gone into doing this in a way that is explainable, that is clear to the financial institutions as well as regulators. Of Documentation is incredible. The ability to actually to create models and then automate the documentation that goes into the details of that. An example that we did with experience lift, which again leverages traditional data plus the extend of SCIRA data plus advanced analytics the IML. We actually did a disparate impact study on that as well, leveraging the the approaches at the CEFP be recommended. We actually went to a third party law firm to get it certified that we did it all the right way, and that's actually now the first question we get from some of our largest clients of like hey, I don't know if I can use this, I don't know if it's going to, you know, meet the requirements. I'm like, look, we've already done all the heavy lifting and and most scores actually don't go through that kind of disparate impact study and we found, you know, is actually had less disperate impact than any other score we could test it against. And so, you know, that's the type of approach. I think that is going to create more comfort in the industry where you're, you know, being very data transparent, you're being very clear about the uses of the data and in many cases we're only using positive data to create a more complete picture. So with boost, we only show positiveill payment history because we actually don't know if you missed a payment or not because we don't have the bill. With riskview five point, no, we only took positive public records data to add to that lift score. So again, where we can have positive data and create a more full picture of the consumer. Again I don't think that's going to run a foul of anything the regulators want. In fact, I think it's a line with what they are driving for, which is driving financial inclusion. Yeah, I obviously we agree and I also the interesting thing you mentioned that I think is often lost in these discussions is usally talk about disparate impact and shows. You know, your lifting power scores were swaring less disperate impact than in the other score you test, and I think that's often the thing we forget, is that the system is, that it has existed for however many years, is not as inclusive as reality would have it be. You, where we to really perfectly understand risk, we would be more inclusive in our lending, and there's always fear about will ai run a foul of fair lending, but I think we sometimes miss the opportunity for these techniques actually improve dramatically financial inclusion because the status quo is it nearly where we like it to be. And I think that's the you know, the point that's easy to miss in this conversation but so critical, and I think that your work transparencies key with that highlight that, hey, we can make things we can. We can only like make you more accurate, not make things worse. We can make things better or protect the US this for just say your for traditionally disadvantage communities. That's a pretty powerful statement. Yeah, and the Aniso analytics are really clear. You can say yes to more people because of the analytics and the extended data sets you're using, and this is also something this kind of hard to understand as well. Sometimes introducing negative data means that you can say yes to more people. So let's just walk through this for a second. Like if I have a population at a certain credit score where they have a ten percent default, that means nine percent of them won't default and one will. If I can bring a data source in that helps me identify the goods in the bads better than the one that's going to...

...default, I can find and say no, no, this is not a good product for you. The other nine I can say yes to. And so whenever I can merge both positive data sources as well as some negative data sources, you're actually in a better position to say yes to more people and reduce the risk of your portfolio. And you know, some might say, well, you're saying, you're you're bringing in more data just to, you know, knock people out of your lending product. That's not fair. I'm like, actually, no, no, it's about seeing yes to nine people by identifying the one where this product is not right for them. That's right. Even, I always use this example, but even like in a subprime, super risky portfolio to a bank, you maybe it's twenty percent losses, which sounds really scary, but means you're saying no to eight good people because you couldn't find the two that were that were risk because it twenty percent of exactly right, eighty percent repayment. It's so you know, in my mind, when we're operating in that kind of world, just any increase. You can have a predictive power pays tremendous evidence because we're declining so many would be good borrower. That's right, or overpricing them where we say hey, we'll let you in, but just barely, and you're going to pay the maximumy go out. We really really honest to the world, better we would. We put you in a very low risk category and that's that's a burden that, frankly, not only the decline borrowers pay, but every approved borrower who pays anything over than, you know, the cost of funds for a financial station is paying a premium because of ours, an industry, lack of understanding of risk, and that's a it's a high premium American consumer and it's it's something we can improve on. A hundred percent agree. And Look, if we want to even take smaller steps to start, like we don't want to reimagine the entire system from the from the starting point. I think every lender should consider this. Okay, you have your normal scoring model. First of all. First benchmarket against you know, either vantage or against experience. Lift. If I can scor ninety six percent, you can always roar any two percent. I feel pretty confident I'm going to do a better job. Yeah, but but secondly, maybe create a second chance. You know, Opportunity. You have an apple cation process. You're going to decline them. Just do a little bit of work, leverage a second score and maybe you can say yes to them in a second chance or even better, a second chance plus you. Maybe you said Yes to them. But your point, I barely let you in the door and you're going to get my top rate. Maybe, if I have more information, I could do a improve your rate step where I use a new score, I use additional data like verification of income and or employment, and the consumer goes through the extra burden of doing that because the promise is you're going to say no and now you can say yes, and so I have a lot of incentive to go through an extra step. Or you gave me this interest right, but maybe you can improve it. As a consumer, I'm going to do jump through a lot of hurdles to get a better price, especially if I'm buying a car or if I'm trying to get a personal line of credit because I want to remodel my home. A slight interest, you know, drop, could be really material dollars and not only you're driving financial inclusion but even a better financial outcome for those of you are saying yes to yeah, that's a the other thing I've seen for second looks that I think is really interesting is, you know, the lenders we tend to work with all over time look at the at the population was just outside of their approval box and go huh, the stuff at the bottom of my approval box is doing really well. Right, and these these alternate scorings are like they said they would be good and like they turned out to be really good. Man, maybe I should like and so there becomes a you know, for lenders you don't have to jump in all the way of saying, Hey, let's forget the old thing, let's just do the created the new thing. They don't pluck, but you let's take the marginal and this the marginal performs. You Go, you know, that becomes a normal and you go, Hey, let's take the next round of marginal and see if these alternative data sources, all you know right works, can help us find the good eight or nine barrowers out of that population and bring them in. And I think my experience is lenders, they get there over time it's a process to say hey, we're watching it, it's getting good, it looks better. It's great to start with something like a second look or, you know, program or an extra consumer stop to get into the program because I think you'll find that to be a one way road tours being more inclusive over time. Yeah, and you know, you talked about at some point now, maybe in the future, if everyone's using these scores and you're not, then you're going to be actually at a big disadvantage. I don't think we're quite there yet, honestly, but I like that vision.

But what I will say is I think, you know, the fintext and the big tech players are more aggressive it looking at these types of data sources, these types of alternative scores, like lifts, like they'll lift attributes, like leveraging, you know, the boost data that we have on the credit report. They're way more open minded than, I think, a lot of traditional financial institutions, and so I think they are going to lead the way and that point it will change the benchmark, the baseline of what is good enough for your credit decisioning models. And I think the other thing that is an opportunity. Whereas my largest financial situations. They want, you know, all the the granular P is, the raw data or the data attributes of these alternativeata sources so they can build their own custom models, which were totally down with, a hundred percent supportive of that. But with something like experience lift, we can take that to a credit union or a regional bank who maybe doesn't have the resources to build their own credit models and again, whatever they have in market right now, I would love to just show them we can beat their metric, we can beat them in a head to head both on who they can say yes to and the risk they're putting onto their balance sheet. So, you know, it's an easy win on that front. I think we're just at the starting mind. Frankly, yeah, I think it's a great point. You made it earlier about highlighted here, which is as you get to higher accuracy and you're doing a comparison with a lift or a boost or a trend advantage or whatever, the alternative scoring technique or approach, maybe you have this really interesting dial that's like, do I want to keep my approvals where they're at and get more people, you know, and lower my risk, or do I want to keep my losses where they're at and increase my book, or do I want to shift my lot? But you you end up with much more control. It's not just bringing more risk into a book or more uncertain it's actually adding to certain day and you go, yeah, I want to I want to reduce my risk, we can do that. I want to keep my risk the same, it just double my population. Can do that too. So it's a really interesting choice that they get to make when you really can think about how do I want to leverage that a better predictive power to change my strategy in terms of approvals versus risk. Yeah, and given the current environment, you know, I actually start often with the financial inclusion as my first lead in of like you can actually help serve the communities you serve today more and you can do it with a better Roi. Now a lot of them they want to talk about their Roi and how how do I make more money with this, but I think more and more financial inclusion is a driver, is a part of their strategy, and so if you can do both, I think that's really powerful. Yeah, and in the evidence is pretty clear and if you just look at distributions of credit scores among protected classes. It's clear that whenever you increase predictive power, you're going to disproportionately help people who have traditionally been left out of the financial system, and that means you're traditionally helping, you know, disadvantaged communities, communities to color, protected status individuals. So it's they tend to go hand in hand just because that's those are the populations that have historically not been well served by, you know, the more traditional approaches on. So the increase and accuracy helps you. It helps you serve those communities. That's a huge not just a good thing to do for the world, good to do for business too, but it's the right thing to do and I think that's a it's pretty powerful motivator for many, many bankers. Yeah, you know, interestingly, I think historically we've tried to not look at kind of race proxy flags or even understand like who's being served, who's being scored, who're not being sport because, you know, a fear of redlining or other things. In fact, I think I see it different trend now, which is we've actually built flags that can be used in analytical environment to how you build your scores, leveraging the cfpb guidelines on how you do that, and more and more we're being asked, show me how my scores perform against different populations and different marginalized communities. Obviously that none of that data is in a production environment, because then you do run the risk of read lining, but I think more and more people want to know how does this perform, rather than kind of sweep it under the rug, and so I think we're just in a different environment. People want to know. They want to know how their scores are performing, how they're underwriting is performing, and can they do a better job? And the answers yes. Right now the answer is yes, almost everybody. You can do a better job and there's ways to do...

...it, whether it's consumer prission data or alternaive data assets and we can bring to bear, or leveraging aimml or doing all three together in the best case. Yeah, it a combination of those, I think is powerful, though, to your point, I think the heiml analytics is almost required to take advantage in a real way of either the user permissioned or the extended FCI a kind of data sources, because is as soon as the data sources multiply and the data points multiplied than like, you know, basic linear regressions and such are not gonna, like not going to give you the power of what that data really has to offer until you can really combine the ministry ways. Yeah, the one, the one climent I would just make the fun part about boosts. The way we did that, which is a consumer engagement model. They have the data to the credit report. You know, to do anything as a lender, you automatically get the data and it's part of your scores, part of your model, it's part of Phyco or advantage. But I agree with you as you extend beyond that, because there's a lot of other consumer permission data sources. I'm interested in that I want to go after. You know, we talked about some pretty basic ways to extend your data beyond the core credit report. We talked about Trendi data. We talked about alterni financial services, be euros. We also have another bureau called the Rent Bureau where we get rental trade lines. Those are kind of easy. I think there's a lot of data that goes beyond that that we're looking at from a consumer permission model. You know, one example we have public records data that comes from risk few from Lexis nexus and basically they're going out collecting data is publicly available on websites, including things like are you a lawyer or an accountant, or plumber or contractor? So these are public licenses that they can get access to. There's ways to do consumer permission data where you could actually go directly to a state and get every single license they have on file and let a consumer claim that they are an accountant or they are a contractor and then add that to their profile. Like we're not quite there yet, but that kind of promise of I bring all of my identity components to the credit decision that are relevant because I can prove I if I know you're an accountant or a lawyer or a plumber or a nurse or a hair salon and you have a certification of that, you're a better credit risk period. I mean there's more data about you that we can make that credit decision on. So enabling consumers to kind of bring all of that to bears so they have a complete, full profile of who they are, can only improve their credit decisioning. Get better rates, say get you know, get get to a yes more often than a know. And for those who weren't even scoreable now to become scoreable. And so that's the promise for me of going even beyond with consumer permission data. That's really fascinating. That gets those kind of sources a data and if you can do it in a way that doesn't add too much friction to the lending process or to your earlier point, if you can do it in a second look way where I'm only adding that friction in, I would have been no scenario. So I'm not stopping the people where I already have good data. I think it's a really powerful model and I think the technologies are only going to go to making that easier to do right. Those are going to give become more accessible data sources, easier to put into the process and take advantage of. Yeah, and that's something every lenders to be thinking about. And I've been working with consumer permission data for many, many years before experience. I was an into it where there's a lot of products that rely upon consumer permission data. The thing I've learned over all those years if you have the right value proposition, consumers are happy to contribute more data. You know, if you're just asking to give data over, then they're going to be, you know, more private, more like what you need this for? Why? But if there's very clear value on the other side of the table, getting to guess for an auto loan or a better rate for my mortgage or, you know, being able to increase my Faculo score instantly, those are value propositions people say yes to and they're there. It's easy and to your point, you don't want to put everybody through that flow, but you threw a second chance flow again. You can get to a yes or you can get to a better rate, and I think the consumer is more than happy to spend a little more time providing some more data to get to that answer. That outcome for them. Excell and then I have one of the questions I did not prep you for but I'm always been curious about, which is, you know, one of the things I think up start in getting into the unsecured persons of the kind of very different kind kind of...

...loan than was traditionally given out by financial stuss very different than what most credit scores were kind of targeted at predicting risk for. One of the things that strikes me is the absurdity, maybe is too strong a word, but the strange nature of using a singular score for products that are very different. Right, like I was able to just our range is like five year Onezero loan or a three year fiftyzero lone. Like you're not like a seven something or a six something for both of those. You could be an eight hundred and fifty four one and five hundred for the other, because there are very different obligations. I'm curious your thoughts on that or what you see people doing to adjust, because it feels like, well, obviously you can say this is a risk your product, we're going to have a different, you know, credit score requirement. It feels like there are people that you know are in the middle that are really very good risks for certain products and very bad risks for other products in a singular score doesn't differentiate between them. While and how do you see, if you've seen any trends on how lenders can or should think about that when they're introducing maybe new kinds of products that are not, you know, well captured by traditional scoring approaches? Yeah, great question. Look, I think consumers for many years were under the false perception that there's one credit score and I think you know more and more people are more aware that there are more than one of them are a score and even with Phicho they have industry scores, they have different versions of their scores. So there is somewhat of a proliferation, maybe even somewhat targeted, versions of different scores. But honestly they're not that well adopted. I mean some who are very narrow use those. So I think the opportunity, Jeff, is absolutely still there. I think we're we're again at a bit of an inflection point. We talked about AIML as a new technique to building better scoring models leveraging broader data sets. We're also entering into a new world, and this is a bit of a buzz word, so I'm just going to throw it out there. ML OPS in automation, right, so machine learning operations is a new approach and we're, you know, on the forefront of that. You know, our ascend data platform enables us to build new aim l driven models in a much faster way. So even before as send the data platform we had, if you go to a traditional like polar data set, have a flat, you know, File Archive, and then you have your data scientist of crunch it for three four months and then it's ready for doing analytics. And then they're going to build analytics for three or four months and then they're going to go through governance and QUEA, testing and and documentation for another three, four five months and then maybe eighteen months later you have a new score. With data platform, we've crunched the middle part of that too. Near real time. I mean you can be building aiml models in hours and days rather than months, two years. We're all now. We're also now automating the kind of data wrangling, the what day did you need? How do you pull the data together so you can actually build the attributes then allow you to build the analytics you want to go do. And on the back end we're crunching that side of it through automation as well. So Automated Governance, automated deployment, automated documentation generation, and so you can think about going from building a new model and kind of nine to eighteen months to days or weeks and and again. We built lift. It took me eighteen months to build lifts, from the data scientists working on it to getting it into the credit report channels, to getting into batch environments to getting in the real time environments eighteen months and we probably, as a credit bureau, create two or three new mm are aiml models for at a production level, at a product level, every year. I have the promise of be able to do twenty, thirty, forty of those a year. And you know, we do maybe twenty custom models for clients every year. We could do hundreds of them in a year based on the new technology that we're leveraging. And this is you know, there's other companies out there, you probably could name them if you wanted to, or doing automated aiml capabilities, but as a credit bureau, we need to be on the forefront of that, enabling our clients so that if you're a credit union and you want to adopt a new model, you can...

...do it in a matter of days or weeks. In fact, I may even walk into your door with a new model that you can benchmark against that I prebuilt for you. And so that's the kind of promise of what were the new world we're entering. And you know I have lift, you know I've experience lift. I plan to have two, three, four, five six versions of that in a year and kind of micro segment them down to a prime auto, and how about prime mortgage? And how about, you know, fintech personal lending, and how about by now, pay later scores like I think we're entering in a world where you do want to do that, because different data sources will actually perform better in different environments. And also the consumer experience you have to worry about. Of what data source are you bringing in and does that require consumer engagement or not? While on a mortgage, maybe that's fine. On A, you know, five hundred dollar personal lending loan, absolutely not. So I think we're going to enter in a world where there's a lot of flexibility on that front and the scoring will not be a bottleneck like it is today, the the score development, course development. I think it's a really interesting point and the thing that strikes me about all of the the trends you've mentioned, from, you know, user remission data to extended fcia data, to the enhanced analytics through Mlai, to things like ml ops, is they're they're all like turning the same way in a flywheel. They're all feed on each other and enable them. It's like, well, MLA is great, but it takes a teen months. Will now actually, the technology is coming world'll take eighteen days instead eighteen months. And well, all this date is great, but I can't take advantage of well, the techniques are here now to take advantage of it and to do it quickly, and so I think as these things all come together, I think that's why, at least, I'm so excited about the space is it just feels like all the pieces are coming into place where there's a dramatically different and better way to approach this question. And now, finally, like all these trends come together to make it really possible, not just in theory but in practice, and for financial institutions to really be doing this in the very near term, which is which makes it, I think, a really exciting time to be in this space. It is, and I wouldn't even say in the near term. Like, if you are thinking about how you did credit decision the last five years, you're going to miss out in the next five and you're going to miss out both from serving your community through driving financial inclusion. You're going to miss out by saying no to people you should have been saying yes to. You're going to miss out on money on the table, and so you know, and it's not some time in the future when the regulators say it's okay. The answer is it's right now, right in front of us. We have the solutions, we have the data, we have the attributes, we have the scores, and so the answer is go now and and we're not going to be slowing down. This is going to be speeding up and the benefit is going to accrue to the entire economy, into our consumers as well as our financial to institutions. And I agree. It's a super exciting time. Excell a lot seems like a pretty good, pretty good place in the conversation. Both of US excited. I do have three questions. I asked all my guests at the end. saw a kind of rappey hire, so I'll find them to you now and then I'll say goodbye to the questions. A number one, what's the best piece of career advice you've ever done? You know, I think the best piece of career advice I've ever gotten is there are no rules to your career. As actually talking to any executive coach at one point and I was like, you know, I don't know, it's either go do this next thing or maybe go and do a start up, but I think if I take a couple more years, I won't even be an option anymore. And he POSI's like, don't create false rules around your career. You can go do whatever you want whenever you want. As long as you understand what you are, what your goals are and what values and beliefs you bring. You can go decide to do whatever you want with your career at any point. No, no rules around your car like that is very unique piece of advice, but I really like that. All right. My second question, what's the best piece of advice you've gotten about the consumer lending space, the Sumer Lending Spaith? Best piece of advice, I think it might have been advice I gave myself, which is you don't have to do it the way it's always been done. You know, when we were building boost, there was a lot of times or I felt like we ended up at a dead end. This wasn't going to work. It was too hard, the analytics weren't making sense. How do we get consumer permission data through compliance and and the regulatory process and and Multiple Times I had to sit down with the team and just say look, we got to find a way through this positive bill payment history is got to be good for consumers and good for lenders. There's got to be a way through,...

...and that's what innovations all about, is being resilient and being knowing your strategy, your vision, and sticking to it and finding a way through. Doesn't you have to do it the way it's always been done? That is so true. I think sometimes the people with the most experience have the hardest time seeing that. There the people who know so ingrained how it's always been done that sometimes you can't. Can't see the forest for the trees the way you've always seeing it done. My last last question is what is one bold prediction for the future so I can drag you back here in a year and tell you were right or tell you were wrong and we can compare notes. Okay, a question I get a lot is is by now pay later? For Real? Well, it's the guy like, is it just a layaway program with new you know, new liftic, or is it a changing consumer behavior? My prediction is it's not a it's not a really hard one, but I've heard people say the honner. It is a new consumer behavior. It is a new shopping experience. It's here to stay. It's going to grow. It's going to grow a lot. And if you think it's just something good experience, like, what do you think is the new part that is that is here to stick? Because I've heard it a couple different takes on what makes is very different than a layaway or a credit card. I mean it's some ways. You if, like by paid in a credit card and paid it off in three months, like it's very similar, maybe a little bit higher costs and credit card. Like. What do you think are the parts that are really new and sticky about the the binopulator space? Well, there's a couple that are very obvious to the consumer and there's a couple that are behind the scenes. I'd say that to the consumer it's in check out. It's super simple. I can, you know, apply and get it in the same time and it's, you know, being used to pay off, not some future credit that I'm going to maybe use tomorrow, but just the thing I'm buying right now. And so it's very controllable and understandable and I'm going to make for payments and very clear, like what I'm signing up for, and so it doesn't feel scary and I think a lot of people have been burned by credit in the past or their own behavior using credit or personal loans. This is very contained, if you will, and I think it's easy, it's simple and it's clear of what you're signing up for. I think behind the scenes some of the differences is the way they're paid. They're not paid by people making payments late or by, you know, having interest rate fees to the consumer. It's all through merchant fees and then so they're kind of paying for their losses through the merchant fees. The merchants have benefit because they're they have a higher conversion rate and they're selling more per basket and that's more than offsets the losses the by now pay laters are seeing. So it's kind of a win win. That happens behind the scenes. That doesn't even impact the consumers and so they have a ongoing benefit to make their consumers happy without having to rely on them for fees. And I think there's a lot of pieces to all of that consumer experience and the merchant experience in the lending experience that are positive. That is a bit of a positive fly will that I think will go well. There may be cases where people use it irresponsibly. They get to ahead of themselves on too many by now, pay later loans and and that will happen with any credit product, and so that doesn't scare me. I don't think that's like, Oh my God, this is just a negative spiral or sending consumers off into in fact, of any lending products I've seen recently launched, this one's very clear and very controllable by the consumer. So I've actually pretty bullish, all right, bullish on buy now, pay later. Just got to get the buy now, pay later into your one of your extended FCIRA or user permission data sets. So the not they tamed on that, I said. But yeah, I will keep my eyes peeled for that. And thanks for a great conversation like this was a lot of fun and I appreciate your joining. Yeah, thanks, Jeff. That was great. Up Start Partners with banks and credit unions to help grow their consumer loan port folios and deliver a modern, all digital lending experience. As the average consumer becomes more digitally savvy, it only makes sense that their bank does too. Up Starts AI landing platform uses sophisticated machine learning models to more accurately identify risk and approve more applicants than traditional credit models, with fraud rates near zero. Upstarts all digital experience, reduces manual processing for banks and offers a simple and convenient experience for consumers. Whether you're looking to grow and enhance your existing personal and auto lenning programs or you're just getting started, upstart can help. Upstart offers an into ind solution that can help...

...you find more credit worthy borrowers within your risk profile, with all digital underwriting, onboarding, loan closing and servicing. It's all possible with upstart in your corner. Learn more about finding new borrowers, enhancing your credit decisioning process and growing your business by visiting UPSTARTCOM Ford Banks. That's upstartcom Ford Banks. You've been listening to leaders and lending from upstart. Make sure you never miss an episode. Subscribe to leaders and lending in your favorite podcast player using apple podcast. Leave as a quick rating by tapping the number of stars you think the show deserves. Thanks for listening. Until next time. The views and opinions expressed by the host and guests on the leaders and lending podcast are their own and their participation in this podcast does not imply an endorsement of such views by their organization or themselves. The content provided is for informational purposes only and the discussion between the host and guests should not be taken as financial advice by companies or individuals.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (78)