Connect With NXTsoft Virtually As We Share Our Knowledge To Assist Financial Institutions!
NXTsoft's webinar series titled Pausing the Pandemic Panic: Ideas and Solutions for Financial Institutions in these Uncertain Times is designed for bank and credit unions to address topics that are affecting financial institutions in these uncertain times.
Couldn't Join Us For The Live Webinar Recording? No Problem! Check Out The Webinar Recording and Transcript Below.
'Modeling High Impact Events In CECL' with Tom Parsons with NXTsoft's Omni Platform.
Good afternoon. I am Tom Parsons and I'm your host today for OmniLytics presentation on CECL and modeling stress events in a CECL model. Before we get started today, however, I just want to introduce you to who OmniLytics is. We've been providing asset liability management and interest rate risk solutions to financial services company for a very long time, we are primarily focused on regulatory required financial risk management. So, some of those products and services, our risk analytics ALM model, which is an interest rate risk analysis tool, some ancillary products that go along with asset liability management, such as cash sources and uses or liquidity planning, stress analytics, CECL, Current Expected Credit Loss, what we're going to talk about today, bond accounting and deposit analysis.
So, some of the highlights I'm going to take you through today is a review of CECL. It's been maybe a few months or even a year since you've considered CECL and it's probably worth a review of what CECL actually is and where it fits into our business. Second, we'll talk about the components and methodologies used in CECL, Q-Factors, modeling stress events, which is a highlight of today's presentation and implementation timeline, then I'll give you some of my final thoughts. So, CECL, Current Expected Credit Loss, what is it? Well, it's a new ALLL calculation and really a drastic change to the old methodology. Bankers, you have to be compliant by January of 2023. Now, that date has kind of been a moving target, but that's the latest date. Public companies have had to comply with CECL since January of this year, but nonpublic or non-traded companies have been pushed to January of 2023.
Just in a nutshell, ALLL, the credit loss is forecasted under CECL versus as incurred and of course it's on every bank's radar right now, as I mentioned before, it was delayed last summer until January of 2023. But we are finding that it's gaining traction again and so we thought this was a timely presentation to make for a number of reasons. One, because well, the subject has come up again and has become important or a hot topic within the banking community and with examiners. But secondly, with the recent economic shift taking place due to some stress events, namely, COVID-19. Where we use COVID-19 and maybe mention it again later on in the presentation (it's certainly a good example of a stress event) Stress events and the methods that we're going to talk about today can be used in any kind of stress event that that might come up.
Just a review of what CECL actually is, Current Expected Credit Loss. It's a new credit loss standard issued by FASB, actually in about 2016 I believe it first came out. And as I mentioned before, been some delays, but right now the target date is January of 2023. But regardless, as it stands today, bankers really must prepare in advance for CECL because it brings a whole new perspective to estimating credit losses. Even though you have more than a year before full implementation, you have to begin learning and strategizing, even collecting data in order to be ready for the compliance to CECL. And we know that the examiners are asking you about this, so they might be checking your plans, maybe they have already, but they're going to be looking for progress. So, it completely changes the timing of the loss recognition.
Instead of waiting until losses have incurred, as we've done in the past, what CECL requires is that you estimate the credit losses and book those upfront on origination date. Now, I know talking to other bankers this question has come up, and many of them are saying, "Well, maybe I can collect all of the loans that I booked in a given month, for instance, or given time period, and book them maybe at the end of that time period." And that's something for you and your auditors to discuss and that's something that we're going to try to solve for you today. But regardless, this change in timing is going to impact your pricing, how you price your loans, so it's going to impact your income and it will also impact your capital planning. So, CECL requires you to calculate and document your expected lifetime losses before you make another loan. So, you need to understand the many various methodologies available to calculate CECL and the different data that each of those methodologies require.
And then there are also qualitative adjustments to be made. We're going to walk through a few of those today. We'll use those in some examples later on in the slide presentation. Regardless, you're going to have to pick the best method for your bank based upon, really where you are today and then the size and complexity of your balance sheet. And really, CECL gives you the opportunity to choose the appropriate level based upon the size and complexity of your organization. There are two methods we're going to walk through today. The Vintage Method you may have heard of, you could call that instrument level or detail loan data, whatever you might call it, we call it vintage. And then there is the aggregate method, and that's a way of producing the results really without a lot of heavy duty detail in the loan data requirements. So, we can download information from the call report and use that as our starting point in calculating.
So, before we get into the various methods, let's just talk a moment about what CECL components exist. Really, regardless of the method that you use, you're going to have information that's going to give you, or need to give you, historical all time loss rates for your specific institution. Then we're going to look at near-term loss rates for your specific institution. So, what does that mean? Historical or all time loss rates is a span of years, probably 10 or more. So, when you start thinking about time frames like that and methods that you might choose, you've got to think of in terms of what data you have available to support that. We know that the call report data goes back many, many years, and we have that information available to us. However, the loan data might not always exist in your loan system.
Near-term loss rates, so we're going to look at slices of time that are more near to today, to give us a better understanding of what your current credit loss rates are and use those as a forecasting basis going forward. We're also going to look at peer loss rates, and there's a reason for that, because it might fill in the gaps where certain data doesn't exist. And lastly, we're going to talk a bit today, I have a few slides on qualitative factors. That's something that a lot of people struggle with as well.
Let's talk first, a little bit about Q-Factors. The reason we have Q-Factors doesn't change between the incurred and the CECL loss methods. And what do I mean by that? Well, we know that some loans, groups of loans, that is, will have losses that are different than their actual historical loss rates. One pool of loans might continue to be forecasted at a certain loss rate, while another pool of loans might experience something very different. Q-Factors is our way of being able to adjust that on a go forward basis. Now, we can adjust loss rates using a number of different methods. One, is the bank's actual historical loss rate using that as an adjustment, but also using peer information and we're going to take a look at that in a few slides down, but maybe the peer information doesn't quite fit and so you have nothing to go on.
Well, Q-Factors gives us that opportunity to make adjustments. So, we use it to adjust the base loss rate. Now, CECL standards treat Q-Factors a little bit different and here's what I mean by that. The Q-Factor adjustment that you use in the seasonal model should be in sync with the Q-Factors being used as assumptions in other models within the organization. So, for instance, if you have a credit stress model that you're using to maybe stress test your commercial real estate loans, if you're using a variable for Q-Factors, in that model, they should be the same Q-Factors you're using in your CECL model.
So, CECL gives us an opportunity to continue using Q-Factors and really Q-Factors must be used where limitations exist. As I mentioned before, if there is missing data or gaps in the data, Q-Factors is a way to fill that in for that. It is the subjective variable in the CECL calculation. Talking a little bit more about Q-Factors and maybe switching gears here, and not specifically talking about what they are, but what they look like. And I've just lifted this particular slide right out of our own documentation that we use when we deliver a CECL report. And something to maybe focus on here, you can just read this, but there are nine main factors that all organizations should consider and these are questions that you need to ask yourself on a regular basis.
So, I'm going to pop over here to the Q-Factor list. We'll just read through these, evaluate your bank's lending policies and procedures. So, they want to know whether anything has changed from the last reporting period to the next reporting period, in each of these nine different categories or questions. Evaluate your bank's lending policies. Have they gotten worse, have they gotten better? Have you maybe you've rewritten them? Second and 2A here is what we want to focus on in the rest of the presentation, evaluate the impact on your bank of changes in economic conditions. And certainly, we're experiencing that right now, but we also experienced it in 2010. It seems like every 10 years we experience some sort of event that disrupts the market and causes recessionary type situations. So, 2A really is something that you should be addressing right now, using Q-Factors anyway, in your current ALLL calculation.
But as we go forward and work this into CECL, this is going to be an important question to ask. 2B, evaluate the impact of local or regional economic conditions. So, as it suggests here, it could be changes in regional employment, so maybe there's a labor strike, or perhaps there is a business that's moving out, moving to a different state, things like that, that could impact the lending at your bank. Third, evaluate your bank's loan portfolio composition. So, how has it changed? Has it shifted from residential to commercial? What's the mix of your portfolio? Fourth, evaluate the experience of your lending management, so maybe somebody retired or maybe you just hired somebody, you expanded them. So, your experiences have improved. Evaluate the volume and severity of your bank's past due loans, has your delinquency changed? What about the number six? What about the review system that you have? Seven, evaluate the quality and collateral underlying your bank's loan portfolio.
Maybe the real estate values have gone down in your area for other reasons and that's impacted the collateral. Number eight, evaluate the existence of any concentrations of credit risk in your bank's loan portfolio. Nine, evaluate the existence of other external factors, so competition and things like that. All nine areas here need to be answered each time you set out to start to evaluate or analyze your own CECL, your credit loss forecast. So, what you're trying to do here is, just to reiterate, we're trying to determine which direction this has changed. Has it gotten worse or has it gotten better and by what degree? Maybe it's just slightly better or slightly worse, or maybe it's significantly better or significantly worse. And each one of those answers, we're going to try to change into a quantitative value in order to influence what the loss rate is for your portfolio.
Moving away from Q-Factors, I'm going to talk a moment about the methods for calculating CECL. And there are really many potential methodologies available to calculate historical loss rates, and they tend to fall into the two main categories, to the actual CECL study or analysis. The first is an aggregate method and the second is instrument specific. So let's start with aggregate, and you may have also heard the term called WARM, Weighted Average Remaining Maturity. With this method, we're going to use call report data, and we're going to pool the loans into the call report categories that you're familiar with seeing, you'll see that in a few moments, when I pull up a slide of an actual analysis. We're looking at the performance of an entire segment of the loan portfolio versus individual loans. It's really the preferred method where no further segmentation is needed.
So, what's meant by that? If you have one to four family, and that's a segment that is allowable by you, by your board, by your examiners, as far as the level of analysis, then we don't need to split that up into more finite categories. If that's the case, then the aggregate method works fine. We're going to gather the most recent loss data, and we're also going to gather all time loss data using call report information. So, aggregate methods are appropriate when the institution is a little bit smaller, maybe less complex, or perhaps even where the data available is limited. So, if, for instance, the loan data, if you wanted to use a loan level model and that data wasn't available, well, then maybe the aggregate method is what you're going to need to use.
Next, is the instrument-specific method and this relies on the performance of each individual loan or a subsegment of loans to best represent the loss behaviors. The instrument-specific method requires identifying and tracking each individual loan over time. And because of that, it can become prohibitive because that data just might not be available. So, we're looking at data on all loans originated in, perhaps, the last 10 years, and if you think about your core loan system, it might not have all that data for you. In fact, if you did a core conversion in the last 10 years, it's likely that you don't have that information.
Many people have done a recent core conversion, and they've lost all of that data up until the time of the conversion. From that data we're going to want the principal balance, the originated principal balance, the effective yield, the loan origination date. We're going to want variables for segmentation. So, we're going to want credit scores, perhaps, loan types, product types, things like that, in order to group the loans together for evaluation. So, what happens here is when you're trying to make adjustments to the CECL base loss rate, it doesn't always line up with peer information and where you don't have loss rates for a particular segment, you might need to be looking at peer data in order to have a better understanding of what the forecast might be, just by looking to your peers.
Well, if it doesn't line up right, then peer data might not be a really great way ... A method for forecasting. There is also, because we need instrument-specific data, there is a risk of privacy issues because we're going to get individual loan information and account numbers and other things in order to track these loans over time. Well, that might run into some issues from a privacy perspective. Well, let's look at a specific analysis. This is an actual bank's data. It's call report data through the year end of 2019. And you'll see in the first column, the balances, the loan balances, in a smaller organization, about a little over a hundred million dollars in loans. I'm going to skip over the next column and talk about the middle columns, where you are seeing the title at the top, bank and peer. Within bank we have all loss periods, loss rate, and we have the most recent loss rate.
We have similar for peer and the calculations are the same too. Let's look at all period loss rate, what that means is, since this is through 2019 and our model, we're going back to the year 2007 and collecting the loss rates over that entire span of time. In the most recent loss rate column, where you see a lot of zeros right now, we're calculating the loss rates over more near-term time. And that's where we have to look at the average life in quarters. We want to know what the loss rate is over the average life of a loan segment. So, let's just focus on one to four family first, with $22 million in loans and we look at the average life column and we see 22 quarters, and these are in quarters.
What we've done through our own data analysis is determined that typically a first mortgage loan is going to have an average life of 22 quarters. Now, your bank or credit union might be different than that and that's fine, we can change those numbers, but for purposes of today, we're going to use 22 quarters as the average life. And what we've done is looked back over the last 22 quarters for the loss rate. In this bank's situation, they're in a fortunate situation, where they've had no losses. But if we were to look at the entire time period, back to 2007, we know that they've incurred losses over that time period of 4.43%.
Similarly, we're going to do the calculation for its peers. Now, we're going to take all the peers in the same state as the subject bank, and in this case, I limited the size of the organization to $250 million. So, we have banks or a collection of banks and their peers that are similar in size at least. I mean, we could also slice those up a little bit differently, but we're just going to use size for today. 7.26% is the entire span loss rate for all time. 0.61, more recently, that is in the last 22 quarters. And then the next couple of columns, we have peer adjustment, we have bank adjustment and Q-Factor adjustment. So, if we are going to use this bank's most recent loss experience as its base, we're going to look over to the top right, CECL reserve rate, that's our final calculated rate and what is it?
Well, it's zero because they've had no losses. Well, that's not going to be acceptable and this gets back to my discussion about Q-Factors. So, we have Q factor adjustment and we have bank adjustment, and we have peer adjustment. We have to adjust that zero to something. One way to do that is to add a Q-Factor. Remember the nine questions, we're going to answer those questions and then drop in a number that tells us what our Q-Factor adjustment should be. We could also look at the peer column and determine that maybe we want to adjust our zero towards what the peer experience is going to be. Or we could look at the bank loss rate over all time and adjust our zero towards the bank all time rates. So, in this case, we'd be adjusting zero towards 4.43, or we'd be adjusting the zero towards 0.61.
Now, we're talking about the current situation without any kind of stress. Maybe this is the business as usual plan, what we're going to use for a reserve rate. But when we introduce stresses, there are going to be lots of options and lots of ropes to pull and levers to pull for calculating the final loss rate. Let's look at this a little bit more closely. All I've done here is sliced out the real estate loan section in their call report. These should be familiar categories to you. Real estate loans, farmland, helots, one to four family first, and that's what I've highlighted here, just so it's a little bit more clear to look at. So, I mentioned before, this is aggregate data from 2007 to 2019, when we look at all periods loss rate. There little recent loss over the last 14 to 22 quarters. The loss rates are zero in every category, except for the owner occupied commercial category. There is a small loss of about 0.24%.
We also have low peer loss rates and if you look in the lower right hand corner of this table, and it was the same on the prior slide, the total CECL reserve, if we were to calculate using the base loss rates of the most recent time periods, the CECL reserve would only be $57,000 and that's just really not going to be acceptable. There this bank's existing reserve is $1.6 million and over their total loans of 106 million. And so, if we were to compare that to the CECL suggested rate, their CECL impact is this, that they are over reserved by $1.5 million. Well, we just know that that's not going to be the case. So, let's adjust. Here in my CECL model, I've made some adjustments, it's the same data from the prior slide, but adjusted and I've also introduced a little bit of moderate stress in here.
It's 2019 data, still 2019 data, adjusted with historical analysis. I've added peer historical adjustments at a hundred percent. I'll talk about that in a second, but that's our objective adjustment, and I've also added a Q-Factor adjustment at 1% and that's our subjective adjustment. And I'll get back to why they're under-reserved here. So, let's look at this again. We started out with a most recent loss rate at zero, but what I did was I adjusted, if you look under the peer adjustment column at 100%, in other words, I said, "I'm going to take my 0% and I'm going to move it up to 0.61%, say that I'm going to copy my peers." But I don't think that's enough because it's just too little for today's environment.
So, I'm playing what-if with my model now, and I'm going to add a Q-Factor here, an adjustment, and probably because of that second question in the questionnaire about what's going on in the region, or perhaps, what's going on globally from an economics standpoint. So, I've added 1%. My total CECL reserve rate is now 1.61% and I multiply that reserve rate by my balance of 22,000,800. And that gives me a CECL reserve, suggested reserve of $367,000. I've only done this for this middle section that we're looking at here, but there are the sections for commercial real estate and other types of loans. And so, at the bottom right, where we add up to get the total CECL reserve, those numbers aren't going to add up from just this section, that's also including other sections. But this would suggest using our peers as their loss rate and also adding 1%, that's going to give us a total CECL reserve of $2 million.
Comparing that to our existing reserve of $1.6 million, that means that we may have to further reserve $416,000. In other words, we're under-reserved at the moment. It's suggesting that our ALLL go from 1.5% to 1.9%. And also, there's another important point to make here. That might be the current adjustment, but what about go forward? Well, that CECL reserve rate column there, that's the suggested rates we use on every new loan that we make. So, as I have mentioned in an earlier slide, the whole point of CECL is to take the allowance as the loans are made, or maybe as a group of loans are made and that would be the rate that we apply. And so, that's going to impact our pricing on these loans as well. It has to be built into the pricing model.
So, modeling stress events, I've introduced a little stress in the prior slide, but here's what we want to do. And let's look again at the timeframe we're in now, we want to assume that something bad is going to happen, and it did in 2010, it did in 2020, again, and next time that comes around, we want to be prepared. So, what do we do? Well, we got to find slices of time that represent those stress events. Well, we don't have data for the COVID-19 of 2020 going on. We do have the data for 2010 during the recession. So, we're going to take a look at that in a moment. We're going to stress with Q-Factors and we're going to introduce peer results and then we're going to test with institution-specific history. We'll look at that in a moment.
And we're also going to introduce what I like to call low probability, high impact assumptions, and what I mean by that, and I use that a lot in terms of interest rate risk management as well. What I mean by that is, I'm going to pick assumptions that are unlikely to happen, but if they did, they might cause a severe situation. So ,again, we're just modeling here. We're playing around doing what if? But it's going to give us a perspective on what we think might happen and how it might impact our ALLL and then ultimately our capital and our loan pricing on a go forward basis.
Here's the same bank, same information, but just sliced up a little bit differently. This is the most recent loss base data from 2012 data. And that's aggregate data from 2007 through 2012. We're capturing the recessionary period, it's a significant loss during the recession period, you can see that the most recent loss rate during that period, and just ignore the all period loss rate for now, the most recent loss rate for one to four family loans, 9.42%. much different than the 0% they experienced leading up to today. We're entering a period of stressful time. We don't know if we're going to experience another 9% loss in the next recession, but that's what happened to us last time. Granted, it was a mortgage bubble and so that's going to impact the one to four family first category significantly.
So, we also look at our peer loss rate, those were significant, 7.24%. Running the same calculations without adjustment, it's going to suggest that our CECL reserve is actually really out of whack here. 3.8, 3.9 million compared to our existing reserve, it's saying we're under-reserved by $2.3 million. But this is just sample data for us to look at and get a feel for what we might expect next time around. Next, we're going to take our 2019 data again, and then stress it with significant information that we've learned from our 2012 analysis and other historical data that we got from our peers. So, this is 2019 data adjusted with historical analysis. We've added bank historical adjustment at 50%. I'll talk about that in a second. That's my objective adjustment. We've added a Q-Factor adjustment of 1%, again, that's our subjective.
And our total means are under-reserved by 784. So, let me take you through that again, $22 million, our most recent loss rate is 0%. We have to adjust that, but we're going to adjust it with some significant stress. And what I've done here is said, "Okay, well, we don't think that the total loss rate over the last 13 years is exactly what we're going to experience, but we think we're going to get half that way." And so, that's what I've done here. I've taken 50% of our loss rate over all time and applied it to our current loss rate. So, now I've raised my 0% loss rate up to, in that column under bank adjustment, 2.2%. Now, I've also added a Q-Factor adjustment because of the stressful time we're going through, that's going to further increase it to 3.22%. That's a little bit more reasonable. We're still under-reserved by 784,000 and it raises our ALLL from 1.5 to 2.25%. However, it might more adequately reflect the experience we might see during a recessionary period.
Well, that's a lot to consider. There are a lot of levers to pull and things to consider in, not only implementing CECL, but then selecting the correct variables for your particular model. So, I put together this timeline to back into when we think an institution should start taking into account a consideration and selecting a model and whatnot for implementing CECL. So, let's just back into this. CECL becomes effective at the start of the next fiscal year, beginning after December 15th, 2022. So, for most institutions, right, really the remaining institutions since the public ones have already implemented, that's going to be, likely to be, January 1st, 2023. And that's where the often mentioned 2023 time frame arises. But look again, carefully, CECL starts January 1st, 2023. Well, what does that mean?
Applying CECL, starting that date, means that you have to reserve day one exposure starting on that day, January 1st, it requires you to reserve the expected lifetime loss on your new loans, on the origination date of that loan. So, how are you going to make new loans in January of 2023 without a calculated and documented CECL reserve estimate? It's really a short answer that you can't, which gets me to my next point. You have to have an operational CECL reserve calculation as of that date. You can't wait until December 31st, 2022. You know how the year end is and how hectic it gets, call reports aren't even due until late January. If you've selected the WARM method and the aggregate method, then you're not going to be able to use that on January 1st. So, even if you had the data, you'd need time to process and document it. So, there's no way you can wait until the end of 2022 to begin the process. You have to be operational with it by September of 2022, but that doesn't mean starting on September of 2022.
At that point in time, you should have your parallel processing done. You should have run parallel, ALLL current methods to CECL by that time. So, to safely make the change from the incurred method to the expected method, you should run parallel for at least a full year. It's really, it's the biggest accounting change to hit financial institution in decades. And it's really a shift and a complete transformation from today's current loss. And to just go from one to the next over one day is just simply not practical. So, you need that time running parallel to fine-tune your model, adjust loan pricing if you need to. You need to understand how CECL is going to influence your loan pricing and then also get everybody in the organization familiar with all of the changes that are taking place. So, again, you need to start by September 30th of 2021 to start your parallel processing and make your fine-tune adjustments.
Well, what does that mean? You have to have board approval. You have to know what model you're going to use. You're going to have to have policies in place and other things, there has to be internal education. And in order to reach that date, September of 2021, then you're going to have to have a selection of your CECL model, at least by June. You got to determine the method you're going to use in order to select a model. So, in other words, you really only have a year from now, and this is May of 2020, in order to select what method you're going to use and what model you're going to use. But in that time, between now and June of 2021, you're going to have to educate yourself. You're going to have to quantify and estimate all of the things that are going to go into implementing a CECL model. So, really, don't let the cost and complexity paralyze you. We know that there's limited time and resources, but you really need to start filling up the CECL gaps and ramping up and getting prepared.
So, some final thoughts, again, it's really time to get seriously working on the timeline. Examiners are going to ask for progress. We know that some organizations we work with have had to document that and asked us for help with documenting that, to give to their examiners for their next exam. We can help you answer these questions free of charge, give us a call. I'd be happy to have conversations about CECL with you, determine whether or not it even makes sense to work with us. One thing we do offer is outsourcing it, so this is not something that you have to do on your own. You can just throw your hands in the air and say, "Hey, take care of it for me." And we'll do the heavy lifting. Thanks for listening in today. I'm Tom Parsons. My contact information is there on the right email@example.com. 847 370 9937. And if you have questions, please send them to me at that email address, I'll be happy to respond, anytime. Again, I'm Tom Parsons, I'm with OmniLytics. Thanks for sitting in today and reviewing the model stress events and CECL review presentation for today.