CECL - Beyond Implementation, Part 1

Virginia Robbins (00:06): Welcome back to PCBB's Banking Out Loud podcast. I'm Virginia Robbins, your host for today's episode on CEC L or the current expected credit loss accounting standard. With me, I have Andy Hines, who works with a public financial institution, and he's already been through the CECL journey. During our conversation, Andy will share with you his priorities, how he made it through the journey and his experiences. Andy Hines (00:31): I think the top recommendation is to really focus on your loan classifications and that trial balance and the loan maturities, which is a data field that most can pull from their core to really ensure that the data that you're gonna need to grab is encompassed in their well, Virginia Robbins (00:52): Because we have a lot to discuss. Our conversation has been broken into two parts today. Part one, Hindi will share with us the lessons learned from CECL, specifically on organizing data and explaining CECL results to his stakeholders. Then in part two, we'll discuss his experience with getting started with CECL, his exposure and participation, and how his institution prepared for CECL. He'll also touch on challenges with spreadsheet. So with that, why don't we get started? Virginia Robbins (01:20): Hi, Andy. Great to have you with us today. I, I understand in your background that you went for flying for the Navy to underwriting the companies whose planes you flew. What was that journey like in your career and, and where are you today? Andy Hines (01:33): When I was at Cornell, I had a Navy scholarship. I really thought I want, you know, I wanted to be an astronaut that, you know, there was a lot of hopefulness for all that before the challenger disaster. Um, you know, I studied biochemistry and statistics and whatnot, so it was always, you know, kind of more of a thinker than a feeler, I guess. Um, and then, uh, went to flight school upon commissioning and, um, you know, always loved airplanes growing up, so that was great. Um, the Navy gave me the opportunity to go back and get my MBA in a program with a bunch of other defense contractors in the early nineties. Uh, and then I kind of revisited topics I really liked, which was statistics and quantitative analysis and finance. I just love the numbers. I like the analysis. I like predictive analytics. I like behavioral finance. (02:33): You know, later with getting into the civilian world, I still stayed with reserves through the nineties. But, um, you know, I would say some of that, you know, when we would try to use predictive analytics for strike planning and so on, I just was interested in that kind of analysis and prediction and so on. So, and there has been a terrific amount of change that all of us in this business, you know, have had to adapt to. Um, I, I was with a correspondent bank in the Mid-Atlantic for a while, um, which was a lot of fun because we banked, um, a, a a whole collection of smaller regional financial institutions, and that was one of the best chapters in my career because again, in your banking career, it's, you get really isolated. You're, you, you don't get a lot of feedback about, you know, best practice across other institutions. Well, there, um, I also, uh, became the president of our loan review function. So, and it was during the great recession, so I really got to see a lot of spilt milk and work with a lot of, you know, people that were, um, struggling through the great recession. Um, saw a lot of regulatory commentary. Um, it was very informative Virginia Robbins (04:00): When working with the regulators. Um, uh, have you found the discussion we've, you know, over time we have seen at times regulators and accountants take, um, different positions on reserves. Yeah. Um, how have you seen the discussions with regulators towards CECL? Andy Hines (04:17): So, so far, um, they've, even during the process of preparation for CECL mm-hmm. , and, um, I believe it was in a fill a number of years ago, financial institutions letter, and they basically laid out to banks what they believe should be the appropriate roadmap to CECL implementation. And I know I used this, I used it in, um, you know, advising to others and sharing with other professionals like myself, you know, organize yourself around those recommendations because they are familiar with that. And so that was during the long implementation process, because they were concerned about risks associated with you being prepared to make the transition on whatever deadlines. And it applies to different institutions. Um, they might be, um, I recall that in the small financial institution, when the regulatory community became much more engaged on a LM analysis and sensitivity and so on, that, um, you know, initially we had to make transitions to black box type models and where the confidence develops, and this is just something everybody has to understand. (05:39): Um, you make reasonable kind of assumptions about it going in, but it'll be a multi-year process as you back test the assumptions in those models. And then eventually there's, you know, there's a, a, a confidence that's developed by, uh, all stakeholders, but it takes time. Nobody is just gonna come right out of the blocks, you know, a hundred percent confident. And I, that's okay. Um, you know, it's, it's not a, because I've seen how this works with other types of, you know, bank modeling and, um, has a lot to do with your process, your awareness that that's how it's gonna work. Um, mm-hmm, Virginia Robbins (06:22): Andy Hines (06:23): In the end, you know, it probably will be better, um, because, you know, you're not, you know, there's a whole nother side of people like me in these small companies that's terrified by these, you know, 20 tab Excel spreadsheets that, you know, are your own black box model that just, you know, they're just too error prone. Mm-Hmm. because again, companies have been trying, once they invest so much into that Excel spreadsheet, you know, we, we went through the turn of the century, we, so we're in the, the aughts, and then you're trying to drag along your methodologies in the 2005 to 2008 timeframe. The call reports changing, wants more detail. And, and you, you know, you get wedded to this, to these internal models that you use and data collection. And, um, and it's actually very non adaptable, you know, if you have to build systems mindful of how it's going to continue to change and evolve, um, mm-hmm. , and even this will, I'll guarantee you. Virginia Robbins (07:37): Um, for you personally, what were some of the biggest changes you saw between, uh, incurred loss and CECL? What were the, did you have any aha moments or any surprises? Andy Hines (07:46): Well, I think it confirmed things that I knew would kind of happen, um, like the one I just described to you where you might have a consumer portfolio where the losses have been a little elevated. Um, and so, but then under the new modeling, because the, the portfolio turns over so quickly, you actually wind up with less reserves. Or on the flip side, you know, a lot of small institutions that had portfolios of seasoned residential mortgage loans, um, you know, they could be in the business. They were, because, you know, they did a good job of making loans in their community. They had very few losses. Uh, but now when you start to do the calculation out into the future, you, you know, it shifts around your entire perspective on potential risks. Um, I have always been a strong advocate of organizing your risk management processes in the bank to parallel, you know, F-F-I-E-C type of, of purpose and classification codes, meaning that, you know, how you manage things inside the bank should look a lot like the schedules and the call report. (09:09): And, and that just helps on all kinds of fronts, because now when the regulator comes in, they don't have to recalibrate their eyeball. It starts with ensuring that the way you have your data organized is consistent with the way someone has grown used to seeing it from your uniform bank performance reports, from your call reports and so on. Um, the kind of information that we were all kind of terrified when CECL first started getting discussed, you know, there'd be lists of like, oh, these 15 data points you need to track all the time. And, you know, nobody's really sure how many of those data points is just like, okay, so first thing you have to do is, is just start tracking everything and you know, your wheels start turning in your head on like, how am I gonna accomplish that? Um, and the, um, the, the years spent committing yourself to, to making what you use for risk management look virtually identical to how they're looking at you all the time from the outside in terms of portfolio segmentation, loss histories, peer groups, Mm-Hmm. (10:18): I mean, you can pull peer groups on UBPR and if you're gonna do any of that to backfill in portfolios where you haven't had loss history, you know, do yourself a big favor and make it more consistent, not less consistent with the kind of information they've already been getting. And I've said that to people about the forward look, whatever you're doing on the A LM should be exactly the same, more or less, you know, what, what you're doing for this analysis. Um, because the extent to which you reduce what appear to be contradictions is, is the extent to which it's one less question you're answering. You know, you just, people can see it, there's a logic to it and so on. Virginia Robbins (11:01): Yeah. And, and it, it's also, uh, FASB's actually looking for consistency. Oh, Andy Hines (11:06): Yeah. I mean, I, again, in looking at banks, you know, when, when we were doing consulting and loan review and all kinds of things, I, I mean, you know, I'd always start out by doing my homework like the regulator does with UBPR. Anything that's outstanding, you know, you're in the, you're on the wrong side of a percentile, just like a regulator. I could build a checklist for myself before I even got in there. And you'd be amazed at how many bankers were using a completely different methodology for risk management. And I'd hear it all the time. They're like, I don't understand why they're asking me these questions. And I'm like, well, I know why they're asking you those questions, , because have you looked at your UBPR lately? And, you know, I get people hadn't, and, and I said, you've really gotta be aware of that. And however, your risk management internally, you know, I would, I would make it more consistent, not less consistent. You may have a few metrics that are particularly insightful to you, but otherwise you're, one, you're not gonna be prepared for their point of view or their questions, what have you. Um, and two, you know, they're like a doctor. They've seen a thousand test results. I mean, you know, they're never gonna, they just know what they know and they're generally right, you know, . Virginia Robbins (12:28): Um, so Andy, I'm, I'm curious, um, given your background in analytics, um, you seem to be like almost the target person to, to build their own spreadsheet. Now there is an issue of model risk and there's validating the spreadsheet and there's main maintaining the spreadsheet and all of that. But there are, you know, uh, in some cases, you know, we have, um, CIL methods like scale, which are, you know, a relatively simple, um, spreadsheet based system. What, what's your opinion of, of, for a small bank moving in that direction? What, what are your thoughts there? Andy Hines (13:05): So, so there's, you know, there are potential hurdles for everyone that's trying to figure out what their roadmap for CL implementation will be. Mm-Hmm. . And I would say that one of the lowest hurdles has been the data analytics tools even to subscribe. Um, I looked at this initially where, uh, kind of like I used to with, um, you know, fair lending type tools, totally internally do my own mapping, you know, because, because you have a better understanding of it, which is great. Mm-Hmm. , you know, um, but the cost of the tools that are available doesn't, I felt that it doesn't justify some of the other risks you have to address, and the other risks you have to address are significant. Um, for example, um, you, our world is always changing and a terrific value you have in engaging with, you know, a big company that's going to host your core processing and, and other things like that is that I always tell people, you know, if, if you are a gazelle and you just come across the field and it's like completely empty, you're like, oh, this is great. (14:31): That's not the first thing that should go through your head. You go, why aren't there other gazelle's here? 'cause there's some kind of danger I'm not aware of. And when you're in a herd, you know, somebody else, you, there's a, a great risk management benefit, you know, to having the right kinds of vendor relationships. So cost is, is one of 'em. I mean, the true cost, even if I buy a piece of software, I work on it. I really have to be honest about my time. This is, this kind of analysis is a cost center. It's not really a profit center. I am a business developer. I, you know, I've gotta manage this. I have to, but I really need to bring business in as well. And I can't just be the analysis department. And that's this thing with the small banks. There's a terrific value in like what we're doing right now, or having working groups or networking or being with a, a, a really solid vendor. (15:29): Because when somebody across the field sees the lion, I don't need to see the lion. Everyone what I'm confident, if they've seen it, I believe them. You know, we'll just start, you know, making changes and adapt to that. Um, so yeah, it's kind of like that. And then finally, obviously is everybody reasonably so is very preoccupied with, you know, managerial controls. Mm-Hmm. . And you just can't, you, you can't be independent enough. So even if we were doing it, I'd have to engage a third party. So there we're back to cost again. Mm-Hmm. , you know, I mean, if I'm doing it completely, I've already got third parties, you know, for certain aspects of it. But it would really, I, I think if you're a really large institution, you probably make it work Smaller ones, you gotta have to outsource it just like a LM and the sensitivity modeling Virginia Robbins (16:28): Or, or your core system or any of the other things that you outsource your Andy Hines (16:32): Core system, or for example, your loan documentation. Right. You know, I used to run into banks that just have been working their own documentation off of shelf documents for consumer loans, you know, for years it worked, never had any problem. And, you know, you could just imagine how non-compliant that can become. But if you're working with a company, you know, that is indemnifying that, I mean, that's keeping up with it, then you can focus on the stuff where you're really gonna make a difference. Because no one can keep up with all of the nuances of, uh, you know, some of the regulatory changes. Just like mortgage documentation. You don't wanna do that at home in-house Virginia Robbins (17:13): In-house. Right. Yeah. Um, so, you know, Andy, we touched, touched on a little bit about implementation, and you are on the other side of that now, um, with your CSO implementation. So, uh, what would be the, when you sit back and look at it, you know, in, in the rear view mirror, what were the sort of like the, the three top or, you know, handful of top best things that you said, gosh, I'm really glad we did that, or in hindsight, I really wish I'd spent more time on this. Andy Hines (17:45): Well, I'm a great believer that, that a lot of your time spent is thinking about many considerations in parallel. You know, it's really not a linear process, and if it, if it gets too linear, then you're overly committed to something you've started doing when you discover that it didn't include something else. So, you know, it takes a good degree of, you know, experience and communication and awareness to know, okay, if I set this piece up, then that's also gonna meet the requirements on this other area. And then of course, everything you do, I know that I'm gonna have to build it so that it can be easily audited. So that mm-hmm, so that there are functional work papers, um, so that it's not overwhelming. So that I, I need, so you're thinking on multiple levels on how this is gonna kind of come together. (18:46): And when I joined this bank, I'd already had strong feelings about the topic I talked about a few moments ago, which is organizing your internal structure to mirror how you already reporting in other areas Mm-Hmm. , you know, for, and why, 'cause you're gonna be more efficient because you're simultaneously answering questions to multiple parties. All those things have to be thought about and incorporated. And, you know, on the roadmap to CECL that the financial institution letter addressed, you know, they were pretty clear, you know, back in 1214 and so on, now's the time to start shifting your portfolios around adding data to open fields and your loan accounting systems and whatnot, so that when you get to the stage of actually putting information into a model, you've, you've already taken a huge potential error out of it. Um, so I'm very, I'm very glad that I had good support where I am, uh, for the idea of reclassifying the loans and scrubbing the loan accounting system and using extra fields appropriately to, you know, mindful of when eventually somebody's gonna ask me for a list of this or a list of that, or the data has to be mapped. (20:22): And that goes to, you know, MDA and CRA. When PPPP came in, I mean, before I even could really digest PPP, I was already thinking I'm gonna get asked a lot of questions about this a year or two years down the road. Uhhuh, , let's, what are the data pieces I wanna tag every loan with so that I'm able to answer those questions later on? Mm-Hmm. Even with CECL, if I hadn't, you know, thought about that and tagged them in the right way, then when it came to doing the modeling forecasting for it, or, you know, even applying the reserve, um, it, it wouldn't be as good as it is right now because they're very unique kind of loans. Okay. So, Virginia Robbins (21:07): So Andy, we have, we have some folks who, you know, thought CECL was perhaps going away, thought hoped maybe it wouldn't apply to them, and are coming to the CECL bandwagon, um, now Yeah. Shall we say. And so for those folks, um, for those bankers, uh, how would you, you know, what, you know, they've gotta do perhaps a fair amount of work relatively quickly now. Um, what would, what would be your top recommendation to them? Andy Hines (21:35): So I think the top recommendation is to really focus on your loan classifications and that trial balance. Um, and the loan maturities, which is a data field, you know, that most can pull from their core, you know, to really ensure that the data that you're gonna need to grab is encompassed in their, well, you know, when I keep harping on loan classifications, the, the reason why I've been so preoccupied with it, why I was preoccupied with it, you know, years ago when more details started creeping into the call report is because the detail winds up in the UBPR, and then you've got somebody that's looked at it, sees where you are percentile wise, and do you have a problem or have you guys just lumped too many loans in the wrong category? Mm-Hmm. . Well now, you know, in order to build up that loss history, in order to have a functional peer analysis, you have to, you have to be aligned with everybody else's call, report data and loss history and so on. (22:42): Um, if, if that data piece isn't taken care of, you're not, you're gonna have a very difficult time with the rest of it. If that data piece, and it's not overly complicated, but you know, you take care of it. And in some cases, you know, lenders are gonna have to jump in, know what they wanna accomplish, and then engage with, you know, data processors to do mass file changes, whatever it is, so that they get the trial balance presentation that, that makes them comparable to peers. That enables them to both summarize their own loss history, but do the comparison to others. Um, 'cause the black box model, if it, if it has that data properly, you know that it can map to it correctly. Mm-Hmm. the kinds of assumptions that, you know, you have to make in it, most of 'em are self-evident. It's, there's no, I don't think there is a huge amount of complexity there or anything. (23:38): Okay. And then you're gonna get an output of it and you should be fine. Okay. Um, all the other issues we talked about, you know, communication, what, you're still gonna have those, but if, if, you know, if I had to say where is the effort, 60% of the effort is in the quality of that data. Um, the model piece, it, especially if a vendor is working with you, is, is not high effort. And the output of it is, is gonna be a lot simpler than anything you've dealt with before. Um, and then finally you just, you know, you've got your communication piece and so on, Virginia Robbins (24:19): Right? So it's, it's making sure that your note department, um, and your lenders and your chief financial officer all working together to make sure that something, something that the note department might have done to help them speed along isn't gonna create a problem for you down the road. Andy Hines (24:38): Yeah. I mean, when people, you know, how did I sell it internally, you know, everybody has a job to do in the bank and everybody's Mm-Hmm. mostly focused on their job. Um, I, I, I knew that, that there were some things that if you can tweak them or improve them or improve data analysis, then all of a sudden you're meeting the needs of other people in the bank too, believe me. That, you know, if you can lead the way and map it out, um, you will get deen engagement of the CFO because he'd love to have it, you know, the fidelity of it improve. So that call reporting looks a lot like financial reporting. Um, you know, the, the people in the, the loan operations areas Mm-Hmm. , you know, they, they'd love to have it, you know? Correct. As well, because many of them in older banks have dealt with a legacy Mm-Hmm. (25:29): of different definitions for loans over the last 25 years. Right. Um, and it makes it very difficult for them every time they're asked to generate a business analytics report or something like that, you know, um, all of these things to the extent that people can understand it's gonna pay dividends for them and what they contribute to the bank in all areas. Um, you know, I think it makes it a lot easier. Um, I don't know. That's what I've tried. Okay. That's been my approach. I think if you just give people checklists, they already have a lot of work. They just kind of, they don't get it. But if, if you can say to them, listen, if we get this wired up, there's, you're just gonna be virtually our call reporting system. You're just gonna be reviewing it. You know, it's not a, and the same thing with booking loans. If we get this wired up, then it's then, like me and some of the other people, I just see a daily trial balance. I can see right away if something's outta whack. But you also have to have your credit people, you know, agree to making it very clear every time they book a loan what the purpose code is because that's, you know, if they don't, so everybody has, has a vested interest, but then in the end they're like, I, this is working a lot better. Yeah. Virginia Robbins (26:48): . Yeah. Andy Hines (26:49): Yeah. It's working a lot better because, you know, we all have to keep doing the same things over and over again where we're providing information, you know, in reports to regulators and auditors and internal management and, you know, people just wanted to, to work correctly. Right. You know, if, if, if it's an institution where, you know, and again, I used to go into some of these places where, you know, they had, here's the report and here's the kind of Rosetta stone you gotta decode, you know, and then trans, and then, because that's how you can figure out how they got to the call report , you know, and I'm like, holy smoke already, you know, and it's like arrows. And I'm like, oh boy. Because, you know, even a lot of the people who are just working hard in the bank, huge amount of anxiety for them that eventually comes back to them, you know, summarize something, they miss one, you know, row and now that through it shouldn't work that way. Yeah. If you get it set up correctly on the data side, as that data goes in, then it's a functional tool for everybody in the bank. And CECL's just one part of it, a LM, everything else kind of feeds off of it. Virginia Robbins (28:02): Yeah. You may, you, uh, remind me, Andy, that at the end of the day, uh, when I served as a chief operating officer, uh, I figured out one time up to a third of my time was spent on either regulatory or internal or external audit issues. Yeah. And if you can reduce that, then that's time spent with customers. So you can move from being, you know, cost center activity to revenue generation activity. And that's, yep. You know, that's huge. That's in, in a, in a time now where we're all looking to ways to improve profits. This is actually an internal investment that doesn't take much effort. It just takes everybody getting on board and moving forward. Um, and you can make a material change in profits. Andy Hines (28:42): I mean, enable aviation, you know, which is my formative thing, you know, and we always loved having just like these short statements about everything, you know, that, that way you always knew what to do. And, you know, we always used to say train, you know, like, you fight, there's no, you know, and what I would get involved with other bankers and try to give 'em insights or help them or whatever, I'd always say, you know, report the same way you manage. So, and your eye will get calibrated. You'll, you'll see things that are out of place don't have different types of presentations for different stakeholders. It just shouldn't be that way. You know, when you look at your CECL, you know, distribution of accounts, it, you know, you go, it, it's instantly recognizable because you see the same thing on the call report basically, you know? Um, that's a good way to be because you become more efficient, you're much more confident about the quality of the data and you're much, you know, you're, you're not doing what you were just describing where somebody's always, you know, asking the chief operating officer, can you make, um, can you do a business analytics report that maybe has this or that just come up with, you know, a, a method of managing the risk internally is the same way that you communicate that risk to, to all various stakeholders. Virginia Robbins (30:11): Yep. Wonderful. So, Andy, as we wrap up here, are there any final thoughts that you have for those that are, you know, just coming to the CECL, uh, discussion or just looking at CECL perhaps, uh, looking at their CECL reports, perhaps for the first time, or even thinking about a CECL vendor for the first time? Andy Hines (30:34): Um, so, you know, I I, I keep learning things all the time. I learn things from chatting with people who are trying to make decisions, reach out. Sometimes it's referrals. It's just, you know, RMA style sharing, best practice, that kind of thing. Um, I tell a lot of people that I wouldn't get too preoccupied with the finer points of, of the model. Um, it is, it, it's more similar than it is different. So, you know, this, uh, an elaborate, you know, picking, well this model or that, did you run three or four models and then compare 'em or something? Um, you know, the, it's, I wouldn't get too preoccupied with that, that fidelity of the data, your whole focus on that we just started talking about. Mm-Hmm. that is such a big part of it. If you've done that correctly, what I think is going to happen over the next couple years is feedback will occur. (31:40): Um, institutions will provide feedback to their vendor partners. The product will continue to evolve the presentation of the product. Um, yes, there are different methodologies and so on, but, you know, um, I, I think, you know, so you shouldn't get too preoccupied with trying to compare and contrast. There are some really quality ones out there that have been around for a while. Um, like I said before, I can't practically see the benefit of trying to do it yourself. Um, it's just there are too many advantages that I, you know, already discussed on that front. Mm-Hmm. . Um, and I, and I, I know I will as we move forward and then begin the back testing side of things, you know, we'll have the valuable feedback we need to continue to, to tweak the, the model. Um, and then I anticipate that it'll be relatively hands off by most people in the company, which is an important thing. (32:48): Mm-Hmm. . And as long as that loan that, 'cause it's just data in the loan accounting system, you know, we don't manipulate, it doesn't have anything to do with risk rating anymore. I mean, you, you know, you're gonna get a pretty pure glass of water from this eventually at the end of the day. Mm-Hmm. . And the big thing will just be comparing what your actual losses were to what, in any given period where the incurred losses, and then you, you'll have ammunition to sort of what if your model a little bit more and true that up. And I, I, I think that the, the labor that has been involved in nearly the last 20 years over the allowance for loan losses, particularly for small companies, it's going to disappear. I mean, it really, I, I think, you know, I I, I don't think that it in, you know, one day it'll seem a lot less burdensome. Virginia Robbins (33:48): Well Andy, here's to that future. Um, yeah, Andy Hines (33:51): I will have other challenges. But Virginia Robbins (33:54): That's for certain, you know, there's always something. Yeah. Andy, I just wanna thank you so much for your time today. I really appreciate it. A great discussion. Um, and, uh, look forward to future discussions. Thanks so much, Andy. Virginia Robbins (34:08): Don't forget to join us for part two. We'll discuss with Andy his experience in getting started with CECL, how his exposure and participation prepared his institution for CECL and why he feels a spreadsheet might not be the right approach for most institutions. For more information about CECL, visit pbb.com/products/CECL. You can watch and listen to all of our podcast episodes on our banking out loud webpage at pcbb.com/podcast.