SPEAKERS
Rich Alterman, Richard Weil

Rich Alterman  00:04
You're syncing up and tuning in to The Lending Link Podcast, powered by GDS Link. With a modern day lender can dive deeper into the future of data decisioning and Credit Risk Solutions. Welcome to the show everyone. I'm your host Rich Alterman, and today we're syncing up with Richard Weil. Richard is an experienced credit risk executive currently doing some independent consulting. Richard has over 30 years of experience working in various risk management marketing roles and several well established banks and fintechs. Most recently, he served as Lili Financial's Chief Lending Officer with the objective of building lending business prior to Lili, Richard spent six years at Varo bank as their Chief Credit Officer, where there was the first FinTech to be come a nationally chartered bank. At Varo he helped establish their credit risk management framework, built their initial lending business and underwriting strategy and shared their credit risk committee. Richard hold both a bachelor of science and applied mathematics and statistics and a master's of science and statistics from the Stony Brook University in New York. In his past, Richard was a participating member of Experian Decision Science Advisory Board, and Equifax's Alt Data FinTech advisory board. In this episode, Richard and I will touch base on how a new lender decides on their initial underwriting strategies, the five C's of credit, the use of third party Bureau and so much more. But before we dive into the interview, please head over to our LinkedIn and Twitter pages at GDS Link, that's G D S L I N K and hit those like and follow buttons. If you haven't done so already, please subscribe to our podcast on Apple podcast, Spotify or wherever you prefer to listen to your podcast. All right, now let's get synced with GDS Link. Welcome, Richard. I hope you're having a great start to the week. Thanks for joining me today. Where are you joining us from?

Richard Weil  01:48
Hi Rich, I'm joining you from Danville, California and Danville is 30 miles east of San Francisco. But like yourself, Rich, I grew up in New York and I was born in Brooklyn. And you can probably tell from my accent.

Rich Alterman  02:02
You were most recently with Lilly Financial as their Chief Lending Officer. Can you please share some details on your role at Lili and a bit on your background before joining Lili and Varo?

Richard Weil  02:10
Absolutely. So at Lili, they offer small business banking solutions for their customers also giving them invoice and accounting software and the ability to prepare for their eventual tax filings. What was the missing link there was these young small businesses needed working capital. And the idea was that we can supply their working capital to them, which is very important for the small businesses as opposed to the Lili customers going to other lenders. And that was the goal there. Like you said, I've been in the business for well over 30 years, my career started in New York at Citibank, where I worked myself up from a credit analyst all the way to a division Credit Officer where I was responsible for all of the lending activities and credit risk management functions. I've also spent a lot of years at Wells and a lot of years at Providian Financial, in various credit risk management roles and marketing roles as well. And throughout the years, I've learned how to lend to people and learn how to lend to people in a profitable manner. And I do remember even early on in my career, when I was taught about, you know, when you're making decisions about lending, you know, think about it as if you're lending your own money, what questions would you want to have? Or what would you want to ask and taking it from that perspective, to that point of view?

Rich Alterman  03:33
Never lend money to friends, it's a lesson you can learn right?

Richard Weil  03:37
Well, unless you know them well.

Rich Alterman  03:39
Unless you know them well. Well, thanks for sharing that. And we'll certainly be getting into a lot more detail on that. But before we really dive into the business side of things, I always like to get a bit personal. I understand in talking to you that you're very involved in adult Baseball League, which was formed by members of the baseball fantasy camp. Besides the health benefit of staying in shape, can you share some other positive impacts that being a member of the league has brought to you?

Richard Weil  04:01
Getting to play baseball, I mean, this league is kids or grown adults from the ages of let's say 40 to 70, who play and it's an absolute pleasure and love to be playing the game you grew up watching and playing when you were younger. It's real baseball, the only thing we don't do is stealing because the catcher can barely make the throw to the pitcher let alone to the you know to second days. But it is just like an immense amount of joy. Every weekend we go out there we play. Of course Monday is a tough because we're all limping on Mondays. And you know, but it is just bring so much joy and so much fun, just to be able to continue to play and it just feels so good to be able to get ahead hit the ball, feel the ball. It keeps you young in many ways. It's just a great thing to look forward to every weekend.

Rich Alterman  04:49
I understand you're the one responsible for getting a leave to be one of your main sponsors, so that you wouldn't be living on Monday morning. So one of my favorite movies is Moneyball. Given your background in math and statistics, I suspect you might be a fan as well. Do you feel the movie did a good job of portraying real events? Or do you feel there was a lot of liberties taken in the movie?

Richard Weil  05:08
Well, I mean, a movie is a movie. So you know, no matter what real events are always going to make it in a way so that the people watching it are going to enjoy it. And they're going to embellish it to a certain extent. But when you're talking about Moneyball, is it basically was a movie about my two passions in life, you know, baseball and statistics. Right, right. And it gave you a different view of what's statistics that people really stuck to in it, you know, Billy Beane, who was played by Brad Pitt. In the movie, he was just very, very adamant about a process of doing it when they didn't have a lot of money. And it was a great movie. I absolutely loved it. And what better movie can there be out there? You know, that covered both of my passions.

Rich Alterman  05:48
Right? Right. That's great. So thanks for sharing that. So let's get down to business. So both Lili Financial and Varo Bank, you helped build their initial lending businesses. This included define the initial product roadmap, vendor selection, underwriting strategies, risk based pricing and loan line assignment strategies, and profitability modeling. So for much of our time, today, we're going to dive deeply into some of the fundamentals lending. So we will focus on several of the items that I just outlined. So let's start with the steps that a new lender needs to take to develop their underwriting strategy. Why don't we do this focusing on consumer lending, and let's assume we're rolling out an unsecured installment loan product. So then, now that we've selected the the product, once again, a consumer unsecured installment loan, as they're offering. How do you begin to define your underwriting strategies to open open forum, so let's kind of start walking from it.

Richard Weil  06:36
I could do this for days, and they'd be talking about the strategy in terms of playing it to, whether it's installment loans or any type of product, the first thing in the industry that most of the people are going to be listening to are going to know about, and because they've heard about this is the five C's of credit. And when you're building an underwriting strategy, you really have to think about all five, now they don't all weigh exactly the same. But I'll just quickly go through those five. The first one is what we call credit worthiness or character, right? It's whether or not the person really has the intention of repaying you or not. And that's one of the most critical ones in the unsecured area. And the second one is capacity, which measures somebody's ability to repay you, you're not making a large salary, you're not going to be able to afford a very, very large loan. The third aspect is the collateral value and in unsecured there is no collateral. But in a mortgage, there's the home and auto loan, there's the car. And there's other types of collaterals as well. The fourth C is capital, which means the amount that the customer is willing to make in terms of the vested interest in that individual. It could be a downpayment, and a mortgage or a vehicle. The fifth one is actually it's called conditions. And what it talks about is the conditions of the economy of what's going on with that particular product or the housing market, or something else that's going on. Now, I in my own right, had come up with a six C, and I've just bring that up, which is called confirmation. And the reason why I think that's very important, what it does it for every loan that you open, you have to make a decision about the level of verification that you do, whether you're verifying the income, are you going to take stated? Are you going to verify it? Right? Are you going to verify employment? Or are you going to verify any of the personal information? So those are the some of the basics that are out there.

Rich Alterman  08:39
Right. And certainly, living in a digital world, confirmations become far more critical. Right? Then maybe when that person sitting in front of you and you have a relationship with that consumer, let's kind of talk about credit worthiness. Obviously, some of this is known to everybody on the podcast today, ours, Equifax, Experian. TransUnion, has been the primary bureaus in the US. But you know, let's kind of talk about the credit worthiness and that willingness to repay, what are the tools that you use to really dive deeply into that? And how do you as you look at that person's credit profile, in particular, how do you kind of align that specifically to a product? So if we're doing an unsecured installment loan, versus let's say, a credit card, how are you maybe looking at that data from a different lens based on the product that you're going to offer?

Richard Weil  09:27
Yeah, well, again, that's a little bit of a difficulty to answer because what you're really looking at is the P&L components of a product. Different products have different P&L, right? So for instance, as an example, if you're making a mortgage, one of the critical aspects of it is your loan to value right? Because even if the customer doesn't pay you, if you're lending at 80% of the loan, loan value, then you've got 20% cushion over there, right so you're underwriting In terms of your credit worthiness may not have to be as strong as if it's in the unsecured product. So that's why it all really comes together with all of the aspects of the products that you have. But if we go to what we were talking about a little bit earlier, rich, which has to do with how do you come up with what your underwriting criteria is going to be? Now, some people, what they do is you can borrow a set of underwriting from another product that you use, hopefully, that the product is somewhat similar to the product that you're offering. Or you could come up with very limited types of underwriting so that you can learn from it. Right. But one of the ways that I've used and regulators do like this is by going out, and if you don't have the data is by purchasing what's called a retro data study from one of the credit bureaus, right, right. And what this allows you to do is, you get data from either 18 months or 24 months in history at the point when these loans were opened, and then you have an observation window, and then you at the end of the observation window, you then know exactly how they performed, right. So what you first want to do is the data that you pull, you want to make sure that first the product is aligned with the product that you're offering. You also want to make sure that the customer base is aligned with the customers that you have, for instance, at Varo, we were targeting millennials. So we had I looked at a younger population, you may also want to take into consideration whether or not there was a recession or non recession in the historical timeframes that you were looking at, in those regards. So those are all factors that you come up with. But what the game is, is basically getting the attributes and scores that you have, from when these accounts were open, which is a retro data set, it's not your customers, it's it's a generic set of customers, and you try to make it look the sample look like the customers that you're going to be targeting going forward. And what the game is here is you can set up with some what's called non negotiables of discrete criteria criteria that you do, if they have this, then you will decline for that. Or you can use a combination of non negotiables. And scores as well, generic scores right off the bat. Right. now, again, what you're trying to do is, you start off with a sample, as you start applying criteria, your bad rates are going to decline. But at the same time, your approval rates are also declining. Yeah, so what you're trying to do is the game is to lower your bed rates as much as possible, by keeping up the approval rates as high as possible in that, and if you have a good credit analyst, they can work with you in terms of playing that game. And then at the end of the day, you can see that you've lowered from the sample that you have, how much you've lowered the bad rates. And the question is, what percentage of the population do you still have available? So that is kind of a very solid way of creating underwriting criteria to solve the credit worthiness aspect of it.

Rich Alterman  13:16
So essentially, I mean, as you mentioned, you're starting a bespoke business, you don't have any or I should say, a greenfield business, you don't have any existing clients. So you're really trying to create proxies, if you will, by saying what is the market we want to go after? And now let me go by that credit file, that retro study where it's not on consumers by have a relationship with yet but rather people I think, look the same as those people that I want to target. You mentioned, understanding what may have occurred during the time that you're doing your look back period, certainly COVID had a big impact on credit reports, and with deferments and forbearance. How do you you know, if you're sitting there today, and I know the Bureau's came out with their COVID attributes, GDS offers a set of normalized attributes, we came out with our COVID attributes. Maybe it's kind of talking about how you adjust for that that event, or economic downturn? How do you kind of sift through that?

Richard Weil  14:11
Yeah, I think what I would want to do is go back to the period of maybe a little bit before then or after that point in time, because I do think that some of the data is going to be, you know, cloudy, because people have remember with all the forbearance and with all the government checks, and that were handed out at the time, people's savings were higher than than other points in our credit history. So it may be a little bit different in terms of what's going to be in the future. Remember, everything that we do in credit is we're assuming that what we recognize in the past is going to repeat itself in the future. Right. And when that's not the case, as such in the COVID timeframe, then there may have to be some adjustments that you you would have to do either So you may want to in selecting your sample, select example, you before it, or after that period of time.

Rich Alterman  15:07
You talked a little bit about the marriage between a score and policy rules. So maybe you're buying a VantageScore or FICO score, you have a custom model, you set some cut offs, but then kind of talk about where you start injecting policy rules on top of the score, if you can.

Richard Weil  15:24
Yeah, well, I think what you're referring to is what I was calling before discrete criteria or non negotiables. Okay, right. So that is part of the analysis that you're doing when you're looking at the bare rate of your sample that you have. And, you know, for instance, you it could be a combination of, or it could be just a safety and soundness thing. I mean, do you really want to have in your sample customers who just recently charged off with other lenders, right, as an example? I mean, maybe the data is not that bad. For those for those populations. It may be small. But if a regulator is looking at that, right, what are they going to say? So you have to think about how, from a safety and soundness perspective, does it make sense to include in your sample, people who have had some very large negative hits in their major derogs, bankruptcies, charge offs, things of those nature that you may want to have what I call non negotiables, which means you have these customers automatically declined out of your pool of populations.

Rich Alterman  16:31
Now, when we look at performance, I worked for Teletrack, prior to joining GDS, and now I'm by Equifax. And it was the first payday loan Bureau and performance was something you saw very quickly, right? You made that loan, and first payment defaults, you know, how they performing. Whereas maybe with other products, it takes a little longer to really draw conclusions about how this person is going to pay. So when we, when we talk about observation periods, we were talking more looking back. But when we think about observation periods looking forward, you know how based on the product that you're offering? Do you kind of start to bring behavioral data into play in your go forward decisions? So I booked that account. Now I want to manage that relationship moving forward?

Richard Weil  17:18
Yeah, very good question. So Rich, what I've done in the past, when I was at Providian, we were mailing 10 million pieces of mail a month, right, so our sample size is very large. So I could tell within three months on a GoPhone, whether or not that strategy was working or not, it gave me a really good, you know, looking at first payment defaults, and, and things of that nature. But you know, when I've been with smaller size solicitations, sometimes I needed nine months, or 12 months to really tell, because what I saw at three months could somewhat change now, though, all you know, kind of consistent on average, but you can't be very sure, or if your sample size is small. So it really depends on your sample size. But like I was mentioning, I could tell almost by the third vintage month, right, how things were going to perform when my sample sizes were very, very large. And that I don't know if I would solidify a decision in three months. But I mean, directionally, it was telling me what I was looking for at that point.

Rich Alterman  18:17
Now, I think one of the interesting balances is kind of that relationship between your approval rates, and your acceptable bad rates, right. And any lender that started in business have to recognize that initially, you're probably gonna have loans that go bad, right? And in fact, if you don't have any loans that go bad, you're probably out of business, because your approval rates are so tight, that you just don't have a sustainable business. How often should you feel entities really be, you know, kind of evaluating how things are performing. And you know, what type of adjustments and last thing I'll add to that is around AB testing. You know, at GDS, when we're talking to new prospects, we really talked about how easy it is in our software to roll out champion challenger strategies. How important do you feel it is to have that kind of AB testing built in and actually use it?

Richard Weil  19:05
Well, I think the underwriting strategy is an ongoing process. It's a constant feedback loop. Now, you can either test it in an AB setting, or you could do it in other ways, as well. But that whole aspect of looking at your underwriting strategy, see how your customers are performing iterating it is something that just never really should stop. It's not like, you know, you do it every six months, you do it every three months, you should always be thinking about how do you improve on it, and the things changing? It's not like you don't just set your underwriting, it's never perfect. You know, there's too many exogenous attributes that are impacting the performance of your portfolio, the economy in variety of other things as well, competition, what other people are putting out there. So you constantly have to be looking at ways to alternate data sources are being brought in. So it's real It's an ongoing job that never stops. And you really just need to have your hands eyes and everything else in ears open in terms of looking at things to improve things, because what you'll also want to do is you don't want to be charging to higher interest rates and fees, and so on and so forth. Because the higher you go, if your pricing higher than your competition, we're gonna get what's called the adverse selection, which is certain something that, you know, no lender wants as well. So it's an ongoing game. And it's a fun game for somebody who really loves data to be quite honest.

Rich Alterman  20:34
And you talked about the five C's, we'll actually be talking about a six C, and I like to throw one in there. And I mean, call it my you as an umbrella. And I think about the utility of the product. And I'm thinking about this, I often think about this from a collection perspective. And I say to myself, if you're collecting on different types of products, the utility of that product is going to come into play, when that consumer is being forced to make a decision as to which thing I paid today. Okay, and in our example, today, we started out by talking about an unsecured installment loan, well, once I get that loan, the utility of that installment loan is really not high compared to pay my mortgage payment, pay my car loan, pay my credit card, which has ongoing utility. So how does does that you or new you, kind of play into some of your credit policy. And thinking about not today but down the road when if that person does go delinquent, and he or she is having to make that choice, that tough choice about which accounts do I pay? If that product has a lower utility than let's say a credit card? How do you kind of bake that into your thought process around credit strategy upfront?

Richard Weil  21:45
Utility is important. And just as an example, my son who just recently took out a credit card, he needed to rent the car, and without a credit card, he couldn't rent the car. So that's utility, you needed to have the credit card to be able to rent the car when he went on a trip, right? As an example. So there's value there that a person wants to have. So what you're you're alluding to is, if if my son wants to go back out and rent another car, he's going to make sure that the payment on that credit card is going to be the same thing as living in a house or making the payment on your auto loan as well. Right? Well, what's interesting thing in the 2008 recession, one of the things that we noticed about the hierarchy of payments is that people were paying their auto loans before their mortgages. Right. So you utility matters? I don't know Rich if I can answer your question exactly. Except that I can tell you for a fact that utility matters. And even though you're right on an installment loan, where you have lower level of utility, the fact is, if they don't make payments, and you're reporting it to the credit bureau that they're not, then you're hurting that individual in the future. So you know, you have a carrot, and you have a stick, and you've got to learn to use both.

Rich Alterman  22:59
We talked about the retro study, started alluding to other data sources. So you know, on a digital lending, clearly, customer not present, fraud has become a lot bigger issue. And a lot of time and effort goes into making sure you have a good broad strategy that can look at first party third party synthetic. So from a testing standpoint, a lot of the fraud solutions do not offer the ability to do retros. Everything has to be more on a forward flow perspective. So can you kind of talk about as you're assessing that initial strategy, we've talked about the credit side of it, but now let's kind of focus a little bit on the fraud ID, is it just knowing best in class products? And these are the ones we need to use? But how do you incorporate some of that thought process into your strategy?

Richard Weil  23:42
Yeah, so the way I've done a lot of the fraud testing has exactly been that which is like a forward looking process, right. And you can do it within a relatively short time as well, right by you can take a snapshot of your portfolio today, you can score them, I don't want to mention any particular vendor right now, because I wouldn't name others. But you can take vendors data, you can score them today. And then you can look at him in three or six months from now and find out which one really gives you the lift that you were looking for. And there are some very good ones that are out there. I know that's not what this hour is about. But the forward looking process is very good way of doing it. And you can get some very quick results as well, sometimes synthetic frauds, they let it sit for a while. So synthetic fraud sometimes, you know in a short window, you have to let it sit for a period of time in order to determine if you have any, but I'm very well aware of it. I know my first charge off at Varro was an old woman from Florida who never even heard of Varo, right. So I understand the importance there of checking the identity of the individual and coming up with tools in order to combat fraud. And the other thing I just want to say Rich about the fraudsters and is that they are so much more sophisticated today than where they were in the past. There are sometimes using AI themselves, right in order to come up with causing fraud. So they're a lot smarter these days. So we have to be smarter as well, in terms of how do we mitigate the fraud risk.

Rich Alterman  25:11
In our last podcast with John Baird over at Vouched, we were talking about the deep fakes and some of the pretty scary things that are going to take place using AI. And, you know, how do you combat what's really become a war, and you have, you know, trying to stay ahead of the curve. So, the second C you talked about was capacity. And certainly over the last couple of years, we've seen a large explosion of the use of open banking data. So companies like Plaid and I'll name a few Plaid, yo li, DecisionLogics, MX, you know, the list is getting quite long. But there too you don't necessarily have that ability to do any type of look back retro today, but is the use of open banking data, something that you've incorporated into your prior roles.

Richard Weil  25:58
I have been using deposit data going back into the late 80s. When I was at Citibank, I moved from the credit card business to the consumer banking group, where we did all credits except for mortgage and credit cards, because they had their own businesses. And, you know, I was young and I was starting off and I had credit criteria in place. And what I did was, I put together, I thought of two things that I thought would be valuable, which was how much money they had in the bank, and how long they've been a customer. And I made a very simple three by three grid, right, so I had nine scores one through nine, that one through nine turned out to be more predictive than the customer's credit score that we had. Right. And I didn't really put any time and effort I just really looked at different, you know, making sure that each of the three groups on the two variables, age and balance, had reasonable sample sizes in it. And that was highly predictive. So I've always known that deposit data has the ability to predict lending performance going forward in that regard. Now, in today's world, what that deposit data can do, when you look at evaluating the capacity to play, what most companies do is they look at what's called the debt to income calculation, where your numerator is your monthly expenses. And your denominator is your monthly income, it could be gross, it could be net, depending on how you look at it, what the deposit data helps you do, it helps you put together a more confident outlook of what that DTI tax debt to income calculation actually is. So it really helps with that. Now, the other thing is that there are other besides just looking at the payment structure, there are also components of the deposit data that can help you in your credit worthiness aspects as well. And you know, I was at I've saw that at at Wells after I left city. And I think it's very, very strong predictor of performance in many different ways.

Rich Alterman  28:01
So you mentioned that you kind of built this simple three by three scorecard, if you will. And we talked earlier about using retro studies to really look at proxies of people that I want to lend to. So when we talk about models, custom models, obviously, you have the scores that are available from the bureaus. But can you talk a little bit about the use of what the industry often refers to as either Quickstart? Or Fast Start models that are industry specific to kind of get the ball rolling when you don't have any experience?

Richard Weil  28:29
Yeah, I think all of these models this day and age is by whatever company it is, and even looking at, you know, the FICO score and the VantageScore, there are all going to be predictive? The question is, of course, when I build custom models, and I've built many of them in my time, they always outperform the generic. But if you're not going to have a custom model, if you don't have data to start with, so those generic models are always going to perform fine. The other thing that the generic models do, that's very important, it gives you a commonality or a language to talk about. I mean, if I had a custom model, it may range from one to 100, not 300 to 800. Right? So you can't compare very easily. If I told you that somebody scored a 700, FICO, a 700 Vantage, you have a rough idea of what loss, you can expect it. So that gives everybody a common language out there. But going back to your question about custom models, you just need to have the time to let the accounts mature, and have a very good sample size in it as well so that you can build a model that's going to be very robust and lasts for a period of time.

Rich Alterman  29:39
So we've approved the loan, and now we're trying to decide how much to lend and profitability. So kind of walk us through that exercise where you have a range of a loan from let's say, 2500 to 25,000. How do you decide who gets what?

Richard Weil  29:56
So this is an incredibly important aspect of the credit underwriting decision process. Your underwriting, whether you approve somebody or not, is really going to control what's called the unit loss rate. So if you opened up 100 loans, and five of them went bad, then you have a 5% unit loss rate. But what the line assignment amount or the loan assignment line alone is basically going to do it helps you look at how the dollar loss rate as a ratio of the unit loss rate is going to come out. That's called what I've called the balance control factor. Okay. So what you basically want is be able to give the better customers that your have larger size loans and the customers that you're approving, that are not as credit worthy to get the lower size loans, what that will do is give you a lower dollar loss rate than your unit loss rate. And the ratio of the two is what's called the balance control factor. So how do you come up with this? Well, again, at Providian, where we had very, very large samples, I had a line assignment matrix, it was 1500 cells. Now, that's probably a lot, you know, I had to come up with, but we had an enormous amount of data. And in order to do that, you really want to put together aspects, the line assignment matrix should be a combination of risk components, income components, and also expectation of use. So for instance, what I mean by that is, if you have a super Prime customer, but they very, very rarely use credit, do you really want to give them a $30,000 line, if they're, if they've never used more than 1000 in history, you could argue, well, you may not be heard, but you don't know which customer that rainy day comes up, where they have some type of life event, and they're going to take you for the full loss. So there's those components of what you can put together in terms of your strategy. And then you can even do it in the sample that you have that you build your underwriting strategy, you can put them through your line assignment strategy. And see you can compare what your bed rate your dollar bed rate is to your unit dad rate to see whether or not your line assignment is being effective, even prior to rolling out your product.

Rich Alterman  32:11
I'm keeping an eye on the time there's a couple more topics I definitely want to get to and like you said we could probably spend hours or days or weeks and we could probably write a book. Oh, wait, I've already interviewed the guy that wrote the book. So wrapping up this chapter here, you know, you've been in the business now as we've we've emphasized over 30 years. And if you had to really think about and I often preach that if we think about underwriting at a core, things really have not changed that much over the last 30 years, right? I'm I'm trying to make smart decisions about who to lend money to and get paid back. So when you think about and looking back over your last 30 years, what would you say have been really some of the most fundamental changes in our industry?

Richard Weil  32:50
Well, one of them is what we already talked about the sophistication of the fraudsters, I mean, our fraud rates used to be in the two digit basis points, 10 basis points, 20 basis points of fraud, you know, that is certainly picked up. And like I said, they become very sophisticated these days. So that's a huge change from when I first started. Also, you know, the model risk management requirements today are much more stringent than what it's been I mean, we even go back to my early days at Citi, we did all the X's and O's in terms of building some of those models and tracking the performance of them, because we had a group credit committee there that we were responsible to report on them periodically on that. But just looking at today's OCC 2011. There's 12 bulletin, that 21 page document has a lot in it that they required today, for every credit business in terms of managing those models. So that is really picked up a lot. And I think from a regulatory perspective, there's a lot more oversight on, you know, the fair lending concerns that people have some of the new data that you have, you have to be very careful about what you're using. I mean, a lot of the attributes, you got to really ask yourself a question. And I always try to ask myself a question. If I declined somebody for using that attribute. Can I explain it to an individual? And can I explain it to a regulator? That's the litmus test for me that I have when I'm using attributes or newer attributes in the model doesn't make sense? And do I feel like I've got a very good reason for using that attribute?

Rich Alterman  34:24
Yeah. And one of our podcasts was actually with Fairplay that's all around taking a look at your, your scorecard models, right to see if there is some disparate impact some bias built into the model, even though there was no intent to do that. But there you are, that that you have some issues, in fact, years ago, right. The big issue around machine learning and AI was the inability to generate that adverse action status. And I remember going back almost over 30 years, and a company who was into neural nets I'm not going to mention the name but had decided to try to move into credit right, but the issue was Like, why did I actually decline this person today, when I might have approved them an hour later, right? So we've gotten a lot better and the ability to deliver those adverse action reasons. And obviously, that's, that's critical today. So I'm not going to kind of dive into Varo's journey to become a bank and reading an article, I understand that it took place in July 2020. And based on an article, I read it, it costs almost $100 million for Varo to become a Neobank. And you know, when you look back on that journey, what were some of the lessons that you learned from that? Where if you were consulting another company, who might be considering becoming a bank, what are some of the things that they really need to think through? And how has what's occurred in the last several months, because of what happened with Silicon Valley Bank? How would that play into your decision today?

Richard Weil  35:52
So let me try to break that apart a little bit, the process was phenomenal in terms of learning what it's like to become a bank, we had a lot of very smart, talented people and Varo. We had a lot of experience, there was a financial plan put together where a lot of people had input in privacy had input on some of the credit products, in terms of cross selling, and the acquisitions of it and the performance of it. But what it really came down to was making sure that we had really sound credit risk management processes in place and the framework in place. And it wasn't just credit risk, which I was involved with, it had to do with all of the aspects of risk, all of the different components of risk, info security, you know, BSA, AML, operations, risk compliance, third party risk, model risk governance, and it also made sure that we had all of our policies well written, all standards in place, and our procedures also in place, all of those things were gone through with a fine tooth comb, as well. And there were two phases of the process as well. One was in the pre approval, which was in the early stages. And then one was a year and a half later, in the final stage right before we were granted, you know, the full approval in that regard. And I know that in the pre approve stage, the OCC and the FDIC had me in a room for three and a half hours, where they grilled me from A through Z in terms of how I put together the underwriting strategy, I actually walked them through the retro data study, and how I put everything together. And you know, I knew in my mind that someday I would probably be having to do that. So I put together a lot of the strategy with the thought that I would have to defend it to people, right. And, you know, look, these government regulators, they know what they're talking about. They've been through this before, and they have opinions about things. But I must say that I thought it went very well, in terms of the review and going through and gave the regulator's the confidence that they had somebody in myself who had the experience in order to put together something like this, and also be able to track them forward. So that was the pre stage, the final stage was a little bit different, because we mature it a year and a half down the road. But I'll leave it at that for now.

Rich Alterman  38:10
Yeah, and this is something you know, when you look at startups that are getting into lending space, and the criticality of having good credit policy people that are going to give the investors confidence that you really know what you're doing, and that you're going to pass scrutiny is critical. So we're coming up near the top of the hour here. So I'm going to throw out a question for you to think about and give us an answer. So a little roleplay. Here, I'm, let's say I'm a someone who's getting ready to go to college, and I love math. I love statistics. And I've been reading up a lot about Chat GPT, and I'm concerned that Chat GPT is going to maybe adversely impact my opportunities moving forward. If I go get a math and statistics degree. What would you say to that person given your current knowledge and thoughts around Chat GPT? And I'll leave it there. And I'll see if I have a follow up.

Richard Weil  39:02
There's so much that we can predict about the future, right? We don't have a crystal ball in front of us to know really what Chat GPT is going to be doing or how it's going to be changing the industry. But I'll tell you one thing, people have good solid math skills are going to be needed somewhere in the world in the world of banking, and the world of finance, in the world of Biostatistics and things of that nature. So if you have a good math background, there's so many things that I believe and in engineering as well, I wouldn't worry about it. You can't worry about the unknown about what's going to happen so far down the road and predicting things, I think, but going you know, look, going to college, having fun, learning to become an adult and enjoying college and getting walking away there with a degree of solid degree and understanding various math skills is going to put that person in the driver's seat for many things. So I wouldn't worry about things that are out of your control. And I just think you just go forward and do the best you can.

Rich Alterman  40:04
Yeah, I'll echo what you said, I tell any of my friends that have children that like math. I say if you look at one skill that is ubiquitous across every single industry, it's math and statistics, right? Everyone uses analytics. It doesn't matter what industry you're in. And to your point to kind of look at it generically. Can I step into a role at an insurance company versus a lending company versus a medical group looking at outcomes for cancer treatment? Right, there's really no limit to how math and statistics get used. Well, thanks, Richard. This is Rich Alterman. We've been syncing up with Richard Weil, a 30-year veteran in the credit risk management area. Thanks for tuning in with us. We hope you can leverage some of what you learn today in your day-to-day roles. And please stay connected with GDS Link and The Lending Link to listen to future podcasts and catch up with the ones you've missed. Thank you and make it a great day. Thanks for listening. If you've enjoyed today's episode, please be sure to subscribe on Apple, Spotify, Google, or wherever you listen to your podcast. To be sure to leave us a review. Follow us on LinkedIn and connect with us on Twitter at GDS Link that's at G D S L I N K. Have a question for the show or have a specific topic you want us to cover? Hit the link in the description to drop us a note. Thank you for lending us part of your day. Make it a great one.