Leaders in Lending
Leaders in Lending

Episode · 4 months ago

Lending During a Global Pandemic: Customers Bank’s Road to Success w/ Sam Sidhu

ABOUT THIS EPISODE

Customers Bank, with a revolutionary crossover between FinTech and traditional banking, is truly disrupting and innovating past the financial industry as we know it. One year later, we look back and ask, how did AI-enabled underwriting models perform in a time of uncertainty?


Sam Sidhu, Vice Chairman and Chief Operating Officer of Customers Bank joins us this week to discuss the bank’s enormous success, how they got there, and where they’re headed next.

What we talked about:

- Lending practices and consumer loan performance during the pandemic

- How AI credit decisioning models helped drive strategic growth

- Managing risk appropriately during the pandemic

- Leading with bravery during a period of societal and economic stress


Check out these resources we mentioned during the podcast:

- https://www.linkedin.com/in/ssidhu/

- https://www.customersbank.com/

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 in your favorite podcast player.

You are listening to leaders andlending from upstart a podcast dedicated to helping consumer lenders,grow their programs and improve their product of rings. Each week, herdecision makers in the finance industry offer insights into the future of thelending industry, Best Practices around digital transformation in one let's getinto the show hi and welcome to leaders in lending.I'm your host Jeff Kelter this week, we're featuring my conversation withSAM sidhout customers. Bank SAM is currently the vice chair and ChiefOperating Officer of custer's bank, but it'll become the president and CEeffect of July. First, two thousand and twenty one I've wanted to have SAM jointhe podcast for a while. Now because custer's bank has been at the forefront of two trends, I think are really important and consumer lending the pushinto unsecured lungs, including student loans and the adoption of thin techpartnerships. Sam Discusses why the Bank chose to invest in these areas andhow they've been able to do so effectively launching and managingmultiple fintac partnerships. We also discuss how consumer loans have faredfrom a credit performance point of view. During the PADEMA and duced macroeconomic stress, environment we've been experiencing, and Sam flipped the rollsa bit and asked me a couple questions about upstarts work and ai for creditand topics like accuracy and fairness. It's a fun, interactive conversationthat I think will provide a lot of insights for everyone. Thisconversation was originally produced as a Weben, so the format s a little bitdifferent and we do refer to visuals a few times. You can find a link to thevideo of the session on the show notes if you'd like to see the visuals, andwith that, let's see the conversation Sam. Thank you for joining me. Iappreciate this is, should be a fun conversation, absolutely thanks, Jolooking forward to it yeah. Well, let me just start by saying congratulationson your recent appointment is, I know your vice chair now but, as you know,cel elect of custer's bank. That's really exciting. Thank you! So much it's, it's been whata year to start a career thanking for those who don't know- and I was in theboard of Customers Bank for eight years and then joined last January and end ofJanuary had six weeks in person with my managing team before we went remotetalk about a heck of a start up to a new job out of the frying pan into the fire, soto speak, real bill trod by fire for those of the audience who might be alittle bit less familiar. Can you give me just a little bit of background onPuspire Bank, who you guys are what you do? The bank in general sure so were weare just over a ten year old start up bank that is now grown to eighteenbillion, an assets and the way that we like to think about ourselves and findourselves we're like a super community bank that has the established expertiseof a typical community or regional bank, but with the modern offerings andplatform of a more technology oriented company. So we're trying to design abank of the future, but we're really at the intersection of banking and Centah,but doing so as already one of the top hundred banks in the country. So wehave a little bit of scale. From that perspective, an you guys have beengrowing pretty rapidly. You say, eighteen billion. In my mind, I go backto when we first started talking. I think it was like God, eight or nine,and there was no- you guys were under the ten billion threshold. So congratsyou have any what's been driving some of that growth. I know this was not myquestions list, but I'm curious because it is a kind of remarkable trajectoryof growth. It's not common in the banking industry o sure. No, absolutelywe have community bank in franchise in the northeast. That's a branch, lightformat with relationship based private banking for small video sizedbusinesses. Using the single point of contact team mom. We only have to OLbranches, which is the same amount as when when we started our relationshiptogether. Many years ago we had gone up to fourteen and back down the twelve.We closed twenty percent of our branches is what we like to say justalong with the industry of, but clearly not that many from a a scaleperspective and the way that we've grown it as we fired on the Cadi side,entrepreneurs and CEOS of mortgage banking, co of lender Finan, a Ce fun,finance, real stute, special Colendi equipment finance and really help thoseentrinus have the autonomy, but then also give them to the support to buildout their team. So we've had a combination of relationship based a can.I growth but then also a tremendous Mont of growth on our consumer portfolio, which really did not exist because we were not a brash face bankwith the health of upstartings and partners. Yeah Yeah is true. I did want to remindyou real, Quick San, that you promise to ask me hard questions here, so I'mnot just asking so do free to interrupt. I think for the audit o want to keepthis as a kind of interactive dialogue and his he said. If you throw in somequestions, will try and get to him throughout the presentation, but we dohave some time reserved at the end for that one of the things that struck meabout your strategy. Strength. Sam, has been the execution of fen TECpartnerships and I'm curious how you think about that kind of build versusby in the the partnerships that you have created because you've been at theforefront, I would say, of the banks. I've worked with in terms of reallyleveraging that strategically. So I just love to understand your thethought process behind how you think about that. As a part of this strategicmix at the bank sure absolutely very good question, and I think that one ofthe things that our current CEO says is that Fin tecs are doing everything thatthanks should be doing but aren't doing...

...big. I chose not to do I choosing to doso as such as difficult operational in from a regular Tory perspective from anallocation, accaparer spectives. So we've always thought a thin tech ofpartnership to allow us to advance into new markets and new channels fasterthan we would otherwise be able to do ourselves and, while the bank certainlyas plans to become increasingly digital across our product Corfiote, we alsorecognize their times when speed to market is critical, when the learningsare too costly or too much of an operational or band with distractionand we'd be hindered by building rather than partment. But one of the otherthings that we also take an this. Is Customers Bank, specific approach andview. Is that we're not afraid? In fact, we welcome an opportunity to be a firstpartner or a fast follow from similar to ourthe. Partnership with outstartcomes with costs, but it comes to the tremendous pensive benefits so, forexample, we're working with not to be named a technology partner today that,but we recently signed up earlier this year and were their first bank partnerand we're sitting in there as they are implementing us for a tech perspective,their documenting, but their on boarding and partnership andimplementation processes for the next bang. That what does that get us? Itgets US Customa Ion. It gets US something that looks and feels good forus to take the cost us more time up front, but then we get something thatfeels more like a custom product, so there's all kinds of pushes and pulls,but we approach it as we approach partnerships as we're very open topartnerships. Yeah, I'd love, your sense on how you,man, you're somewhat newer in a full time role at the bank, but one of thethings that's unique to me is your ability to execute, and I mean thatboth in terms of finding and on boarding a partner and then managingthe relationship, I see banks that really fail at both to have maybe theappetite for partnerships, and then they can't get it through all thevarious committees and it gets mired in internal process and doesn't goanywhere in. Similarly, I do see banks that are able to sign a partnership,and then you know what I came from the Sotheran. We always called it shelf.Were I didn't go it's great. It's signed a like. It doesn't actuallyproduce results for the business and you guys have been effective at bothbringing new partnerships live and making the meaningful parts of the bankand the business and I'd be curious. What you think it is that you dodifferently or better or uniquely well in that kind of process internally atthe bank, to enable you to do that, because I think it's something many ofthe that tennis probably are trying to do and struggling with theeffectiveness that you guys have managed your thoughts on how you dothat. What makes you different I'd love to hear absolutely and before I talkedabout our strength, I also talk what our weakness is. We do exactly what youjust subscribe. We have also made mistakes along. The way where we'veshare signed up was some very complicated software. That's supposedto change our entire organization, and I I I call it to equipe. We received aCD ROM in the mail and we never cook in it for an started, a install it andfigured out the use. Everyone does things like that. There key larinsalong the way, but I think what makes us different than the typical bank is.We have that we call our tech team, the tech team, not it. We truly think aboutthe strategic value of the partnerships and we've made sure that we'redeveloping the right capabilities internally that could be from a cultureperspective. It could be from an operations perspective. It could alsojust be from an Abi Capability Perspective to make sure that we havethe muscle memory for these types of partnerships and it started with somedevelopers and then it involved into a true sort of a tech organization, butreally what really helped us was not a player of best foot for and said. Thisis what we want to be. We were building a Fin tact with an organization and, asa result with bank gobal and as a result, we built a lot of thesecapabilities as a traditional commercial bank. Originally, we wouldnot have otherwise had a necessity of this. who were grateful for that DNAthat was put into our organization. You know Qualifie, plus years ago yeah. I,like the phrase muscle memory to because I do think, there's a degree towhich the first is the hardest, and these are often in banks, untread patsand it gets easier, and I think you guys have done a good job of breakingthat Nick ground and then saying hey now. We know like that. We've learned,we got a little better, O our process and our team gets a little morecomfortable. So I think that's a good way to think about it. I did want asking. Of course we want totalk about consumer lending, so I apologized the audience of taking to onto get there, but I wanted to ask before you guys, ended up being verylarge in the pppat program throughout Ovid, which you know given your size.The time I think you are what is a punched way above your weight class interms of your. What you did through Ppe can tell me able about how you how youdid that what was it that will enable you be at successful on that and whythat was important to you. So I think a lot of P banks were serving their existing customers right,they're just going Hay. I want to make this available to the banks already thebusinesses I serve. I think you quite a bit broader to try and support theprogram holistic ally, I'd love to know why and then how you execute it soquickly on that sure- and I think we didn't go as intentional as it mightseem, looking back to it, but we started a digital application processwhich were the Bank of our side. No one really a different application processand we stood went up with him. Seventy two hours and when we initially youknow, came up with the idea of the concept, but it was no different thanfilling out a form or type in the form in just it was not a dynamic ter I, butwhat we notice of that process, as we...

...kept getting daily day by day, smallerand smaller loans, where, as the average loan size was called threehundred and fifty thousand or five Tortosa at the bank were for customers,we started seem loans as small as hundreds of dollars, and we recognizethat there was demand out there. That was being unserved we'd en took ourtypical body to be in this case, although many of that last be kind offelt like seas, in the sense that you had of hundreds of dollars. I bet itdoes yeah and- and we set up in tech, parker ships with the cabbages beyondthe Lendes of the world, who had many of these direct relationships wherefolks that were being failed by banks, because banks were either had a manualprocess when they were overwhelmed by trying to design the system tounderwrite a program that normally he would stand up in two years. But thiswas an emergency pandataria and an so. We had a good combination of been smallenough but big enough, but I think that allowed us to sit in the sweet saucespot and actually be able to get up. As you said, quenchable our way, we didfour hundred thousand loans over a proximus under four hundred thousandloans, eleven billion dollars of PEP PE funds. Dis. First I mean for a bankthat was nine billion, an asset just a handful of years ago. Eleven billionand funds disperse for Peppi is quite an achievement. They congrats in andwell done. That's an impressive execution, I think so. I will switchnow a consumer law since thing we do and the thing that we work on together,but custer being made a real decision a number of years ago to enter consumerloans in a pretty substantial way, not just in partnership with upstart, but Ithink broadly and I'd love to get your thoughts on that, because it's notsomething many banks. Do I see most banks, think consumer is mortgage andanything it's not a mortgage like. I don't really touch it any scale than anyou. Guys chose a different path, so I'm curious why that was strategicallysomething you wanted to invest more heavily in sure, absolutely very goodquestion and we we had the bank. I recognized because we were branch likewe saw ye know the bank was required, but the cart Menein team in twothousand and nine. It was two hundred million in assets and it greworganically to when our relationship start that started to day billion. So,as you can imagine called branches, two hundred million twelve franches, eightbillion that clearly didn't come from the consumers of deposits. We not cometo the consumers yeah hank global. We set up as an opportunity to raise fordigital bank deposits and consumer process. We now wanted to match fund and wewanted to create a diversification of the franchise and we felt thatsomewhere between ten and twenty percent was the minimum threshhold wewanted to get to. We knew we weren't smart enough tobuild it ourselves to your earlier question. We also knew the becomingextremely adept to know within it. For a period of time, from a model,perspective would mean that we need to purchase a lot of data more build a lotof data ourselve, which means you lose a lot of money yourself to be able tocreate the right low. So what we did was, as we started purchasing loans,flow arrangements and referral arrangements. With some of the talkconsumer lenders en we learn from them, they helped educate us. We help them,they helped us. We started with a lot we narrowed to a few, and then weleaned in with the best partners I'll start as an example that really helpedachieve our you know longer term goals and then what started as a wholesalebusiness has now involved into it in a direct franchise, enhancing businessand that's the way that we planned it and that pass that we've been on. Itwas a path. This time it was a was an intentional path as a year at, and you guys are in a number ofcategories of consumer Lans. This point right. What is how many differentthings? I know we do on security together and we can talk a bit aboutthat. But can you tell me about what your brought or consumer portfoliolooks like at this point sure? So, as you can imagine, we have a smallresidential. You know portfolio. We also have an historical manufacturhousing for fillings, part of our consumer Bocke. I want to like Li a seebusinesses with the bank. We a we have the the personal loans we have studentloan refinance and we're looking to you know and momentote and I'll call thenext tobes all good categories lending I like that, and so let's talk about coved andthat's the title of the Web, nor right like consumer lending in force to covet.I think when I was talking, the banks call it three years ago. There would bea concern that was both hey. This is theriskiest thing we could do right like you're, not there's no house, there'sno car there's nothing to take back. This money is just it's not even like astudent loan. They can survive bankruptcy like an unsecured consumer,Lana, the riskiest kind of loan in many ways that a bank and engage in- andthere was nervousness about that, but then that nervousness was certainlyenhanced by. We were in a detail end of a pretty good period of years. From acredit point of view, is this a moment in which to engage in? Am I going toreally at the end of the boom times in get an best in this risky category andleave myself at risk during what is inevitably the common terms of adownturn. Tell me how you guys were thinking about that particular risk atthe time, because I know we started working together a number of years agoin the consumer space and I'd love to understand the evolution of what yousaw through the last couples of months...

...a year and a half or so of a veryunusual economic circumstance. But certainly you know one that I think iscould properly be described as a macrocosmic of some real portion. Sureyou know I'll start with how we viewed when we entered into the business, which I would call thedigital personal loan business a couple of years ago, that we felt that therewas Ousak to be made. So we could. We felt that we could at that. Yes, therewould be charge ops. Yes, we would quote uncle, be writing off loans andlosing money on some of those. Having said that, we felt that we were justpreportionable being paid for that risk, so that was the you know, intentionthat we had when we started the business once the pandemic cat. Itwasn't too far into the pandemic, where we started to say this is theopportunity for us to prove everyone wrong and to say that we were right andwe had some very tough conversations that you remember very early on an thepandemic, and there was a lot there, a lot of parallels initially withhurricanes or natural disasters, that you could view that we thought were avery important parallel banks or building reserves, planning for apotential of call, em insurance reserves for a potential downal turnand potential credit losses. But we didn't necessarily anticipate that. Iwould go that far now. None of us anticipated that the stimulus would beso large and would be so sustaining and continue in, and that's one of thethings that I think has been a little bit of the while we had three caps onthe fetter. Now we have two because, at the end of the day, credit a creditrisk is lower today to be completely frank because of all the the ostanesflow into the system in the non fire Hayas a man soter that we anticipated,but at the same time there was a six to nine month period of time where we feltwith or alfader margin at increased, even more significantly than we dogetting paid for. prepended can I know some of these answers, but were you actually increasing your kind of targetreturns and pricing on the loans for the ALPI will say for those who aren'tas familiar as somebody who plays broadly in the industry, we saw adramatic pulled back from banks from capital mark. It's just a high degreeof nervousness on which I think to your point was. It was a massive opportunityif you believed in what you were doing. What is it warn? Buffin says wheneverybody else is nervous, get greeting and everybody was nervous, and it was amoment. I think where you could go and take advantage of that. How did youthink about? Were you tightening credit boxes where you raising rates like? Howdid you think about managing the fact that it was a non zero risk that youwould see deterioration and performance? And then we can talk about what we hado? We had what actually end up seeing in the portfolio, but I'm curious howyou thought about managing that increased level of risk, even if youwere trying to take advantage of the opportunity yeah. So there were, therewere competing the competing sort of decisions that we have, so one ismessaging and investor concern, and the second was is taking advantage ofshowing that we were not tourists as well to be a sort of active in market.It's important to show that you're not going in and out same goes for amorgage warehouse business. That's why we've hatch us a tremendous yearbecause we've been able to show those customers and to show the industry thatwere willing to be there but good times, and in tough times and as such we gotrewarded in the good times. The past twelve months we did cut the credit box,which again is an output versus n inputs in our overall relationship, andwe raised it from at the time I think pre Ben Dan Lik. By about twenty. Youknow twenty points, and that was more an a little bit to appease some of theconcerns that were out there in the market, because not everyone canunderstand the analytics and a monding that goes into thinking about this, butwe and we paused all pool purchases which we were still doing in and weleaned it, and I think that it served us well and it allowed us to fire a lotmore customers and- and we've been ramping up since I think we have sloweddown to thirty to fifty million a month. The middle of last year ended the yearbetween fifty to seventy five a month and we're in about a hundred milliondollars a month or so an origination today, Yep we're glad to be doing aunde million a month with you cause. It's been, it's been great, so I guesswe should back up. I forgot to do this earlier and talk about the nature ofthe relationship. You know we have with customers, Bank we've, obviously thepartner together for a number of years. How do you think about how thepartnership works? Maybe I'll give from other sarge point of of view? How wethink of how we work with thanks, broadly, would love to just start. Withyour perspective on on how this partnership works. For you yeah, it'sbeen, it's been a great partnership for us. It's been a long term partner.Again we were an early early bank partner of Love Star, which has allowedus to have a nice Protegi, furnish Ip a nice two way, partnership and be andallow us to help upstart build their business, but importantly, allowupstart to help us build our consumer long distese. So it's been mutuallybeneficial from that perspective. What I would say is is that in tough timesyou recognize, who are your best partners, your best friends, O yourbest team members and from a and the froze of the beginning of the pandemic,when it's hard to remember those times right now, when we really didn't knowwhat was going on, we didn't know how long this was going to last, or wethought it was going to last eight weeks and we'd be through this, andeverything would be back to normal. We had, I think, a very good experiencewith upstart from from a e from being...

...open and transparence in a servicingrelationship, human relationship, I'm thinking about the customers at the end,as opposed to just the dollars in a sense, and I think that we valued thata lot and that speaks to why, a year, just over a year later, our partnershipis growing and deering. So I think that's a very important aspect of whoyou parker with and our relationship. I've always enjoyed the missionalignment. I think we have as well about providing high and consumerexperiences and focusing on the end. Consumer is the goal which I think hasbeen a true north for us for those who are less familiar with upstart, we arean eylandt platform so providing, unlike mini fin tex. We are entirelypartnered with banks and have been from the our earliest days and provide kindof digital origination. Experience, including probably the thing we're bestknown for is the application of Ai to credit underwriting and deep ranalytics to find those borrowers who are more credit worthy than theircredit score might indicate, and I think we can we'll talk about that in aminute Sam, because I think there's been a lot of question about. Is thatreally going to work? How does it work? Is it going to survive a downturn, andwe have some data on that now, as we've come through the pandemic, the otherarea, we really apply machine learning and artificial intelligence to is thekind of simplification of the Bar wors experience and and how do we getthrough Kycier fication, the income verification without asking for a lotof documents and those things really end up, making a huge difference in theconsumer experience. But importantly, in the economics of the program in thecost if to bear to actually process those loans, I think one of the reasonsso many banks have not been in unsecured lending is because the costsare too high to justify the the revenue and when you can reduce those costs,you can make this really economical and we offer the ability not only to havethe origination experience but also to drive demand to that experience forpartners, so we're actually helping you guys force that hundred million a monthand then servicing on the back end for partners who want so and we've beendoing all of that with custers bank. You guys were the first partner thatwas actually keeping loans on balance sheet with US way back in the day, asopposed to programs that were selling loans into the capital markets, and sowe've evolved a lot together and are looking at now. The second product wehave entered, which is on a refind, eventually auto purchase financing,which I know we're talking about how we get into that game together. So we'vebeen excited to do that. With that background, I did want to ask youreally quickly you one of the things I've been talking about a in creditunderwriting for ten years now, almost to banks and one of the questions itwas hardest to answer for any banker was hey. You show me some great dataand it's all from like good times and yeah there's some like betterdifferentiation and the credits better. But what happens when times are not sogood? Is this going to deteriorate much more so than the thing I've watchedthrough a number of cycles like credit score? Did you have those concerns andthen how would you categorize the performance of those models through thecurrent maccoom IC stress environment, brought on by Ovid? Absolutely we hadthose concerns. Our investors have those exon concerns because it was anew business for us as well, and I would actually pose it back to you andI'll. Ask you some of the questions that I get asked in Er that you havehelped us answered. The cautin you'll have you'll be able to t. We can talk alittle bit about the overardent, just our direct relationship and that'llhelp get to the question that you're asked one question that we could ask:Why do you need a thousand or fifteen hundred? However many bird? Isn't itreally just FICO and one or two others, and why do you need all these and howheavily waited are be thousand one and the eleven hundred one the sixteenhundred and first variable yeah? So we have something like sixteen hundredvariables that are contributors to our model today, and it's really amazingthat you find that there is very little power. Even I think you got to get overa hundred variables to see something like half of the explanatory power ofour model and then every little variable is not super important.There's you can take out anyone variable, including a credit score, andit wouldn't really change the outcome, because they're of the sixteen hundred,many that are saying similar things, they're related, but it's the the onesthat are related. It say slightly different things and understanding howthat really reflects a difference on credit worthiness that gives youUPLIFTS and I think, to take advantage of this. I think you really need acombination of three things and not many have put all three together andthat's the sixteen hundred, I think, of the data like a big spreadsheet right.Sixteen hundreds like how many columns do I have and sixteen hundred columnsis great, but then I need to make sense of that. I need a lot of rows because Ineed a lot of people who look very similar in one calm of a similar creditscore and look different in others to understand how to find that slightlylower credit score bar or who truly is credit worthy and we know they're there.The beginning in sight, we saw was in a subtribe pool like if twenty percentdefaults, which sounds awful until you realize it means eighty percent ofpeople page you back. Could you just find the eight and that's really wherewe started, and so you find that if you have those calms, you need a lot ofrows. We've now across all of our partners originated Oh for ten billiondollars in unsecure consumer loans, and so that's hundreds of thousands of rowsof data, so to speak in the spread sheep, and then you can't use your oldlogistic regression goes to a score...

...card that prints out on a five page PDto take advantage of those things right. You can't say if this variable goes aup x, you need really sophisticated algorithms and those algorithms requireboth the large number of columns and the large number of Ras. So we foundthat it's really important, that if you have those rows and those columns andcan apply those techniques, then there's tremendous ability to actuallyfind those relatively credit worthy. Yet not yet high credit score barbersand even to find those high credit score barers, who actually represent asomewhat higher degree of risk. So it really does take all the variables youcan take one out, but every one of them adds a little something to the modeland it's the sum of all those little somethings that ends up making a reallytremendous difference in your understanding of the credit worthinessof a specific consumer. I think that we've gotten to understand that we'vegone to see the benefits of that now. I think one of the things that we talkabout in the banking business, both in the consumer side, as well as thecommercial side, is that there's the gut there's the relationship, there'sthe shaking someone's hand looking at them in the Aran, knowing them in thecommunity. That's what really helps reduce crepis CASO. How do you thinkabout that Visata, a digital relationship with the customer thatknow one from the bank as as Mac yeah? I think it's a great question Ye.Obviously our models were all trained on a performance of loans that didn'thave that relationship and I think, particularly as you get into larger,different kinds of loans of same mortgages. That becomes really crucial,and actually I love the model. You guys are developing because I do think thata hybrid of the two will be valuable in the future, particularly less forcreditors, point of view and more from a service point of view. But I do thinkyou see- and we see this in pools of loans- that for customers with longstanding relationships with a particular bank, they perform better ona loan from that bank than they do want a generic long. There's elements ofthat that we see and will price with our bank partners, but they have a longstanding relationship. But I think when you can see it, it means you can alsosee the absence of it, and so you can properly price and manage that risk. SoI think there will be real value for banks that can cross, sell theircustomers on multiple products and thereby take advantage of the lowerrisk that comes with a long term relationship on understand how to priceand quantify that lower risk. O It's not just a hey you're. A customer willgive you fifty bips off, but you can really understand how much less risk is.This customer WHO's paid off? Three loans with me than a new consumer, butof course, there's also real value for the bank and being able to bring a lowfriction low cost experience. It can bring new custers into the bank becausethe great thing about that loan that's to a non existing customers. That's anew customer and you can sell them another one where they won't be a newcustomer and you can help them with other needs and I think, being able todo both as really important. We've certainly seen that, particularly insmall loans, so unsecure lungs, like people want fast. They want easy, theydon't for the longest time. My Co would go on to his bank and I won't it's abig bank, so I won't name it in the Va web, an art to make him feel bad. Butif you went and said I'd like an unsecured loan, they would pull up aneight hundred number to schedule an appointment in the branch and he mayhave liked shaking somebody's hand. But in that moment he wanted to know howmuch he qualified for and they weren't giving him that information. So I thinkproviding the best consumer experience does mean meeting them where they are,which in this case may mean in their pajamas on the couch over here, andthey want an answer then, and I think you can ultimately combine that withrelationships and high touch service as it's needed, but not necessarily foreverything, which is, I think, the default. The many banks go down to yeahyeah. Absolutely like the point you made about it's a little bit of selfselection as well right at exactly trying to find customers who want thisjourney in this channel as well, and you touched a little bit on the therelationship and the cross selling and owned a relationship, and you know howI'm sure you get asked a lot of questions as opposed to just talkingabout our experience and when the end up when a bank partner comes to himsays I am in Pennsylvania, New Jersey and I want I want to referral loans inPennsylvania, neerest and that's all I need. How do you think about the theold branch bottle and the geographic lovel versus a national model and adigital French? You know type type, feel yeah, it's a good question. I seebanks a go different ways and I'm always recommending the banks to thinkof a digital product as a way to build a national footprint. I will say:There's so many banks that want to go digital, but they want to know that thecustomers are in a footprint where they can walk into the branch and talk tosomebody if they want, I think often they feel like their ability to crosssell. Other types of products is related to everything else. I've got issold in the branch. So if I want to bring a new customer in, I need to havethem able to come to the branch to cross sell, but we really push them tothink about a way to broaden. We can obviously restrict when we're doingreferral flows for we're helping you find new customers. We cannot. We canrestrict that to the geographic footprint of our banks. Let's say I'min these five states can just get me launch mirror and that's fine, but Ifind many of them are increasingly interested in using these digitalexperiences as a way to broaden their experience to see what it's like tohave a customer in estate. They haven't been in and dip there to in the waterof moving towards a national digital experience, as well as there inpersonal experience, and then the other thing that is, we do work with ourbanks that have more, you know, substantial existing branch footprintsto say how do we bring this experience into the branch? How do we enable thebranch employeeto talk about this kind...

...of product to the consumer and thenoriginated in the branch? Maybe on the consumers phone, but in consultationwith that banker? We've had a lot of success, but some of our partnersdriving adoption within their existing customer base through their existingbranch network. It actually can work quite well, so we always want to serveyour current customers. But think of this is a way to expand both yourcustomer base and your geographical footprint, because there's a lot ofopportunity and demand out there for that absolutely couldn't agree more sodid I fully answer. You know your question with the reverse questions, Ithink so the question I'd have for you is: Do you think this is going to be aturning point moment that you know we cannot just say yeah models performedwell, but we can actually quantify in it might as well as well. We're heremight as well throw some of these slides up and show the qualifications Ithis is. I pulled this out of year. I think analyst called DX, I'm sureyou're familiar with this, when we actually look at to quantify what youand I were talking about the performance through the model. Do youwant to tell the audience here what this slide is showing Sam I'll give acontext of the experience we've had more broadly, but the lovey walkthrough what you say: Credit Performance through the cycle on thisportor. Absolutely so what you on the top line is the industry of consumerloans and consumer installment loans. It goes on secure consumer loans andforbearances prior to looking at the date track and through the pandemic andthat's the red line. So you see that an he got up to. It seems likeapproximately sixteen percent of so lo in some sort of before parents. Now, ifyou look at a customers bank direct, you can see our our increase was lesssharp, but obviously then the margin and Adela was significant. If you alsopulled out upstart from that, you would see that upstart and our relationshipwith outstart did even better than the Blue Line, which is a Blandan line, butacross Er over all portfolio, and I think a lot of that has to do to withnot only the underwriting, but also when I teston earlier, which is theServos of service oriented nature, but a start services, the loans that theyhelp us O for referral perspective, but yeah. I think that service and runningthe the referral in t the deferr program. Sorry on the talents, wasreally important for context. It's the shape is remarkably similar. This iswhat we saw for the industry and across all of our lending partners. We havethat, and you saw this kind of increase, but a much more utie increase inoverall delinquencies- and this was this payment. Impairment for theaudience is a is a combination of both people who are in a deferr program orpeople who are in a number of days late on a loan. So it's the total impairment,whether they're, just not paying or whether they've entered deferre andwhat's remarkable, it's come back to normal. The industry is still slightlyabove normal, but as of the state into last year, but the up star Port Foliis,really back to where it was before, which is to say we saw a very small,increased relative industry and almost a complete return to normal and then,as we talked about the AI models, this is one of my favorite charge, but it'sreally hard to read, but I think it explains why. If you looked at thegross numbers, the customers bank impairments were much lower than theupstart overall platform, and that's because, as we work together sand, youguys don't take, you have a limited risk appetite. We have other lenderswho sell, unto the capital markets and don't retain and they're willing tooriginate loans that are riskier as we perceive them than you are, and this isa chart of you know how risky upstart perceived a loan in terms of tears andthe credit score, and what you can really see is now our banks that werelimiting risk by the upstart risk prediction, so not only better cridperformance in good times, but this is impairment rates when you are in tears,one through four in the less frisky category of loans, your aparment weremuch lower than you were, if you were over here and what I'ms declared as themore risky loans in a way that just wasn't true, if you were using onlycredit score to do that right. If you tried to limit your Ceteris by justlooking at credit score, you didn't have the same kind of relationship andwhere you were really limiting risk up impairment here versus what happenedthan this, and this is why? Because you guys, are looking in the lower wist.Hers is why your you know portfolio had less impairment, a less losses and goodtimes and less impairment during the current crisis. It's so. I think thisdata has been really quite stunning in some ways to see how much morepredictive of risks and impairment to the portfolio the upstart model was.You can just look at the average column for credit score and the average row atthe bottom for upstart tear and go well. If I have had to pick one of those tolimit my risk, it's not hard to figure out. If I want to want to use thecolumns or the rose it's, I guess the question I have for you coming out ofthat I'll stop sringe house. Do you think this is if that was a big concern?Okay, you've seen the models and I've seen the accuracy, metrics and goodtimes, but what's it going to happen in bad times? To that, do you think,coming through this experience will be a point when more banks become open toleveraging these technologies because you've seen them through a cycle, andyou now have some evidence to look at not just assumptions about how they'llperform, and I think so far that data is really compelling absolutely, and Ithink that this goes to the point I talked about. This is an opportunityfor us, customers, Bang to show case a fact of our franchise. That's more theoutput, but the fact is we got data and the data is what makes this muchstronger is yes, the out the output is what allows people to feel morecomfortable, but the fact that that we went through it makes the outstart moteas an example that much better yeah.

That's that's right. We certainly spenda lot of time in proving our models for how to adjust to R and in a way that isUN pressing, like the rapidity of change of the economic situation. Inthis sistance was so much greater. Normally you got it takes months tohave this happening here as a matter of days, but I think we learned a lot andI think, we'll be even better positioned the next time we go througha cycle, but certainly the data to me seems to indicate that if this was yourconcern, is a bank what's going to happen during a stress period, there'spretty strong, compelling data now that it's really not a major concern, so I'dlike to turn over audience questions in a minute, but before I do, I justwanted to end by asking you you're taking over the helm of the bank. Youguys have been really on the cutting edge of inch of interesting stuff insome ten tech partnerships and consumer, leading. What's on the horizon, whatare the next two or three years for or customers bank under your direction?Look like sure I think that's wear a bank and transition, and it's notnecessarily because of leadership. It's also just partly because of maturityand grows a across ten billions. The first time, just over a year ago, withinto the fourth quarter, two thousand and nineteen, and we had paused at ahundred ten billion for Durban, release and their reasons relating to our banklobal ownership and investment. So I think that there's a resumption ofgrowth there's a building out of a bunch of our businesses, but reallyit's it's thinking about the future of a bank where banking is going were notburned by legacy technology and a legacy branch network which, in somecases, going to have some benefits. But really, I think the last year has shownthey were trends that were already hapning they've, been accelerated,their getting momentum to go banking, misial customer requisition bank as aservice. These are all things that were already a part of customers bank in avery small way, but we'll be leading in more heavily over the next couple ofyears and the way that one of the things that we need to do is is focuson change management within the organization as tech forward as we are,we might be better than ninety percent of the banks. That's not saying much tothe non against the non banks of the Tech Company so but a lot of work fromthat perspective, but I think that we're looking to increase and lean intoour consumer loan portfolio and build it into an overall overall strongrelationship with our consumers, both in the asset sign, as well as a liveboy side as well as across to side, and I think that's taking what has provento be a strong. We led with the asset generation and now we're going tocontinue to try to create so strong customers, not just through the life ofthe loan. But though for life through digital means. I think that's going tobe a very important aspect of how we approach not only or consumer loanborrowers, but also, importantly, over all, to have been contradic, create aOmni channel digital experience that doesn't feel like a digital bank versusa branch bank versus a commercial back. And that's interesting, and I like that.You talk about leading with the assets. It's one of the things I'm starting tohear more from banks. I talk to, which is this question of. Can we lead withLendy? The traditional model has always been you sponsor a little league team,you bring in deposit accounts with relationships, and then you and thenyou offer them loans as their a customer, and I think this method ofactually bringing people into the bank through a revenue generating product.Frankly, it's better to start up with a want. It is where the revenue comesfrom, but I think it's really interesting because I most banks I see,are still hesitant to lean that way, but I think the world is shifting thatway where, if you can do it, why not and it's great to have a partner who'spushing at the front of that absolutely have to, because it's tough to acquirea deposit customer Dittay that what are you offering other than rats? So that'swhy you need to think about what is a customer need? How do I service withthe customer e end a hen kin o? I G get a feel over all relationship. Yep,perfect thanks for joining Sam I've got a bunch of questions coming in from theaudience Jinny. I don't know if you want to ask, but I can pop on here:I've got one that came in that I'm actually kind o interested in thecontext of this relationship. Question which is you mentioned that you startedout with some hold on purchase programs, but you're now in an origination model,at least in terms of your upstart relationship? How did you think aboutthe importance of originations originating in your name versus buyingloans on the back end as more of an asset allocation thing? Why do youthink that's important to you yeah? So I think that one is a wholesalebusiness and, and one is a franchise and handsome business, so the wholesalebusiness we were very open about. We were learning, we were purchasing, buteuro price taker versus a price maker, and that's really as simple as it was,and so they were transitory loans booked on someone else's paper. Weowned them, we made a premium paid servicing and- and it was truly anacidly relationship and I think that's the transition that we needed to makeas we continue to learn. Build outer models build out the other partnershipsto be able to figure out, but how do we build the best digital first consumerbank within our commercial yeah? I like it. Let's see what we got here in thequick. My questions pulled up out of out of the ten billion on originations.What is the average Lanman? It's a great question. Sam, do you know theaverage Lon amount and your port folly, I'm sure Ed's going to pin it to me inslack. If you done- and I think m between twelve and fifteen sand is my-is my guess and probably knows best that I'm sure I does best. I thinkyou're right is between twelve and fifteen, and the upstart platform is awhole. It's a little lower in, and that's really a relationship of thebroader risk profile that we have across all the lending partners that wehave and the lower risk loans tend to have slightly larger loan sizes. Youcan imagine hiring come consumers...

...higher cret score consumer that thecubby port folly is a little bit larger on an average Lan size basis than theupstart port fly. As a whole, but you can think of the unsecure consumerloans like as as three five years between one and fifty sand, buttypically you're, seeing a ten to fifteen thousand dollar average loansize and the chunk of that being credit card refinancing high interest atrefinancing is a big is a big chunk of that. Let's see. Okay, we've got other cooksto AI models, replace human underwriting, a hundred percent, but he thinks Sam a hundred percent,definitely not a hundred percent. You can't even book a hundred percent theloan without without sometimes having women intervention, so I think the wayto think about the Nautilus you're helping, but some cases you're,building a credit box, you're opening up a door based upon a box and you'retrying to relate replace, as I mentioned, for a little bit of the gutof the relationship manager based on a lot more data than the relationshipmanager has when they're making that cut face decision, yeah and I'd sayfrom that Star Platform. Point of view. All the credit decisions are fullyautomated. So there is a complete replacement from the risk analyticspoint of view. We're not saying hey, come in and tell me you think thisperson is better risk. There is still instances where for fraud, preventionfor verification for other things, there will be human interaction andagain we try our best to limit those because they do have and we find if wecan go to a no human interaction model. We convert roughly twice as manyinteresting parties into loans, that's good for everybody, but we also don'twant to do that at the detriment of preventing fraud. Maybe you've keptfraud below thirty bips across our history across the platform. Sobalancing us to is really important and we did that's not to say you can't callif you have a question you want to talk to somebody we're absolutely there tomake that happen, but at the same time, as I say, a lot of people don't rightthey just they want to get through it and the more. We cannot ask you to upload a pace up or a W T or a bank statement to complete the process. Thebetter off it is for the consumer, the lower cost it is for us and for thepartner and generally the better performing of loans will be so we dofind it that it works. That way, let's see Itri to look through myquestions. Oh here we go during the pandemic. Howdid you monitor and assess the performance of your personal importfolio, and how did you get confident going back and expanding originations?What was the cadence for the process by which you were looking at Hallis wasperforming and getting comfortable as you to increasing the originations aswe as you got a little bit farther in and said, hey things are going well,let's start growing again what what did that look like? So there was a dataelement and a human element to that. So we, for example, with all of our will.We call sort of our service by other portfolios. We started evaluate them ina daily basis if we could, on a weekly basis at a minimum, we had weekly callsso each of art, our services. We were gathering our bone data and developingour own proprietary views. We were evaluating hurricanes and naturaldisasters and Puerto Rico and Catrina, and trying to think about all of that.What history could help us to inform in addition to what the data you know wastelling them to send some, I think, as I mentioned earlier, some of ourpartners really showed strength relative to others and and we leaned inwith with those partners, yeah yeah, that's great, and I think weenjoyed the ongoing Dialin. It was nice to have a good partner who was lookingat the data and no making rational decisions where I think some weremaking fear base decisions, and I think that process worked well and Kudos toyour team. I think for being on top of it and well respecting of the day aunderstanding of it. So I've got two questions here, related to fair lending,and this is probably, if what's going to happen when the economy turns out,is the number one question I get asked about a models. A fair lending isprobably number two, and so I'm gonna assume you'd. Rather I answer them. I IO. Yes, that's. I think whenever you look at theapplication of a inl to to any area, the question of fairness is central andit was central for us as we began developing models many years ago, andso we actually went and had a conversation with the consumerfinancial protection bureau before we started lending with any of our bankpartners on. How should you think about fairness in the context of the modelswere building in the data we're using? I will say we were native enough thatwe walked into the CF enforcement office as opposed to their office ofinnovation, because we just we weren't sophisticated, regulated inities yet,but what what we came up with them was a process to really think through. Howshould you evaluate fairness, understanding that the world as itexists, is not totally fair? If you look at the distribution of creditscores among different classes of people, African Americans versusCaucasian Americans, they're not the same, and so if you're, using evensomething as standards, credit score, you're going to have a higher approvalrate for White Americans, and you will for African Americans or HispanicAmericans, because they just don't have the same credit core distributions.It's a that's true, so we can't just compare it as everybody being treated.Exactly the same, so we came up with a waterfall test mechanism that we builtwith in conjunction with the Bureau and Perform Cortelyon the portfolio andprovide results to the bureau. The result of that in the end, was theissuance Bythe Bureau of what they call...

...a no action letter which is hey, hey,we've looked at this. It makes sense and we don't see any cause for concernon specifically the topic of fairness and frankly, what we found is that in aworld where, where in accuracy is high right, where we're turning downeverybody below six, eighty, eighty five, ninety percent of who might begood borrowers- that when you can have a more accurate model, everybody canwin, and that really means two things: one. When the CF compared our resultsto what a traditional model might have done with similar risk levels on thesame portfolio, they sat, they found it for every demographic. We couldincrease approval rates in lower interest rates right, so we're charging.People last were proving people more than a traditional model across allclasses, and then the question becomes. Are you? Is that impact beingdesperately felt by a protected class or not in there? What you find is itthis day that can often offset traditional biases things that are noteven redistributed today and so we've not found any cause for concern and ourwork with the bureau and that's something we will continue to do. But Ithink it's an important question: It's not an easy question to answer, whichis why we went right to the front of it, but we really believe very stronglythat the uses of these techniques and alternative data points can helpimprove the fairness of the overall credit system because of how muchunfairness and inequality exists in the system today and we think we're seeingthat in the results of our testing our work with the bureau. So it's don'tthink of that is a back foot question I think of that is something we reallylead into is frankly, why we got into the business in the first place, Gustask a fall question of Matin. I think that if you are the CPA, just generallya regular, a static model seems easier to take. You know no action, so howdoes a regulator or how so the bank partner think about a regular orunderstanding and getting comfortable that dynamic an yeah? It's a greatquestion. So I think what you have to do is move the level that you'rethinking about the model at upper level, which is to say, I'm not worried aboutthe specific like credit score box. As much as how do I determine what thatbox should be? What's the process that the trains the model? Where is the datacoming from? How are we assessing fairness? How are we assessing accuracyand how are we overseeing things that might go to production and if thosethings are standing the same and I'm overseeing that process, then theoutput of that process should be good. In fact, as a model is trained, one ofthe benefits of having defined the test for fair lending is that we can run iton every version of the mop and go hey. We tested it under the old model, we'vetested it under the new model. We still have good results, so we're comfortablegoing alive and that's why I don't know that frankly, there's any otherplatforms doing that, but thinking about evaluating, overseeing theprocess level and the oversight that the criteria to launch the testing thetraining process. That's the core thing I think versus saying hey I want to. Iwant to prove every change and ficus born. Of course you guys do, approveand dictate every change in credit score box, but I think that's where thethinking evolved at the regulatory side, where the banks have to get comfortableto is saying, I'm not don't need to look at the code for every model change.I need to understand how it's being evaluated. If that evaluation processis changing now, that's a different issue. I want to see very clearlywhat's happening, but if you're using the same training and testing andevaluation method and the model gets a little bit smarter, I'm comfortable,that's still effectively the same model and it's just a little bit smarterversion of it and that's. I think that the shift and thinking that has tohappen to effectively take advantage of these things. So we get a question. Tisupstart provide the digital application that a customer would fill out or onlyprovide the underwriting engine. My answer is up to you. We and ourpartnership with customers bank. We provide the digital experience. Weprovide the flow of loans and the servicing and we have other partnersthat just want to consume the underwriting engine and do that througha programmatic interface. I think that's not as common and unsecuredloans, because, frankly, not a lot of people have a great experience forunsecured loans today, but it is available. If that's how you'd want toconsume it. Jeff. I have a follow of question that dish. You know what isupstart ANP o. How do you think about yeah? It's a great question. We measureN ps very carefully, typically on originated borrowers soon, afterorigination in our PS hovers right around eighty. I don't know what ourlatest number is cross. All our partners and, frankly, our bankpartners that are retaining customers, bank being sable, typically haveslightly higher n pss that are programs that are selling to the capital markets,because the number one thing that influences people's happiness is thereright and because you guys are able to offerslightly better rates than those lenders that are song or loans to hedgefunds and others. Typically, that results in higher customer satisfactionbut Ittie, I'm being told as eighty one right now I never rate. Eighty one is apretty high ps N and we monitor that weekly daily following the trans,because it's really important to us as a metric of is the experience forproviding meeting the customers needs and that's how we got into thisbusiness with the consumers our true north, and this is one of the ways wemake sure that we are staying true to that true north and just as a point ofreference, I'm sort of many folks in the going to call know their own andPSS, but where would you say top ter thanks R or Ober? I D N? I don't wantto make anybody feel bad say, but the...

JD power looks at the N ps for theindustry as a whole. I want to say it's in the S O S. I know there are some oneto lower in that. Actually NTS for those horn is familiar. Is a netpromoter score and one a ten? Would you recommend this product or service to afriend or colleague, and I think it's they don't call me to like eight to tenor eight to ten are considered promoters, seven to five or considereda neutral and form blower detractors, and you take your promoters and minusher to tractors. So a negative one hundred is a possible score, not just as zero, and we do see some ofthe financial inity who go into the negatives where there are more peopleunhappy than are eight to ten is like really got to be very people to be veryhappy, and I think you either told me: customers banks it's right at eightyone. So that's well above where I think anybody, I think the only people in theindustry, Financial Menestry that have an at or probably, USA, who typicallyhas a really high customer satisfaction score. Most of the rest are in theteams. To this, I think, is pretty devil I, the best top tier isconsidered fit. Fifty sixty w o the o o depends on how you measure it and whoand what service. But it's eight is quite high and looks a lot more likemore to typical tech products in the financial and in the broad consumerindustry. Then then financial service a well. So I guess people are asking ifapproval rates have gone up and if consumer demand has gone up in twothousand and twenty one, we saw certainly a dip and consumer demandgoing through the pandemic in our minds. It's mostly recovered. I think Ibullied a little bit by stimulus, but we've not seen a substantial slackeningand demand from consumers and, frankly, the average approval rates gone up. Theapproval rates depend a lot on the marketing mix we may send like wherepeople are coming from, but I will say on the same set of barers every coupleof months: the model gets noticeably sparker and whenever the model getsnoticeably smarter, we can approve a few more of any given population ofbarbers and lower the rates for a good population more because there are stillmany good bar wers that we don't recognize it such and every time youpull one or two losses that you could predict and not lend to. It gives you ahandful of borrowers that you thought were too risky that you could pull inand actually better understand. It is credit worthy, and that happens for usevery month that we may choose to Mont Mail, a slightly riskier group that wedid last month and offset that from a gross a privilage point of view. Butevery month the model is getting smarter and I think sand with a task,it's maintaining his performance from credit and that we're not seeing as itapproves more people a deterioration of the credit performance. It's just USreally finding those properly credit worthy people more accurately than wedid a couple months ago. Upstart partners with thanks and credit unionsto help grow their consumer, Lone Port Molios and deliver a modern all digitallending experience. As the average consumer becomes more digitally savvy.It only makes sense that their bank does to upstarts a islanding platform,uses sophisticated machine learning models to more accurately identify riskand approve more applicants than traditional kind of models which broadrates near zero upstarts. All digital experience reduces manual processingfor banks and offers a simple and convenient experience for consumers.Whether you're looking to grow and enhance your existing personal and autolining programs or you're just getting started up star, can help upstartoffers an into in solution that can help you find more credit worthyborrowers within your risk profile with all digital underwriting on boarding,long closing and servicing it's all possible with upstart in your quarter,learn more about finding you borrowers, enhancing your credit decision, ingprocess in growing your business by visiting upstart for Dash Banks, that'supstart for Dash Banks, you've been listening to leaders and lending fromup star make sure you never miss an episode subscribe to leaders andlending in your favorite podcast player using apple podcast leave us a quickrating by tapping the number of stars. You think the show deserves thanks forlistening until next time. The views and opinions expressed by the host andguests on the leaders and lending podcast are their own, and theirparticipation in this podcast does not imply an endorsement of such views bytheir organization or themselves. The content provided is for informationalpurposes only, and the discussion between the host and guests should notbe taken as financial advice by companies or individuals.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (31)