WEBVTT - Why Americans Are Falling Behind on Auto Loans At Their Highest Level Ever 0:00:02.720 --> 0:00:19.599 Bloomberg Audio Studios, Podcasts, Radio News. Hello and welcome to 0:00:19.680 --> 0:00:23.320 another episode of The Odd Lads podcast. I'm Joe Wisenthal 0:00:23.440 --> 0:00:25.840 and I'm Tracy Alloway. Crazy. I feel like I just 0:00:26.440 --> 0:00:30.000 do not have any feel right now on like the 0:00:30.120 --> 0:00:31.760 state of the consumer. 0:00:31.800 --> 0:00:32.040 Really. 0:00:32.080 --> 0:00:36.560 I mean, you hear k shaped economy, labor markets slowing down, 0:00:36.960 --> 0:00:40.120 then it's like lowest layoffs in years. You go outside, 0:00:40.280 --> 0:00:43.040 everything looks booming. Like, I just have no feel right. 0:00:43.159 --> 0:00:46.199 I know, consumer sentiment is terrible, but consumer sentiment is 0:00:46.320 --> 0:00:48.919 terrible for years, and people keep shopping. I have no 0:00:49.000 --> 0:00:49.640 sense of it right. 0:00:49.720 --> 0:00:53.600 Well, consumer sentiment actually came in higher than expected most recently, 0:00:54.080 --> 0:00:56.360 big surprise, but I was gonna say, are you not 0:00:56.560 --> 0:00:59.000 out shopping for Christmas present? 0:00:59.080 --> 0:00:59.680 It's insane. 0:00:59.760 --> 0:01:01.680 Yeah, there's a lot of people buying a. 0:01:01.600 --> 0:01:03.400 Lot of stuff, buying a lot of stuff. 0:01:03.440 --> 0:01:06.600 But I think this gets to the ke shaped economy point, 0:01:06.640 --> 0:01:10.000 which is, if you have a cohort of wealthy people 0:01:10.120 --> 0:01:14.440 who are buying more, it more than offsets the lower 0:01:14.480 --> 0:01:18.120 income people who are buying less at lower prices. So 0:01:18.200 --> 0:01:19.559 it's really hard to tell. 0:01:19.800 --> 0:01:22.959 It's really hard to tell. One thing that definitely feels 0:01:23.040 --> 0:01:25.920 different if you look at aggregate measures of household balance 0:01:25.959 --> 0:01:29.959 sheets like this is something that is very different than 0:01:30.080 --> 0:01:33.280 sort of like pre grade financial crisis. The general view 0:01:33.840 --> 0:01:37.759 is that the American consumer or the American household has 0:01:37.800 --> 0:01:40.119 a very big cushion. There is a lot of home 0:01:40.160 --> 0:01:44.520 equity built up. There is not a thin layer. Obviously, 0:01:44.600 --> 0:01:47.840 anyone with money and any sort of investment account has 0:01:47.880 --> 0:01:52.080 done phenomenally well. We're according to this December twelfth, yesterday, 0:01:52.080 --> 0:01:53.880 I think the s P five hundred hit yet a 0:01:53.960 --> 0:01:56.240 new all time highest. If you have any sort of 0:01:56.280 --> 0:01:59.040 home equity build up, if you have any sort of investments, 0:01:59.320 --> 0:02:01.800 you are doing very well. On the other hand, of course, 0:02:01.800 --> 0:02:04.400 people are stretched from years of inflation. We know that 0:02:04.560 --> 0:02:06.320 hiring has slowed down. 0:02:06.400 --> 0:02:06.920 We know that. 0:02:07.120 --> 0:02:10.840 You know, we see these headlines delinquencies for cars have 0:02:10.960 --> 0:02:13.079 like shot up. But I've been seeing these headlines for years. 0:02:13.080 --> 0:02:15.120 I don't totally know what they mean or how apples 0:02:15.120 --> 0:02:17.760 to apples they are with the past. I just don't know. 0:02:17.800 --> 0:02:19.280 I just I'm very confused. 0:02:19.360 --> 0:02:21.720 Yeah, you know what's really interesting to me just from 0:02:21.720 --> 0:02:24.920 a financial perspective. Yeah, if you look at some of 0:02:24.960 --> 0:02:29.920 the bonds that were actually built on consumer loans, the 0:02:29.960 --> 0:02:33.960 weakest ones are now from the like twenty twenty to 0:02:34.080 --> 0:02:35.360 twenty twenty two period. 0:02:35.639 --> 0:02:40.800 Oh see, this is another interesting element of measures like delinquencies, 0:02:40.840 --> 0:02:43.360 and why I sort of wonder like how comparable they 0:02:43.360 --> 0:02:47.720 are because okay, partly a delinquency measure is a snapshot 0:02:47.720 --> 0:02:49.480 of a moment in time, right, a snapshot of health, 0:02:49.560 --> 0:02:53.639 but it also inherently reflects something in the past, because 0:02:53.639 --> 0:02:56.799 it reflects, you know, what we're lending standards at the time, 0:02:56.960 --> 0:02:59.720 right exactly, so you know, and now it's a period 0:02:59.760 --> 0:03:04.880 of interest rate booming itself. Yea, give the money to 0:03:04.919 --> 0:03:08.040 anyone anyway. We need to get a better picture of 0:03:08.160 --> 0:03:11.320 exactly what's going on. How stressed is the consumer. How 0:03:11.400 --> 0:03:15.400 much do these delinquencies just reflect the profligacy of lenders 0:03:15.480 --> 0:03:18.520 during the boon times when rates were nothing, et cetera. 0:03:19.280 --> 0:03:21.680 And yes, we need to figure this out, especially we're 0:03:21.680 --> 0:03:24.080 in the middle of shopping season and all that stuff. 0:03:24.120 --> 0:03:25.800 So I'm very excited to say, we really do have 0:03:25.800 --> 0:03:28.440 the perfect guests. We're going to be speaking with Recard Bondebo. 0:03:28.480 --> 0:03:31.600 He is the executive vice president, chief strategy officer, chief 0:03:31.639 --> 0:03:35.960 economist Advantage Score, a credit scoring company. Recurd Thank you 0:03:36.040 --> 0:03:37.360 so much for coming on the podcast. 0:03:37.480 --> 0:03:38.640 Thank you for Havmius Snana. 0:03:38.920 --> 0:03:42.360 What is Vantage Score a US credit scoring company? What 0:03:42.400 --> 0:03:42.960 do you do there? 0:03:43.280 --> 0:03:45.520 So, we're the largest credit scoring company in the United 0:03:45.520 --> 0:03:48.480 States and we have founded almost twenty years ago by 0:03:48.520 --> 0:03:52.360 the Free Credit bureas, TransUnion, ECOFAX and Experience, and we 0:03:52.400 --> 0:03:54.800 were sort of created with a very specific mission in 0:03:54.840 --> 0:03:58.520 mind to drive greater competition and credit scoring. Prior to us, 0:03:58.560 --> 0:04:01.040 there really wasn't a lot of choice in space. We're 0:04:01.080 --> 0:04:03.080 also there to drive more innovation and create the most 0:04:03.080 --> 0:04:06.000 predictive scores, which was a big ass from the banks 0:04:06.000 --> 0:04:08.080 at the time, and also to be able to expand 0:04:08.120 --> 0:04:11.120 access to millions to enable everyone who really is credit 0:04:11.200 --> 0:04:13.880 worthy to be able to get access to credit products. 0:04:14.000 --> 0:04:16.400 You mentioned the banks just then, can you expound a 0:04:16.440 --> 0:04:18.200 little bit more on your customer base. 0:04:18.640 --> 0:04:22.120 Yes, so we used Obviously, the primary use case that 0:04:22.160 --> 0:04:24.159 most people think about when it comes to credit scores 0:04:24.520 --> 0:04:26.839 is for lending. Right when you know you're applying for 0:04:26.839 --> 0:04:28.520 a loan and they want to evaluate whether or not 0:04:28.760 --> 0:04:30.200 you're going to be able to perform on that loan, 0:04:30.240 --> 0:04:32.200 they will often pull your credit score as part of 0:04:32.200 --> 0:04:35.240 that process. But It's used in many other stages as well. So, 0:04:35.279 --> 0:04:38.120 for instance, many people who are applying to rent in 0:04:38.160 --> 0:04:40.240 a new apartment building may also get asked for it. 0:04:40.360 --> 0:04:42.680 When you are trying to get a utility. 0:04:42.320 --> 0:04:44.560 Bill in New York, you get asked for it. 0:04:44.600 --> 0:04:48.520 Yes, you certainly do. And utility bills, telephones, anything that 0:04:48.560 --> 0:04:51.800 involves a long term commitment on payments generally. Now you'll 0:04:51.800 --> 0:04:53.640 often get asked to provide your credit score. 0:04:54.000 --> 0:04:56.320 So just explain for the way you were started by 0:04:56.360 --> 0:04:58.240 whom twenty years ago, we. 0:04:58.120 --> 0:05:01.920 Were a joint venture and ECHOFACTX and TransUnion, the three 0:05:01.960 --> 0:05:03.599 national credit reporting agencies. 0:05:03.680 --> 0:05:07.320 So what is the difference between these major companies that 0:05:07.360 --> 0:05:10.799 we've all heard of that provide a credit score, et cetera, 0:05:11.120 --> 0:05:13.440 that they such founded? You like, what do you do 0:05:13.520 --> 0:05:14.279 differently than them? 0:05:14.560 --> 0:05:16.880 Well, so what they do is they're the ones who 0:05:17.040 --> 0:05:20.520 collect all this data from lenders and others on your 0:05:20.520 --> 0:05:23.479 credit performance. Right, So they call credit buros they collect that. 0:05:23.520 --> 0:05:27.520 They're highly regulated. But then most lenders can't just make 0:05:27.560 --> 0:05:29.560 sense of all of that data on its own. They 0:05:29.600 --> 0:05:31.840 need some guidance to have to translate that into a 0:05:31.920 --> 0:05:33.120 what does that mean? Right? 0:05:33.160 --> 0:05:33.400 Okay? 0:05:33.480 --> 0:05:36.680 And so that's where a scoring algorithm comes into effect. Right, 0:05:36.960 --> 0:05:39.280 And so the scoring algorithm helps to take in all 0:05:39.320 --> 0:05:42.000 these hundreds of different factors about you to try to 0:05:42.040 --> 0:05:45.080 then determine what does that mean about your propensity to pay? 0:05:45.360 --> 0:05:49.640 Okay, you mentioned predictive analysis as well, What exactly is 0:05:49.680 --> 0:05:50.680 that and what's that based on? 0:05:51.040 --> 0:05:54.080 Well, so when credit scores are created, right, the aim 0:05:54.440 --> 0:05:56.680 the goal is to try to evaluate what is the 0:05:56.800 --> 0:05:59.560 likelihood that somebody's going to default on a payment over 0:05:59.600 --> 0:06:02.040 the next time twenty four months. So when you see 0:06:02.360 --> 0:06:04.760 that score, the score is actually a translation of a 0:06:04.800 --> 0:06:09.279 probability right or on odds right to evaluate what is 0:06:09.320 --> 0:06:09.800 that risk? 0:06:10.279 --> 0:06:12.560 And how did we end up with the system of 0:06:12.839 --> 0:06:15.520 FICO scores in the US, because it has like an 0:06:15.640 --> 0:06:16.679 interesting history. 0:06:17.000 --> 0:06:20.279 Well, back in the day fair Isaac they created the 0:06:20.279 --> 0:06:22.200 first sort of known credit score. They were the first 0:06:22.200 --> 0:06:24.920 ones to realize that there was a. 0:06:24.640 --> 0:06:28.400 Looking at fi and fair Isaac is that yes? 0:06:29.040 --> 0:06:31.960 Okay, yes, And so let's go back a bit, right, 0:06:32.040 --> 0:06:35.640 So in the old days, lending was not necessarily the 0:06:35.640 --> 0:06:38.240 most fair system that there was, right, or want to 0:06:38.279 --> 0:06:41.080 call up your previous employer, They may call your landlord. 0:06:41.120 --> 0:06:42.600 They may just ask around and if they don't know 0:06:42.600 --> 0:06:44.560 anything about you. You know, there was a lot of 0:06:45.120 --> 0:06:46.280 judgment involved in. 0:06:46.240 --> 0:06:48.039 Lending, a lot of racial discrimination. 0:06:48.440 --> 0:06:51.000 Well that certainly was built into that system, right, And 0:06:51.080 --> 0:06:53.640 so then there was a law created, the Fair Credit 0:06:53.680 --> 0:06:56.039 Reporting Act, that said, like, you can't do that, you 0:06:56.120 --> 0:06:58.880 need a better system that is fair and that is 0:06:59.000 --> 0:07:02.760 a better quantitative ability to assess people's risk, right, And 0:07:02.800 --> 0:07:05.800 that created then this need to be able to consolidate 0:07:05.800 --> 0:07:08.200 all this quantitative information in a way that lenders could 0:07:08.200 --> 0:07:10.600 easily use it. So Faiko was the first fair ISAAC 0:07:10.760 --> 0:07:12.680 at the time was the first to create that and 0:07:12.760 --> 0:07:15.280 they did very well doing so. But you know, then 0:07:15.320 --> 0:07:19.160 there was a need for competition innovation, and there was 0:07:19.160 --> 0:07:21.120 a lot of frustration around the time of twenty years 0:07:21.160 --> 0:07:23.360 ago that there was only one game in town and 0:07:23.800 --> 0:07:26.160 it didn't score about twenty percent of the US population. 0:07:26.240 --> 0:07:28.120 It still doesn't, And then a lot of lenders were 0:07:28.120 --> 0:07:30.200 felt frustrated, like if it doesn't work for twenty percent 0:07:30.280 --> 0:07:33.120 of the population, there's a problem. We need something different. 0:07:33.160 --> 0:07:36.200 And so then buwers took the unusual step of actually 0:07:36.240 --> 0:07:39.520 coming together to create an alternative and that became Advantage School. 0:07:39.720 --> 0:07:43.600 Interesting can I as a consumer go credit score shopping. 0:07:44.160 --> 0:07:46.920 So first of all, there's different ways to use it, right, 0:07:47.000 --> 0:07:50.280 So a lender will typically choose the credit score that 0:07:50.320 --> 0:07:53.560 they're going to use for being able to underwrite alone 0:07:53.560 --> 0:07:56.040 with you, and often they'll use many more factors than 0:07:56.120 --> 0:07:59.640 just a simple credit score, right, particularly the more sophisticated ones. However, 0:07:59.680 --> 0:08:02.520 when you're trying to understand what your situation is, you 0:08:02.520 --> 0:08:04.080 can there are lots of different places you can go. 0:08:04.120 --> 0:08:05.640 You can either go the credit bureaus, you can go 0:08:05.640 --> 0:08:07.600 to the likes of credit Karma. There are many different 0:08:07.640 --> 0:08:08.560 services out there. 0:08:08.440 --> 0:08:10.560 But I can course the lender to look at a 0:08:10.600 --> 0:08:13.880 specific Yeah, look at this one over here, it's great, 0:08:13.920 --> 0:08:14.800 get a second opinion. 0:08:15.080 --> 0:08:17.280 No, I'm afraid not. That's not how it works. It's 0:08:17.360 --> 0:08:19.360 it's really you know, the lenders try to determine what 0:08:19.440 --> 0:08:21.520 is the most appropriate score for their product. And there 0:08:21.520 --> 0:08:23.480 are many many different schools out there. There's schools that 0:08:23.960 --> 0:08:26.800 in some cases built specifically for types of products like 0:08:27.000 --> 0:08:29.440 order loans, and the other schools that, like our schools, 0:08:29.440 --> 0:08:31.120 that are generic that can be used for any type 0:08:31.120 --> 0:08:31.640 of product. 0:08:31.920 --> 0:08:34.719 So you collect more data. And you mentioned that there 0:08:34.760 --> 0:08:37.640 is this wide swath of the population that wasn't being 0:08:37.679 --> 0:08:41.479 captured by the credit bureaus? What do you do additionally 0:08:41.520 --> 0:08:45.200 on top of them to expand the pool find potentially 0:08:45.240 --> 0:08:48.160 credit worthy borrowers that they had been missing before. 0:08:49.000 --> 0:08:52.800 So the thing is that the quality and the types 0:08:52.840 --> 0:08:54.959 of data that's been collected by the credit bureaus has 0:08:55.000 --> 0:08:59.080 improved significantly over time. OKA, And so when we started 0:08:59.080 --> 0:09:01.360 creating our algorithms, we're in, you know, the current version 0:09:01.400 --> 0:09:03.679 that's now being adopted for mortgages, the version four we're 0:09:03.720 --> 0:09:05.320 releasing version five this year. 0:09:05.800 --> 0:09:06.439 We actually go. 0:09:06.440 --> 0:09:09.080 And rewrite the whole thing each time so that each 0:09:09.120 --> 0:09:11.080 time we can come up with the most accurate way 0:09:11.240 --> 0:09:14.360 based on the current data is available and our current 0:09:14.400 --> 0:09:17.679 ability to understand how consumers are behaving, because that behavior 0:09:17.760 --> 0:09:20.720 changes of a time. Other companies, what they've done is 0:09:20.800 --> 0:09:23.400 they've built a model long time ago. They don't like 0:09:23.440 --> 0:09:26.439 to necessarily revealed everybody how it works, the secret source. Right, 0:09:26.559 --> 0:09:29.079 So when you're seeing there's a chief risk officer and 0:09:29.120 --> 0:09:31.360 there's a new model coming along, either you want to 0:09:31.440 --> 0:09:33.280 understand that it's going to pay very similarly to the 0:09:33.280 --> 0:09:35.840 previous one to be okay with it, or you need 0:09:35.880 --> 0:09:37.679 a lot of transparency and how it works, so you 0:09:37.679 --> 0:09:40.040 can get comfort in this new model. So I think 0:09:40.040 --> 0:09:43.040 there's the big divergence and strategy. We go back to 0:09:43.080 --> 0:09:45.520 boots and redo everything from scratch each time, but in 0:09:45.559 --> 0:09:47.599 the same time provide an awful lot of transparency and 0:09:47.600 --> 0:09:50.080 a lot of tools so that lenders can get a 0:09:50.120 --> 0:09:52.640 really good understanding of exactly how this is going to 0:09:52.679 --> 0:09:54.600 work and how it's going to behave in different situations 0:09:54.640 --> 0:09:57.080 and they can test it out right, whereas the other 0:09:57.080 --> 0:09:59.440 one is still working with many limitations that have been 0:09:59.440 --> 0:10:02.160 in place since the very first models. And because of 0:10:02.200 --> 0:10:05.560 those limitations, that's a big difference in why we've scored 0:10:05.559 --> 0:10:07.000 a lot more people. So I'll give you a very 0:10:07.000 --> 0:10:10.000 concrete example. So, for instance, one of the limitations that 0:10:10.160 --> 0:10:12.600 the others have is that if you haven't had any 0:10:12.600 --> 0:10:15.800 credit activity for the past six months, that you're not 0:10:15.840 --> 0:10:18.360 going to get a score. So you just imagine somebody 0:10:18.400 --> 0:10:21.439 that works with the military has been deployed overseas or 0:10:21.440 --> 0:10:22.040 anything else. 0:10:22.120 --> 0:10:22.320 Right. 0:10:22.640 --> 0:10:25.920 But the good thing is, starting about fifteen years ago, 0:10:26.200 --> 0:10:28.559 the Bureau started click and storing data so we could 0:10:28.640 --> 0:10:30.960 use time serious data because who'd have thought that time 0:10:31.040 --> 0:10:35.760 serious data could be useful in prediction, right, completely strange idea. 0:10:35.840 --> 0:10:36.000 Right. 0:10:36.040 --> 0:10:39.720 So with finan Score four, we started using trended data 0:10:39.760 --> 0:10:42.680 times series data, and with that, obviously we can see 0:10:42.679 --> 0:10:44.760 back twenty four months. So yes, if there's a gap 0:10:44.800 --> 0:10:48.120 in six months of history, it's important, but we're still 0:10:48.120 --> 0:10:50.720 able to see what happened before then, right, And that 0:10:50.760 --> 0:10:53.280 gets rid of tens of millions of people when you 0:10:53.320 --> 0:10:55.040 have that constraint. There are the constraints in there too, 0:10:55.040 --> 0:10:56.880 So people that are new to credit, so if they 0:10:56.880 --> 0:10:58.720 haven't had a full six months of history again, they 0:10:58.720 --> 0:11:02.040 won't get scored. Aren't any tradelines, they won't get scored, right, 0:11:02.080 --> 0:11:04.040 And so what we've been able to do is to 0:11:04.559 --> 0:11:07.439 deal with thos constraints in a different way by a 0:11:07.920 --> 0:11:10.800 using time series data, b using some other data points. 0:11:10.840 --> 0:11:13.839 So we were the first to use utility payments and rent. 0:11:13.960 --> 0:11:16.200 I mean, who'd have thought that your ability to pay 0:11:16.240 --> 0:11:19.040 your rent could somehow again be useful and trying to 0:11:19.080 --> 0:11:23.040 assess your risk, right, And so including those new different 0:11:23.040 --> 0:11:26.040 types of data, realizing that these constraints can be changed 0:11:26.040 --> 0:11:28.400 now that you have time series data. But then also 0:11:29.000 --> 0:11:31.800 guess what you know? Math has evolved as well, okay, 0:11:31.920 --> 0:11:34.520 And so you know what we realized too was that 0:11:34.960 --> 0:11:37.240 you can be really smart and use some new methods, 0:11:37.320 --> 0:11:39.559 like you know, some AI methods for instance, like clustering, 0:11:39.800 --> 0:11:42.120 to really understand, well, look, here's a group of people 0:11:42.440 --> 0:11:45.120 and they behave in a certain way, and by doing 0:11:45.160 --> 0:11:46.959 that in a better way, we can then figure out 0:11:47.000 --> 0:11:48.520 what is the best way to measure this group of 0:11:48.520 --> 0:11:51.280 people here versus this group of people here, and doing 0:11:51.280 --> 0:11:53.880 that well enables you to build much more predictive scores. 0:11:54.160 --> 0:11:56.200 And so that's an important yance too that not a 0:11:56.200 --> 0:11:59.440 lot of people always realize is that it's not one 0:11:59.520 --> 0:12:04.400 formula that's calculating everybody's score. People will get depending on 0:12:04.600 --> 0:12:07.160 what their credit file looks like and their history looks like. 0:12:07.480 --> 0:12:11.040 They'll get divided into different segments and then each segment 0:12:11.120 --> 0:12:13.559 is scored according to that and that again increases the 0:12:13.600 --> 0:12:16.800 ability to score people accurately and score more people, like 0:12:16.800 --> 0:12:18.960 in our case it's thirty three million more people they're 0:12:19.000 --> 0:12:20.640 able to score, so it's quite substantial. 0:12:36.320 --> 0:12:40.559 How did the models deal with breaks in previous consumer patterns? 0:12:40.600 --> 0:12:43.960 Because we have seen some major ones in recent years. 0:12:43.960 --> 0:12:49.560 So after the pandemic, we had a phenomenally tight labor market, 0:12:49.600 --> 0:12:52.199 and we saw a lot of wage growth for lower income, 0:12:52.760 --> 0:12:55.800 a lot of spending that was sort of unprecedented in 0:12:55.840 --> 0:12:59.480 many ways. How do models actually incorporate that sort of 0:12:59.559 --> 0:13:01.120 big shit shift in the trend. 0:13:01.960 --> 0:13:05.880 So I think that's something really important to try to understand, 0:13:05.920 --> 0:13:08.360 and it's not easily understood by many. So give me 0:13:08.440 --> 0:13:10.120 a second here, I'll try to break this down. 0:13:10.280 --> 0:13:11.679 When people say give me a second, I have to 0:13:11.679 --> 0:13:14.160 break this down. Please break it down. 0:13:14.240 --> 0:13:15.439 You have a minute. 0:13:15.559 --> 0:13:15.959 A minute. 0:13:16.040 --> 0:13:18.120 Honestly, that's kind of what I'm enjoy most about listening 0:13:18.200 --> 0:13:21.360 to your podcast. And so look, the first thing to 0:13:21.440 --> 0:13:24.840 understand is that this credit score is not an absolute 0:13:24.880 --> 0:13:28.839 measure of risk. It's a relative measure of risk. Let 0:13:28.840 --> 0:13:30.400 me break that down for you, right, So what it 0:13:30.440 --> 0:13:32.760 means is that you know a score. If somebody has 0:13:32.760 --> 0:13:35.400 scored seven to twenty in one month, and then somebody 0:13:35.400 --> 0:13:38.920 else has scored seven twenty three years later, the risk 0:13:38.960 --> 0:13:42.000 will be different. And the reason for that is very deliberate. 0:13:42.360 --> 0:13:45.680 When we are evaluating a person, right, because of the 0:13:45.800 --> 0:13:48.720 laws of the Fair Credit Reporting Act, right, we're allowed 0:13:48.720 --> 0:13:51.000 to look at the things that are about you. But 0:13:51.040 --> 0:13:52.920 there are things that are going out on at the 0:13:52.960 --> 0:13:56.520 same time in the economy that impact risk of the 0:13:56.520 --> 0:14:00.480 population as a whole. Right, So that's that's why it's 0:14:00.480 --> 0:14:03.240 so important when you're looking at things like credit scores 0:14:03.280 --> 0:14:06.960 to understand when was that score pulled right, because the 0:14:06.960 --> 0:14:10.040 score of seven twenty in twenty seventeen had a very 0:14:10.040 --> 0:14:13.840 different characteristic of a score of seven twenty in twenty 0:14:13.880 --> 0:14:16.680 twenty two. Okay, now, but the thing to bear in 0:14:16.720 --> 0:14:19.520 mind is it's an excellent relative measure of risk. So 0:14:19.560 --> 0:14:21.960 at any one given point in time, you know, somebody 0:14:22.000 --> 0:14:24.000 with a seven twenty is going to be much better 0:14:24.040 --> 0:14:26.560 performing than somebody at six thirty, And at the same time, 0:14:26.640 --> 0:14:28.880 somebody eight forty is going to be much better than 0:14:28.880 --> 0:14:32.160 both of them. Okay, And that holds consistently true, and 0:14:32.200 --> 0:14:34.600 so that's why it's very important to include when you're 0:14:34.600 --> 0:14:38.160 making lending decisions. But lenders have to be thoughtful, right, 0:14:38.240 --> 0:14:41.240 they have to. As they're making decisions about, you know, 0:14:41.240 --> 0:14:42.840 how many people they want to be able to underwrite, 0:14:42.880 --> 0:14:45.040 and how to think about risk, they need to also 0:14:45.080 --> 0:14:48.560 start thinking about these external factors as well, so that 0:14:48.600 --> 0:14:52.600 they can then set their underwriting criteria to meet the 0:14:52.680 --> 0:14:54.840 kind of level of risk that they're willing to take on. 0:14:55.040 --> 0:14:57.000 I had never thought about this, but of course that 0:14:57.040 --> 0:14:59.240 makes so much sense. So it's like I could have 0:14:59.640 --> 0:15:03.400 an excellent credit history, I could have ex employment pay 0:15:03.400 --> 0:15:07.160 on my rent, but for example, if the economy is 0:15:07.200 --> 0:15:09.360 going down the tubes, I may still yet be a 0:15:09.440 --> 0:15:12.560 risky credit because I may lose my job at some point. 0:15:12.640 --> 0:15:14.880 And so this idea that it kind of has to 0:15:14.920 --> 0:15:19.440 be relative because the underlying conditions that affect everyone through 0:15:19.480 --> 0:15:22.440 outside of our control, but they are still important from 0:15:22.480 --> 0:15:24.560 the perspective of the lender exactly. 0:15:24.600 --> 0:15:27.000 And at the same time, if you get declined for 0:15:27.080 --> 0:15:30.200 a credit card, you can't be told that the reason 0:15:30.200 --> 0:15:34.480 you're getting declined is because unemployment has hit five percent. Okay, right, 0:15:34.640 --> 0:15:37.200 that doesn't work. The laws are very specific. The reasons 0:15:37.600 --> 0:15:39.680 for why you're not getting the top score have to 0:15:39.680 --> 0:15:42.360 be explained, and they have to be based on attributes 0:15:42.400 --> 0:15:44.680 and data obviously that are specific to you. 0:15:44.960 --> 0:15:48.120 What's the most important external factor when it comes to 0:15:48.200 --> 0:15:51.640 credit scoring, Because I've heard arguments for obviously the labor market, 0:15:51.680 --> 0:15:55.680 the unemployment rate, but also wage income and therefore real 0:15:55.840 --> 0:15:58.880 disposable income. How do you weight those different factors? 0:15:59.280 --> 0:16:01.320 I mean, I think if your lender, it's going to 0:16:01.480 --> 0:16:06.400 really depend upon what types of consumers you're lending to, right, 0:16:06.560 --> 0:16:11.160 particularly now we're seeing such divergence across consumers in terms 0:16:11.200 --> 0:16:12.200 of who's doing well. 0:16:12.040 --> 0:16:12.640 And who isn't. 0:16:13.160 --> 0:16:16.160 And so, for instance, if you are a lender focused 0:16:16.200 --> 0:16:19.280 on people that are kind of below prime let's say 0:16:19.280 --> 0:16:22.280 that not completely subprime, but that near prime group, and 0:16:22.800 --> 0:16:25.680 you're focusing on ail loans, and you know you're in 0:16:25.960 --> 0:16:30.800 regions like Texas or in certain areas, then obviously understanding 0:16:31.160 --> 0:16:34.080 the economic conditions that are affecting those people, like a 0:16:34.080 --> 0:16:35.840 lot of those people would be working in certain types 0:16:35.840 --> 0:16:39.600 of industries. What is employment like in those types of industries, right? 0:16:39.960 --> 0:16:41.560 Or is it people that are in the gig economy? 0:16:41.720 --> 0:16:41.920 Right? 0:16:42.000 --> 0:16:44.880 And so it is quite nuanced and it's not necessarily 0:16:44.880 --> 0:16:46.400 one thing, and it's going to depend. Whereas on the 0:16:46.440 --> 0:16:49.040 other extreme, you know, if you're handing out black cards 0:16:49.240 --> 0:16:52.760 and your audience is incredibly affluent, then again it's less 0:16:52.760 --> 0:16:54.800 about the risk because at that point your risk of 0:16:54.840 --> 0:16:57.840 default is probably one end ten thousand. So they're just 0:16:57.880 --> 0:17:00.200 trying to make you sure that it is absolute the 0:17:00.320 --> 0:17:00.880 risk free. 0:17:01.040 --> 0:17:04.440 Right. So someone in a highly cyclical industry, Like I 0:17:04.440 --> 0:17:06.960 don't know, truck drivers in Texas or something that we're 0:17:06.960 --> 0:17:10.320 taking out auto loans would probably be seen as riskier 0:17:10.400 --> 0:17:14.760 or the labor market would depend more for them, whereas 0:17:14.880 --> 0:17:17.720 if you're taking out a black AMEX card or something 0:17:17.760 --> 0:17:20.320 like that, probably real disposable income. 0:17:20.880 --> 0:17:22.040 Yeah, okay, so. 0:17:22.000 --> 0:17:25.679 You mentioned different segmentation. People talk about this K shaped economy. 0:17:25.800 --> 0:17:27.480 Is that real or is that immun. 0:17:27.600 --> 0:17:30.600 I absolutely believe so, but I think that it's a 0:17:30.640 --> 0:17:34.119 little bit more nuanced. And so, you know, one of 0:17:34.160 --> 0:17:37.000 the things that we spotted late last year and we're 0:17:37.000 --> 0:17:40.520 tracking into this year was that we want the first 0:17:40.560 --> 0:17:41.960 to see that it was a K shaped economy. But 0:17:42.000 --> 0:17:43.440 a lot of people were making the assumption that the 0:17:43.480 --> 0:17:46.359 K shaped economy was being driven by income levels. But 0:17:46.400 --> 0:17:47.879 when we were looking at the date at the time, 0:17:48.040 --> 0:17:51.040 we were seeing that those that were in sort of 0:17:51.040 --> 0:17:52.960 the higher income level in our case that's one hundred 0:17:52.960 --> 0:17:55.600 and fifty thousand above. So that's not your you know, 0:17:55.680 --> 0:17:58.160 people who are running hedge funds and sept and you know, 0:17:58.240 --> 0:18:01.600 but still it's the relatively better to do cohort. They 0:18:01.640 --> 0:18:03.639 were actually seeing the highest year of year increases in 0:18:03.680 --> 0:18:07.200 delinquency rates at the time, so we knew that okay, 0:18:07.240 --> 0:18:09.520 home in a second, it isn't as simple as this, 0:18:09.640 --> 0:18:12.040 So you know, we spend a lot of time and 0:18:12.280 --> 0:18:14.000 sort of a lot of banks and we kind of 0:18:14.000 --> 0:18:15.800 collaborated with them to try to understand what's really then 0:18:15.840 --> 0:18:19.200 the differentiator. And then what seemed to be really a 0:18:19.240 --> 0:18:21.320 part of this is wealth. So you know, a lot 0:18:21.320 --> 0:18:24.040 of people don't necessarily differentiate income and wealth, but they 0:18:24.080 --> 0:18:27.280 are separate. And so you know, when you're looking then 0:18:27.320 --> 0:18:29.720 at a high income cohort at the time to try 0:18:29.720 --> 0:18:31.560 to see, like, okay, well, which ones are doing well, 0:18:31.600 --> 0:18:35.320 which ones weren't. Home ownership was the biggest differentiator because 0:18:35.320 --> 0:18:37.879 they had a bigger cushion, something they can rely on. 0:18:38.040 --> 0:18:40.679 And then obviously other aspects as well, like stock ownership 0:18:40.800 --> 0:18:44.560 and small business ownership, et cetera. But home ownership is 0:18:44.600 --> 0:18:46.720 the one that has the bigger effect because there's more 0:18:46.760 --> 0:18:49.439