New Constructs' Friday Webinar: Intelligent Capital Allocation 21 April 2023

Great. Okay, guys, we're going to go ahead and get started. I'm Pam Pesek. I'm joined here today as usual by Kyle Guskey II, and Lee Mineta Kohler is joining us as well as Cody Fincher, both from the analyst team. Let me see here. We have a great agenda to cover today, and before I get into introducing that to you, I did want to, as usual, Go through the disclosures and disclaimers, as New Constructs is not an advisory firm. So please familiarize yourself with this slide. I'm going to kick us off by giving a quick review of what we're talking about on society, and then Kyle will be going through some of the latest research on Netflix and Tesla, and Lee will be going through our focus list updates. We will round it out with Cody talking about our data checks, which is a little bit of behind-the-scenes look, and he'll be using a particular example recently with AIG. You're welcome to ask questions throughout. There's a little question widget, so please don't hesitate to use that. I'll be monitoring that, make sure that we get as many questions throughout our time together. As possible. Let me get over to Circle. Here we go. So this is our questions for webinar section. It's new in Society. I'm going to be posting a comment here every week. I would really love to hear your suggestions for topics, any questions that you might have that you want to make sure that we cover, any feedback on previous webinars. Or on what you would like to see happening in the Society. We're having some great discussions. I really appreciate everybody bringing their comments and participating in some really fascinating conversations. Giulio pointed out a really interesting potential for suspension that our team acted on. And also Jim Fain brought up a really interesting question about the difference between our unattractive or dangerous stocks and our focus lists. And this is something that Lee will touch on later, but we have some different research products at New Constructs. Some are very quantitative and some use that quantitative work and then extend it with qualitative work, which I think adds a lot of interesting value. And so that's often why you see the difference between the model portfolios and the focus lists. Also, thank you, Howard, for posting this email, very brief email exchange between Howard and David on ROIIC that was also, I thought, very useful. So please hop out to Society. Please send your feedback. You can always DM me on Society as well. So I'm going to switch back.

I wanted to add quickly, Tam, that All the analysts and all the software developers at New Constructs are all on the society and interested in engaging and interacting with people. So if you have questions about anything, that's one of the best places to get a hold of us quickly and easily. We also get awesome answers to questions from, from other customers and members of the society. So it's a great place to ask the community what's happening and what they think.

I wholeheartedly agree. All right, with that, I'm going to go ahead and turn this over to Kyle.

Great.

I'll leave it on the slide there just a second. We're going to get into Netflix and then Tesla. And so I'm gonna actually jump into the website, we'll look at those reports, and then we'll actually dig into the reverse DCF. I'd like to explain how we're actually making those Scenarios that put together those charts you see on the slide right there. We'll get into the assumptions and what drives those scenarios and kind of play around in the DCF so you can get a real feel for exactly what we're doing when we put together these reports. With that, I'm going to take over screen share and first take you straight to the research page of the website. Which is where you can find the latest research. We put out a lot of stuff this week. We'll talk about most of it in here, but definitely check this out each week to see the latest reports we're putting out. But first we'll start with the Netflix report. We put out right after their 1Q23 earnings. We wanted to take a look at what happened in the earnings, what it means for the stock as far as future growth. As far as the actual price of the stock and the expectations baked into the stock price. So a quick overview kind of on what happened. We've seen for quite some time now, Netflix was the growth story, growing members, growing subscribers. That kind of switched last year. They started actually not projecting forward memberships anymore. It was all about revenue due to the kind of introducing of the ad business. But so far we've seen slowing growth now and falling margins. Netflix revenue of 3.7% year over year in the quarter. Guidance isn't for much more, 3.4% year over year next quarter, and well below some of their previous goals of double-digit revenue growth. And so you've got a slowing business, margins falling in the wrong direction as well. And so when you put that together, like I said, we'll model that in the DCF, kind of get the picture of Why the company does look the way it does from a valuation perspective. We've seen they're another streamer at this point. The industry has evolved to where they were once the first streamer. Now we've got streaming companies everywhere. Every service seems to have their own app, their own platform. And so they're now just one of many at the point, at this point. And then add in the free cash flow burn. We've seen negative free cash flow in 10 of the past 12 years. One of those years was 2020 when they significantly cut back on production. They've cumulative negative $7.6 billion over the past 5 years alone. This is a heavy cash-burning business and it's not going to really change because the name of the game is new content. And so they have to keep investing in that content. And so when we put all that together with our, all our previous research, you can see on Netflix, and we've got links in this report to all our previous reports right here at the top. Puts it together to be a highly expensive stock, overvalued. This is the chart we like to talk about. Here's their past NOPAT. Here's what they've got to do in the future to justify the current share price. This was $330 right before earnings. And then in a still optimistic scenario, we're looking at $175 a share. And then we detailed each of the assumptions in those scenarios here, and that's where I'm going to jump into the DCF model and actually show you where we put those together. Let me jump over to start the ratings tab. You can just type in Netflix, open up the company model. It takes a second, so I've already got it opened up here to keep this moving along. But what we want to look at here is the forecast page, which is where we put in the assumptions on our DCF model. We've got 3 scenarios: neutral, optimistic, pessimistic. Those are standard names. You can make them whatever you want. If you want all of them to be optimistic and put in 3 optimistic scenarios, go for it. But for this analysis we used, I used the neutral and the optimistic, the optimistic being what do they have to do to justify the current share price? In that analysis we looked at— let me get to that so I can get the right numbers here. So we wanted to say, okay, let's look at optimistic scenario. What do they have to do to justify that $330 a share price? And in that case, we're going to try out different margin assumptions, different revenue growth assumptions. We can change tax rates. We can look at net working capital as a percent of revenue, fixed assets change as a percent of revenue, which kind of drives that invested capital growth. And so all of that combined goes into the DCF and we can figure out what each scenario looks like. In the case of the current stock price, we said, okay, what if they get percent margin 17.7 here? Which would be their previous high from 2021, right? 2022 was 15%. So even though 1Q was declining to justify the stock price, they've got actually improved margins to their previous high. And then we looked at revenue of 13% compounded annually for the next 10 years. There's that 13% out through 6 to 10 years. And then we looked at the change in capital We kept it at 10% of revenue. And so these numbers down here at the bottom, it's just kind of trial and error to kind of see where the change in invested capital falls. We wanted to keep it at 10% of revenue, which is well below what they've done in the past, but we want these scenarios to be sort of a best case. We're not trying to be overly punishing the company or making a scenario that is overly pessimistic. We want to be conservative and we want to be optimistic to see what they actually have to do to justify that stock price. And so when we put all of these assumptions in, we would save it off, and then we can go over to this decision page is where it takes you back to. And you can see this optimistic scenario. That gap that was previously over 100 years is now 9 years. The revenue CAGR goes from 3.6 to 13.2. And so what this is saying is, in order for Netflix to justify the current stock price, they have to achieve these assumptions for 9 years. And we can see those results actually in the DCF review tab here. We can move it out 9 years, update to the optimistic scenario, and jump out 9 to 10 years or so. We can see what the implied revenue in those years would be. There we've got our 13% revenue growth, we've got our 17.7% margin assumption, and then we can scroll down and get to that shareholder value per share. Right here, $318 in year 9, $328 in year 10. So when we wrote this, it was $330. So it was that 10-year scenario. This is what Netflix has to do to justify $330 a share today. When we look at the actual report, we quantify it in a little other ways because looking at this, you might just say, okay, cool, they have to do that for 10 years. What does that actually mean? And so that's where we can Compare it to peers. In this scenario, Netflix's revenue would be $109.2 billion in 2032, which would be 1.4 times combined Fox, Paramount, Warner Brothers combined, and 1.3 times the actual TTM revenue of Disney right now. And so to justify that current stock price, Netflix has to be bigger than Disney completely and bigger than most of all the competitors, the big ones at this point. And so With those expectations, we can say that doesn't seem realistic in our mind, given again falling margins, slowing growth, revenue growth expectations for 3%. We're looking at 13%. So we can go in and put in a more realistic scenario. We looked at dropping margins from that 18% to 14%, which is the 5-year average, 13.7% here. And then revenue growing at consensus for the first couple of years. And continuing out through year 10. So we're still being optimistic. These future year growth rates could be lower, could be higher, but we're gonna give 'em consensus estimates there. And we kept that growth in invested capital at 10% of revenue. And in that scenario, again, we can go out and see in the DCF review tab exactly what Netflix should be worth today if that were the scenario of growth that we expect. So we're gonna do that. We can actually go in and see. About 10 years, you're looking at a shareholder value of $175 a share. And then you can actually scroll down here and see where all of the kind of growth rates, what they equal in those— in that scenario. So you'd be looking at an 11% revenue CAGR, ROIC of 12% average, and 13% in that year 10. And we can see Each of the drivers of this model and what that has to do in this scenario, or what it's actually happening in this scenario. And again, in the actual report, we then juxtapose that against what it would look like even at $175 a share. Revenue would be $87.6 billion, 104% of Disney's TTM revenue, and still 111.1% of the combined revenue of those 3 from above. So still got a very optimistic scenario, but again, want to be conservative here and say that $175, almost falling in half, is still an optimistic scenario. So it really highlights just how much downside is in Netflix's stock based on our DCF model.

Kyle, I've been using reverse DCFs for a few years now. I know you have too, but I remain excited and impressed by the ability to connect stock price back to performance quickly. That's one of the pieces of magic here is, right, we can set a timeframe, we can set a stock price, and look at the kinds of performance hurdles that the company has to meet in order to justify that stock price. I think lots of times you hear in the news about subscriber growth or loss or revenue growth for Netflix, and it's hard or not necessarily connected with price or the connection's not as clear. And the DCF really gives us an easy way to understand that relationship. It connects operations with valuation in a much clearer way than you get otherwise.

Yeah. And I see we've got a question here on how do we account for the change in revenue for Netflix on sharing passwords? Right now they've been kind of non-committal on it. They were previously going to roll it out a couple months ago, seemed to slow down. Now they're talking about sometime in 2Q23, and that's where we can plug it into the DCF. If we want to say that Hey, it's going to increase revenue or decrease revenue. We can put in those assumptions right in the DCF. And so if you wanted to, hey, they're gonna grow quicker in the first few years because more people are gonna sign up than consensus, we could go in and model that. And if we wanted to say here the same 13.7— actually, I'll just change one of the other scenarios so I don't have to go type all that in in the middle of this. But if we wanted to say that neutral scenario, 15% in the first couple years, You could save that off with that 14% NOPAT margin. We can see what it changes from the $175 a share scenario. It's going to save this off, it's going to regenerate the model, then we'll jump into the DCF review and see exactly what it means for the stock price. So just one second.

Last week we also talked about how you can do this same kind of analysis in the download XLS tool. One of the advantages of the XLS is that it updates a little more quickly than the website does. Yes.

Quickly, as in instantly. So, and these always seem to take longer when I am sharing my screen. But maybe that's just a feeling.

They're much easier to see on the screen, though, when you're using the web version.

Yes.

So yeah, we'd go into that neutral scenario. With those higher revenue growth rates in the first couple of years, everything else stayed the same. That $175,000 went to $192,000 almost. So you can see changes, how we can model changes in revenue very quickly, very easily, and see exactly what it means for the stock price. Again, could do that with any company, any stock margins instead of revenue. You look at all that differently. With that said, I'll jump into Tesla as the other one. If you've been reading our research for a while, that we have long called Tesla one of the most overvalued companies in the market. At times that has been proven correct from a share price, other times not so much because the stock definitely does not trade on fundamentals for most of the time. But fundamentally what we've been saying for a while is that competition is coming, margins are going to head down, and Tesla is going to have to continue to burn cash for many, many years to build out the production capabilities and scale up the business. Not only to maintain competitiveness, but especially to justify the stock price. We've got massive expectations for future production, future deliveries, and all of those can't be met with the current facilities and plants they have right now. So we know they're gonna have to build more. We can actually look at exactly what these scenarios we put together, what they mean for some of the production assumptions or production implied production to each of these scenarios. But the chart down here, we actually put together 3 scenarios for Tesla instead of 2 like for Netflix. But to justify that current share price, we're looking at a scenario where profits are on par with Apple, which is the most profitable company in our entire coverage universe, well above the combined NOPATs of peers. which in this case are Stellantis, Toyota, General Motors, Honda Motor, Ford, and Nissan. That's that combined group. And then Toyota down here, just to give another reference on where each of these scenarios falls and what that implied share price is. So we've got clear visual aid to show just how high the expectations are for Tesla at $185. I know it's come back a little bit since earnings. We think that has been justified given what earnings showed with Margins following the stock price, or the price cuts that have implemented this year are only going to drop prices— margins further, we think. And then as competition scales up, it's just going to turn into more of a traditional auto market, which has long been average margins, slower growth than what we're seeing with the expectations baked into Tesla. And so we can actually look at these models as well in the DCF on that forecast page there. The similar story in this case, in scenario 1, to get that current stock price, we kept margins the same. Optimistic scenario, 13% margins, which was the same as 2022. That's almost 2 times Toyota's margins. So we're not being pessimistic on the margin for what they're gonna do to justify the current price. Had revenue growing 28% compounded annually. I think the growth in first quarter was 24% year over year. So we're trying to be high on revenue growth there, which is again above consensus, 26 and 21 or so. And so you can really see just what is baked into the stock price. On the invested capital side, we have invested capital growing at 14% compounded annual growth rate over this forecast period, which is magnitudes below what they've done in the past. We think, again, trying to be conservative. But we would expect that invested capital growth to actually be higher moving forward as they have to build out more plants, build out more production. And so we really wanted to just put together a best case. The bulls definitely love Tesla and what they think it could do in the future. So we're here to show you, here's what they have to do in the future to justify that current price. So we take it out 10 years, we've got a price of $196. I think at the time this was trading around $185. That was—

let me double-check there.

Yeah, the $185. And so you could see what that implies is revenue of $941 billion at, again, NOPAT of $122 billion. So these massive assumptions for future profit growth in the current stock price. And so when we Hear people argue that they expect Tesla to grow year over year production at doubling, tripling. That's already baked into the stock price for our DCF model. And so they basically have to do more than that to have any upside for current shareholders. If you're buying at today's price, they've got to grow revenues to almost a trillion dollars 10 years out. And we looked at some of the— what kind of market cap assumptions If we assume different selling prices on vehicles, we can look at how many vehicles they would have to sell. And what we looked at specifically was we kind of take out, okay, that $940 billion in revenue. I think if they currently have automotive revenue is like 87, 88% of total revenue. So we look at that, what that implies with an average selling price, and we can see how many cars they would have to make. And so in this scenario, depending on selling prices, it would range from 12 million to 30 million, which would be anywhere from 37 to 95% of the projected EV market. You've got expectations essentially that Tesla is the entire electric vehicle market 10 years from now. And so that is what is baked into the stock price. Anything less, there's downside. And so we look at 2 more scenarios there on exactly what that downside could look like. So we looked at the neutral and the pessimistic, you know, growth slowing a little bit in later years. Instead of 28%, we looked at 23%. And then in the kind of pessimistic, growth slowing to a much slower, more traditional auto market as you get out in years 10 beyond. Revenue or margins, same thing, falling 13% down to 7.3%, which in that case I believe was equal, yeah, equal to Toyota. So again, if we look at a market that looks much more like traditional auto going forward, the downside in Tesla can be significant. Because I think in these scenarios, we're looking at $75 a share and $28 a share. In that neutral scenario, we can go out here. And I actually think I might have moved one year ahead, we were looking at 2031. But there's that 78 in the neutral, which again was margins 13 falling to 7, consensus growth, and then continuing that consensus growth out to 9 years. In that scenario, in our report, we show that at an average selling price of $50,000, they'd be selling 9.9 million vehicles, which would be 31% of the market share. So even if they can get 31% of the market share, you're still looking at a 59% downside. without some drastic change in profitability or revenues completely changing. And then on the pessimistic or kind of very conservative case, on the market looking much more like traditional autos, we see a scenario where Netflix $29, $30 a share, and that's still not as negative, I guess you would say, as it could be because we're still giving them credit for, for 4.8 million cars sold, 15% of the market in 2031, at an average selling price of $40,000. And we also are only including that invested capital growth of 14% a year, which could be much higher than that, as it has been in the past. There's definitely still an optimistic scenario to put them at $28 a share. And so we really can model out any different scenarios in the DCF and really put together quickly what we're looking at and what the stock price implies. See some questions over here.

Do you want to— some metrics questions, Kyle, do you want to take those now?

Yeah, I can actually show on the—

Great, thanks.

Model here, just the economic book value, NOPAT converting to economic book value. You can see our calculation here in the model on the economic book value page. And we start at NOPAT, we've got the WACC discount rate, some of the adjustments, and then we get down to that economic book value right there.

The basic— the basis of economic book value is NOPAT divided by WACC, right? So that's the foundational amount of what economic book value is. It's the no-growth value of the business. And then after we've divided NOPAT by WACC, we make those valuation adjustments, which are the same kinds of ones that we make in the DCF. to shareholder value, or to get the shareholder value from forward value.

Yeah, we can see that economic book value in the right below, we actually put it in a per-share value as well. If you're looking at a no-growth value for Tesla of $23. And then the other question on WACC, I'll pull up the WACC review page. Lee, I think you're more familiar with the changes in WACC, you could speak on that. But here we've got the WACC page. On the company model that explains exactly how we're calculating WACC, each of the different aspects of it, all the way down to where we're at current period and historical periods, what WACC has looked like as well. But—

Sure. Great. We calculate a cost of debt for companies and it's based in part on our credit ratings and the rates that companies are paying in the public markets from a data provider. So as we don't make any specific inflationary adjustments, but I think that because we're updating that data daily, I think that we see those changes to the cost of debt move into our models fairly quickly. Yeah, we're using the fair value that we're— all of our WACC stuff is fair value, is using fair values rather than— So it's all market value based. Yep.

You can see some of the increases in cost of debt from the 2020 lows, risk-free rate up from almost zero.

Right. So that stuff gets updated. The cost of debt after tax gets updated in 2 ways based on our credit ratings. So as the company's creditworthiness changes, we change how we score them. And we assign different cost of debts based on their creditworthiness. And then those— the relationship between credit scores and the cost of debt that we use in WACC is driven by what's happening in the marketplace, what companies that have those kinds of ratings are paying in the real market. Hopefully that answers your question. I'm happy to follow up with more detail.

Yeah, we can extend on the society as needed, of course.

Perfect.

I think that sums up what I was doing on Netflix and Tesla. Again, if you have any questions, we can definitely answer those later in this or put 'em on Circle. We can help you out there as well. But with that, I'm gonna turn it back to— I think Tammy will take over the slides and Lee can talk about the focus list updates.

That sounds great. So we talked a little briefly earlier about— someone had asked about the differences between things like our Most Attractive and Most Dangerous stock lists and the focus list. So the Most Attractive and Most Dangerous lists are ones that we've maintained at New Constructs for a very long time, since at least 2004. Those lists are algorithmically generated, so they just look at the metrics, and we select the 20 best large-cap and small and mid-cap stocks and do the same thing for the worst stocks on the Most Dangerous list. So those are driven primarily by our rating system. In 2017, we added the Focus List as a new type of model portfolio. And the Focus List is still driven by our models, ratings, and metrics. But it is a more analyst-curated list, so it is a human-managed portfolio. One of the things that we say, and I'll put this in the chat, let me go ahead and put this in there, is a quote from Martin Leibowitz that we talk about in the training of our analysts, where we say that the results of any equity valuation model should be viewed as only a first step in a truly comprehensive assessment of firm value. And we think that's true. In order to do those kinds of assessments, you've got to have good numbers and good metrics, but ultimately the value of any asset's the present value of its future cash flows. So understanding history accurately gives us a better way to understand what we think is possible and likely to happen in the future. I think you can see that kind of analysis happening when Kyle talks about Tesla and Netflix. But the human-curated list involves us being more forward-looking, doing more analysis on the position of the business in the market, what the economy looks like broadly, what the sector and industry looks like more specifically, and where we think the company's gonna go moving forward. Kyle and, and the rest of the content team do a good, a great job of keeping that list up to date and managed. It does mean that sometimes there are stocks on those lists that we have recommended that have dropped to neutral or even maybe dangerous, and Kyle reviews and maintains those on a regular basis. We made 3 updates to the focus list portfolio, the human-managed one, in the last week. We took gains on McDonald's and Oracle out of the long portfolio. Let me pull up these numbers quickly. Oracle's up 98% since we added it in 2018, compared to 48% for the S&P 500. Its ROIC has dropped from 22% down to 14%, and we think there's really only about 14% upside for the stock right at this time. So we're gonna take gains and sell it. McDonald's has slightly different situation. They are outperforming the S&P since late 2018 as well. The economics of McDonald's are sort of better than ever. They've been a positive free cash flow generator for at least 22 years. But this situation is that price appreciation since November of 2019 has made the stock less appealing to us. It's a great company, but we think maybe it's reasonably priced now rather than being undervalued. So we decided to go ahead and take gains there as well. And then on the short portfolio side, we've decided to add Lucid to the shortlist. This is a zombie stock. We added it earlier this month. I think we've talked about it in this webinar as well. So it's a zombie stock. They're destroying cash, they're destroying value, burning cash. They're missing production targets. They're laying off staff. They've got stalled growth. Their valuation implies enormous growth, at least as big as Tesla, in order to justify their valuation. So I think the hurdles to Lucid becoming a valuable stock are enormous. So we've added it to the shortlist. Anything you want to add about any of those, Kyle?

Yeah, I was just going to add on the McDonald's. That's really where the difference from a kind of quantitative numbers qualitative analysis. We start with the rating and then we look into the actual DCF forecast. And when we did that with McDonald's, we found that they would, in order to justify just the current price, they'd have to essentially get margins to something they've never seen achieved in company history and continue growing 6 or 7% a year for the next, I think, 15 years is what the model looked like at the time. And so it was really a scenario where Again, cash flows are great, but when you actually build out the DCF model, you realize just how high the assumptions are. And it was just one of those scenarios where the risk-reward was sort of good business, not great stock. So in that case, we wanted to take it off the focus list just because the expectations are kind of baked into the stock as it is. And so while they may achieve those, there wasn't any upside even if they do. And so to kind of predict or look for any upside in the company, They'd have to go well above anything they've done in the past or anything that could look future-wise margins and revenue growth.

So, and, and, and users have a couple of ways to keep up to date with those model portfolios, the focus list, long and short, and the most attractive and most dangerous and our other model portfolios. Kyle publishes updates about those to the research section of our website every time we make an addition or subtraction from any of those lists. So there's always an article you can read about any of them. Those long and short links at the top of this slide take you to those pages. And then also always on your portfolio page, you can always look at the current holdings of those portfolios or in those portfolios. We also have add dates so you can see how long we've held them. Great. Can you swap over the next slide, Dan? And we don't have as much time to talk about this one, so I just want to at least Tease it a little bit here, but we just published a long idea on Suncor Energy on our website. I'd encourage everyone to go take a look. I think it's an excellent investment opportunity. I can get a, got a couple of little points I wanted to note about it before I pass it on to Cody, but it's a Canadian oil and gas company. They do exploration, production, and they have products, oil and gas products. That's a heavily integrated firm. They say that their breakeven price for oil is in the mid-40s or over $100. That's down from the 60s for them since 2014. And they believe that they can get their breakeven price down to the mid-30s in the near future. That seems believable to us. Their NOPAT margin has been improving. In 2017, it was 13%. It's up to 21% this year. Only Petrobras in the industries has better NOPAT margins, and their ROIC has been improving as a result of that as well. With OPEC cutting their production targets more than expected, we think that it's got a price floor for oil and that there's probably around 40-plus percent upside in Suncor. So we think it's an awesome investment opportunity right now. And yeah, the report's got some of these DCF scenarios that Kyle's been talking about that you can go through that show what kinds of performance the company's got to achieve in order to justify its current stock price. It's pretty low. Its price-to-economic book value is below 1.

Yeah, I think on the flip side of the Netflix-Tesla DCF, when we look at the long side, we're really looking at how bad does something have to get for the current price? And then like this case of Suncor Energy, its current price implies that in a decade, NOPAT is half of where it is today.

Yeah.

And so that's where we get that low expectations in the current stock price. So anything above that, there's a lot of upside there. And so you can dig into that report and see scenarios we put together and how we get to that 48% upside, even in a conservative scenario.

Yeah, yeah, it looks like maybe the stock price has been pushed down by some safety concerns that Suncor's had in the past and some other news-generated things, but the company has been a cash generator for a while and it looks like they will be going forward. They're, they're set up to, to generate a lot of free cash flow into the future.

Yep.

Great. Hey, yeah, please check that article out on the website. It's a good one. And I'm going to go ahead and turn it over to Cody Fincher. He's going to talk a little bit about AIG and how we make sure that we give you the best data we can.

Yeah, thank you, Lee. Sorry about my video's a little dark here. My webcam really doesn't like the lighting in this room. So I'm going to talk a little bit today about our— how we maintain the most accurate datasets and financial models as we possibly can. Some of the methods that we use to do that. So to start it off, a few kind of key points. First one is we source all of our data directly from filings from the SEC. So we like own all of the data. We get it straight from its original source. We're not reliant on any intermediary to get that data. We use a staff of highly trained analysts that are experts at the intersection of Finance and accounting, and that's who's collecting all of the data straight from the filings. When we're going through that collection, we use over 600 different data checks to make sure that we're collecting all of the right information as best we can. So first off, what is a data check? When we are going through The filing and collecting information, that information is being inserted into a database, and data checks are effectively just queries that run on that data trying to spot any potentially erroneous information that we've collected. A really easy example of that is, say, if I'm going through a filing and I collect a value for total assets and I collect a separate value for total liabilities and equity. If those 2 data values are not the same, I'm going to get a data check result. For anybody with an accounting background, you know, total assets will always equal total liabilities plus equity. So when that's not the case, that's a situation, a very basic one, where we would get a data check. Like I mentioned before, there's over 600 of those things. They have kind of a wide range of complexity and, and what they're trying to do. But data checks are a pretty foundational aspect of what allows us to collect data so accurately. The other thing, like I mentioned before, while we're collecting data, we are also data and sanity checking it kind of simultaneously. We are also modeling the results right as we are collecting data. So we are at the time of collection also updating all of our financial models, like producing New NOPAT values, new forecasts. And that's a pretty big advantage for us in our ability to make sure that all of the data is right. Oftentimes the easiest way to find out that you collected incorrect piece of information is to plug it into a model. And if it's a really bad piece of— if it's a really bad input, you'll get a very noticeable output that shows, okay, something Kind of has to be off here. So I think that's something that actually seems simple, but it's actually pretty unique in the industry. I think oftentimes data collection and modeling are pretty divorced activities. It might be one company that does one, and then they package up the data and sell it to somebody else, and then somebody else will use that data to create a model or just do some type of analysis. And If you get bad data, sometimes it's not until it's already off to the next person until it's noticed. So we're able to do all of that kind of in-house because we are both collectors of the data, but we're also users of the data. So pretty, pretty big advantage for us in terms of getting the data right. Yeah. So I wanted to provide that background and then talk about kind of a specific example Where that type of methodology really helped us. So we've got this screenshot up, but I think I will take over the screen share for a second and navigate so I can show you guys kind of the, the tools that we use when we're looking at this. So many of you are probably familiar, like, with this portfolio page where you can see the rating. The tool that we— or the screenshots that we have up there on the slides is just from this marked-up filings tool. I think this is a really cool thing for our clients to be able to use. It basically makes every data point that we collect and every input in the model, you can audit it all the way back to the source, all the way back to the financial filing. So our clients can interact with the filing in a similar way that we do when we're initially collecting the data. It offers like an extremely high level of transparency into our processes, and we're able to do that because we feel really good that we got the data correct.

So this is—

Cody, let me stop you. I think you need to switch screen share. We're still seeing the ratings table.

Oh, okay. I'm sorry. That did pop up in a new page. Let's try this.

There we go.

Thank you, Kyle. Yep. So this is the tool I was just referencing. This is where we call this the market filings tool. And you can click around to different areas of the filing and see the data that we have collected. So this is a specific example that I wanted to talk about because I think it's pretty interesting. This is AIG. This is their most recent annual filing. And this is just a great example of how the fact that we model the data allows us to catch errors that probably no one else would catch. So the table that I have pulled up right here, this is AIG's balance sheet, and the specific data point that we're interested in is not necessarily this loan value, but the— some of the information in this line item, and that is the allowance for credit losses, basically their loan reserve. So you can see like this number right here corresponds to the loan amount of Close to $50 billion. And then they have some additional information about the reserves on those loans in this text. So it's important to understand why these reserves numbers matter from a modeling perspective. The really quick explanation of that is that differences between the loan reserve and the actual charge-offs can flow through net income. And those things can distort earnings pretty heavily because the loan provision is kind of just a management assumption. So one method that a manager can use to manipulate earnings is to kind of just play with that loan loss reserve and have a different amount of money flow through net income. So we always collect these reserves values so we can back out those Non-operating earnings that companies are recognizing in net income in order to produce like better NOPAT numbers, better adjusted numbers. So that's why collecting these reserves matters. So we always collect the current reserve as well as the previous reserve so we can calculate that difference. And this is a pretty common place for reserves values to be disclosed. They're often disclosed right in the line item that talks about loans. And you can see here, the value that was disclosed to us was that in 2021, they had $629 million of a reserve. And in 2022, that number went up to $38 billion. Pretty huge change. For context, these numbers are usually within like 10% of one another. So the fact that we're going from $600 million to $38 billion is massive and highly unusual. But that's the data that was disclosed to us. So at the time of initial collection, this is what an analyst would take because this is what the company is offering. And it appears correct. I mean, we collected the right values off the balance sheet as they were disclosed. But this gets flagged pretty immediately once you try to model these results, because what you wind up getting is Like a close to a $38 billion earnings adjustment that we're making, which doesn't really make any sense when there wasn't kind of a corresponding action going on on the income statement. So that gets flagged to analysts with our data checking process and tells an analyst that even though you collected this data, even though it's collected as it appears on the balance sheet, something else has to be going on here. So that would make us go back into the filing and look around to see what other information is available. And we were able to find something else that got us to the right number. We can find that in our income statement adjustment section. We have an area where we talk about the changes in reserves, which is here. And just clicking on this allows you to like jump around in the filing. Really cool tool. But this shows the— in the notes of the loans and the corresponding reserve. And you can see here they're giving us different numbers. It's the same number in 2021, this $629 million number. But now in 2022, they're saying $716 million. Much different than that $38 billion number they gave us on the balance sheet. What we found was that they were, in our opinion, kind of misclassifying this value here, this $37.6 billion of loan receivable. They decided to include that in their reserve disclosure on the balance sheet. They clearly made a different decision in the notes. Doesn't really make any sense why they gave us 2 different disclosures, but we were able to pretty quickly determine that this is the correct disclosure. The one on the balance sheet, in our opinion, is incorrect. So we were able to go in and collect the right data value, even though—

I hope for AIG's sake that the value from the notes is the correct one as well, because otherwise they own a huge percentage of bad loans.

Yeah, I mean, that would be going back up to the balance sheet. That's a great point, Lee, that if you just look at this in context to the loan value, they've got a little under $50 billion in loans and they're On the balance sheet, they're telling you they want to have a reserve for pretty much like 80% of it. That's, that's nuts. And if you did have 80% of your loans that you were actually concerned about, we'd be talking about AIG maybe going out of business. So why they want to disclose it that way, I have no idea. But yeah, that's just a quick kind of Behind-the-scenes look at kind of how our data checks work. A pretty interesting example of where it allowed us to get data right that was a little complicated to get right because it's, it's difficult to work with a situation where a company is just giving you bad numbers.

We've got, we've got a question here related to that, Cody, because I can pass along how they ask if we need to do require manual intervention in cases of mistakes. And this one we didn't. We were able to use the notes to to get the correct value. It's— well, it's manual intervention, I guess, in the sense that this gets flagged to analysts as a problem situation, and an analyst has to go in and correct the data as needed. The question here is like, what percentage of filings do we find these kinds of errors, and do we ever notify companies of these errors?

Do we ever notify companies of the errors? That's something that I believe we Not for this specifically, because we were able to get the right value in the notes like Lee mentioned, but I do recall a few times where we found typos in financial statements that make it to where, like, you can't reconcile net income or you can't figure out how to add up all the components to total assets. And that's something that we have alerted companies about because there wasn't really a way for us to know how to do anything with it because They just didn't give the right information in the statement itself. Typos like that are pretty uncommon. These are audited financial reports, so somebody should be catching it before we catch it, but it has made it all the way to us before.

I'd also like to note that we have in the past also tried talking with the SEC to alert them to these kinds of errors as well. Those conversations didn't really go anywhere. And my— the SEC was generally unhappy that we pointed out these errors to them because it made for more work for them. Yeah, I believe, like you said, these are pretty, pretty uncommon types of errors, but they're— but they can make a meaningful impact on the model. You got to get these things right or you're going to have bad, bad modeling results.

Yeah, absolutely. That's a great question, though.

Any other questions? I know we just have a few more minutes. I'm going to go ahead and take the screen back from you, Cody.

Go for it. Thank you.

I have one more thing I'd love to note, Tam. Absolutely. Q1 filing season is, is coming up late next week. Companies have a longer period of time where they can file their annual reports than they do for their quarterly reports. So the span between when we get— when we have annual filing season and when we have Q1 filing season is relatively short compared to other times in the year. So by the end of next week, we expect to see meaningful numbers of Q1 filings for most companies coming in. So you'll start seeing more model updates for new financial disclosures from companies here shortly.

That's fantastic. All right, folks, if there's no more questions, thank you so much for joining us on this Friday. I hope you have a lovely weekend. We'll see you out in society for further conversation, and if not there, then hope to see you again next Friday. Take care, everybody.

Bye. Thanks, everyone.