FICO: A High Score Business - [Business Breakdowns, EP.111]

0:03This is Business Breakdowns. Business Breakdowns is a series of conversations with investors and operators diving deep into a single business. For each business, we explore its history, its business model, its competitive advantages, and what makes it tick. We believe every business has lessons and secrets that investors and operators can learn from. and we are here to bring them to you. To find more episodes of Breakdowns, check out joincolossus .com. All opinions expressed by hosts and podcast guests are solely their own opinions. Hosts, podcast guests, their employers or affiliates may maintain positions in the securities discussed in this podcast.

0:45This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. This is Zach Fuss, an investor at Evernic Capital. And today, we're breaking down Fair Isaac Corporation, or commonly known as FICO. FICO is best known for its consumer credit scores product, which has become a common language across the world of consumer loans and banking. Less well -known, but a major piece of the business, is FICO's software offering that helps financial businesses with fraud detection, customer relationship management, and loan origination. Between these two offerings, scores and software, FICO earned $1 .3 billion last year.

1:26To break down the business, I'm joined by Dev Cantesaria, managing partner at Valley Forge Capital Management. In going through the business's history and business units, Dev explains why it would be tough to design a better business than FICO on paper. Please enjoy this breakdown of FICO. All right, Dev, looking forward to this conversation to break down Fair Isaac or FICO. So I thought a good place to just set the stage would be start with a basic overview of the business. What initially attracted you to the story? And then we'll have the opportunity to dive a bit deeper into the value chain and some of the nuanced aspects of the business, which you find particularly unique and underappreciated.

2:07So maybe just an overview to kick things off. We try to look for the best business models in the world. And we have been following Fair Isaac for many years. The story with Fair Isaac was a dominant franchise. Its scores business is one of the highest quality business models on the planet. But there wasn't a lot of organic growth during the early years of the company. Its growth was completely related to volume, and there was really no change to pricing of its scores product. But that changed in 2018, which got us interested in the company. If I look at what their basic business model is today, I know they have both the scores component and a software component, I want to better understand what those key products are and the value they provide to their customers.

2:54So today, the revenues of the company are evenly split between scores and software, but scores provides over 75 % of the operating income of the company because that side of the business has much higher margins. But as background, the company, as you mentioned, has two main business lines. The first is its consumer credit scores, which help with assessing credit risk for consumer loans, such as mortgages, autos, credit cards. The second line of business is software. This portion of the business delivers data analytics for things like loan originations, fraud detection, customer management, customer communications.

3:34The company is evenly split in terms of revenue between these two pieces. If we talk about scores first, FICO scores are a common language used by the various industry participants to talk about the credit quality of a loan or of a consumer. To date, it's the best way that we have to assess and monitor consumer credit risk. The company delivers over 15 billion scores every year. And as I mentioned before, these scores are used to originate loans of various types, also used to monitor loans. They are also used to market loans to certain target populations. And the scores also play an important role when you want to package loans and sell them to Wall Street in securitizations.

4:17So scores, I would describe them as an important common language that all of the industry participants speak. Today, over 250 million Americans can be scored using a FICO score. And scores are very inexpensive, generally less than a dollar each. And some scores are even a penny or a fraction of a penny, depending on the use case. To give you an idea of the far reach of FICO scores, they're used in over 90 % of the credit lending decisions. They're used in over 99 % of credit securitizations. 90 of the top 100 largest U .S. lenders use FICO scores. And it's not just lenders. There's over 700 insurers that use FICO scores, a third of all retailers, more than 200 governments, more than 150 pharma and healthcare companies.

5:07So FICO scores have found many uses over time. I think it's important to not differentiate, but perhaps explain. There are the credit bureaus, which are in and of themselves great businesses, and then there's the FICO score business. How do the two businesses work together and differentiate and perhaps compete in ways as well? I think some of the history of the industry is important to go through here. Before the 1960s, there was a significant character component to lending decisions. You could imagine a farmer in a small town walks into a community bank looking for a loan. He didn't have audited financials of his farm.

5:45He didn't have a credit card history. There was no easy way to know what all of his outstanding loans or liabilities were. And so the loan decisions back then were largely based on trust and familiarity and the value of the collateral that you were loaning against. There were very few objective factors to work with. And so most of these loan decisions were highly subjective. There was also no concept of a variable interest rate depending on your credit risk. So you would walk into a bank and they would have a single rate displayed for what a home loan would be, and the bank would have to make a decision of simply yes or no.

6:20So Bill Fair, who was an engineer, and Earl Isaac, who was a mathematician, they worked together at Stanford Research Institute. And they saw an opportunity to bring some standardization and objectivity to the lending process. In 1956, they started a consulting firm together called Fair Isaac & Co. Each of them contributed $400, and it began in a studio apartment in San Rafael, California. And the story is that they sent letters out to the 50 largest lenders in the U .S. explaining their ideas, but only one responded, and that was the American investment company, AIC. So this company hired them to analyze their credit files.

6:59And in the beginning, the factors used to assess credit risk were quite primitive, like age, occupation, income. What they really needed to refine their models were large data sets. So combining the mathematical models with large data sets, consumer characteristics, default history, and this was important for them to refine and improve their models. So Fair Isaac sold its first credit scoring system in 1958. And if you look into the 1960s, AIC and other similar companies began using rudimentary computer -derived credit scoring. Companies had to send surveys to prospective borrowers, but consumers didn't have the patience to provide 50 -plus variables.

7:38The scoring models needed to focus on 10 to 15 factors that were most important. The scoring system is also back then were not universal. GE Credit Court, which was a large lender at the time, might have had 15 to 20 different scoring systems depending on the underlying populations and the types of loans that they were looking at. The scoring systems were extremely helpful to lenders because they could tighten or loosen standards based on changing the cutoff score level. This is something that they didn't have before. The lenders could also compare loans across locations. They could track them over time to get an accurate sense of overall company risk.

8:16Prior to credit scoring, there was no easy way to assess the changing credit risk across thousands of home loans, for example. They also now had the ability to charge variable interest rates based on the credit risk. So this was the initiation of the subprime lending market. In 1974, the Equal Credit Opportunity Act was passed, and this accelerated the need for some type of objective scoring in loan process. At that point in time, it became illegal to discriminate on factors such as gender or marital status. Lenders needed to show regulators and consumers some objectivity in the process. So through a series of mergers and acquisitions in the 70s and 80s emerged the three main credit bureaus that we have today, Equifax, TransUnit, and Experian.

9:01And the first standardized credit bureaus risk score, Prescore, was released in 1981. Prescore incorporated the credit bureau information for the first time. So FICO had the mathematical models. The credit bureaus had the consumer data. These companies were combining that to form Prescore. Previously, it was the banks and retailers that had this client data that were fed into the FICO models. And so the scores for a long time were custom built for certain banks. And these models were not shared amongst the various lenders. They were very specific to each bank or each financial institution. Prescore was the first generic model that was not custom built.

9:41The scores were generalizable across lenders and a wide variety of loan decisions. And the scalability also lowered the cost of scoring, which could now be standardized across the entire industry. The score that we know today, the modern FICO score, was released in 1989. That's a score that ranges from 300 to 850. It is said that the 1 to 300 range was left for another scoring product, which I don't think we've ever seen. And that 0 to 99 was meant for special codes to identify incomplete data. People often wonder why a score has this odd range of 300 to 850. And then I think the other critical industry development to mention is that in 1995, Fannie May and Freddie Mac institutionalized FICO scores by making it required for all mortgage originations that they would handle.

10:33So a business with a very interesting history that's settled into this monopoly -like structure, to me, lending seems pro -cyclical. Mortgages, auto loans, I would assume volumes tend to rise when the economy is booming and rates are declining or volumes decline as the economy slow. In a rising rate environment, people may be skeptical of the continued trajectory of that growth. What is it about this business that allows them to grow at a steady pace despite what would be natural cyclicality in my uninformed view? Now, you are correct in pointing out that the Scores business, the business -to -business side, is cyclical.

11:12And when there's a recession, there's less mortgages, less auto loans, less credit cards being issued. And so the best data point we have is going back to the financial crisis. In the 2007 to 2009 timeframe, scores revenue declined about a third. That is one of the reasons that people have a negative view of the company at times. But overall, scores volumes will increase over time due to the many additional use cases we have for scores. And due to the recent change to how they price scores, and we'll talk about that maybe a little bit later, they're able to add special price increases across different categories of scores.

11:51And that's allowed the company to maintain revenue or even grow scores revenue despite volumes being flat or down. Perhaps a related question, or maybe adjacent, how does the software offering coexist or partner with the score side of the business? Or are they somewhat independent of each other? The software business can be a standalone business. Many investors have thought that there was a benefit to separating the two business lines. There is some cross -selling. Many of the software solutions are sold to banks and financial institutions, but those are likely different departments. I believe that they can largely be standalone businesses, but Scores generates a lot of free cash flow.

12:38And at the moment, the software side requires heavy investment. Now, most of that heavy investment occurred over the last five years, and that investment is declining. But there is at the moment a nice financial complement of the two businesses in that you have a Scores business that has very high operating margins, north of 85%, with its special price increases growing steadily in the mid to high teens. And then you have a much younger platform software business that has been created where very fast growth, but it's young and it requires investment. The cashflow from the score side helps support the software side, but not a lot of direct complimentary selling or cross -selling between those two businesses.

13:22To focus the conversation on the scores business, just given how robust that offering is and the opportunities set ahead of them, We haven't mentioned much in the way of competition in that business. My understanding is they have a pretty dominant market share, but who are the competitors and how are they organized? So let me start off with some background on the Scores business. The Scores business revenue is about two -thirds business -to -business or B2B, and one -third of the business is business -to -consumer or B2C. The latter is far less cyclical. The B2B Scores business, there's mortgage scores, there's auto scores, and then there's credit card and personal lending scores.

14:02Those three categories are all roughly the same size. The cost of the score depends on the type of score, whether it's mortgage, auto, credit, and how that score is going to be used, whether it's going to be used for originations, monitoring, or marketing. Mortgage origination scores carry the highest pricing, and marketing scores can be less than a penny each, and those carry the lowest pricing. Scores today is a very U .S.-centric business. So over 95 % of scores revenue today comes from North America. There's a small contribution from areas like the U .K., Brazil, and China. And if you look at what a FICO score is, it's composed of five major factors.

14:42So for a consumer, that would include payment history, the amounts that you owe relative to the limits that you have, the length of credit history, the frequency of new credit, and the types of credit used. And so those are all weighted to form your FICO score. But it's very interesting how FICO scores are sold. FICO scores actually have three layers of customers. It's important to mention again that FICO only produces the mathematical models, and these are then combined with the credit bureau data to create the scores. So Fair Isaac does not gather or retain any consumer data of its own. So the direct revenue for scores comes from the credit bureaus, which represent about 75 % of scores revenues.

15:26But the end customer is, in fact, the banks and financial institutions which are making the credit decisions. The credit bureaus sell consumer credit reports to these banks and financial institutions. The FICO score generally represents less than 5 % of what is called a tri -merge report. That is generally about $50. So a tri -merge report is simply all three credit bureau reports combined together in one consolidated report. So the three credit bureaus are simply passing along any price increases that FICO makes. And they often layer on their own markup at essentially 100 % margins. So the credit bureaus, interestingly, although they're the customer of FICO, they're agnostic to FICO's price increases.

16:13The financial institutions will usually take the costs of the credit reports and include those in the closing costs that consumers pay. So if you go for a mortgage, you know there's a large list of closing costs. Well, the credit reports are often part of what the consumer pays. So although FICO is increasing prices at a higher level, you have to go many layers deep before you get to who actually pays for those price increases. It's worth mentioning that scores are extremely cheap relative to the loan sizes. So you can imagine on a $400 ,000 mortgage decision, a FICO score is less than a dollar.

16:49And if you look forward, you could say, well, listen, even if a FICO score costs 10 times more, let's say $5, it still offers tremendous value to the ecosystem or the loan originator. Prior to 2016, FICO had not increased the prices of its scores for 25 years. As I mentioned before, their growth was entirely based on changes in volumes of scores. The company began instituting special price increases for its various scores in 2018. They first started with mortgage scores and raised prices there 30%. And then the following year, it was autos by roughly that same amount. And the year after that, it was the credit card scores.

17:30And they continue to rotate these annual price increases across the different categories. These special price increases add about $60 million of annual revenue to FICO. and they come along with incremental margins of 95 % plus. Now, I wanted to talk a little bit about the different types of scoring models. As I mentioned before, there's about 230 million Americans that can be scored using the standard FICO score. But the eligible population has been increasing, especially with the introduction of their new FICO 10 suite. The company launched the FICO 10 suite in January 2020, and they say it's about 10 % more predictive than the previous score, FICO 9, which launched in 2014.

18:15All of this is relevant to get to your question about competition. Let me mention a few more things about FICO scores today. They come in a lot of different flavors. So FICO 10 comes as also FICO 10T, which uses more recent trend data, like what has been happening to this consumer over the last 60 or 90 days. FICO XD uses cable, cell phone, and utility bills to assess credit risk in people that have limited credit history. This particular variety of score adds another 12 % of the potential population that is scorable using FICO scores. Ultra FICO links to bank accounts to assess activity and balance sheet strength, and that score can add 20 points to your regular FICO score.

18:59There's also a FICO resilience index, which identifies customers who are more resilient to economic stress relative to other consumers that have the same FICO score. And there's also industry -specific scores. So there's the FICO auto score and the FICO bank card score. The reason I'm telling you all of this is when we talk about competition, the main competitor for FICO scores is Vantage score. And Vantage score launched in 2006 by the three credit bureaus. And it has tried to differentiate itself by saying that it can score 40 million more individuals than the standard FICO 9 score. These are mostly individuals with thin to no credit history.

19:40With the launch of the FICO 10 suite in January 2020, FICO 10, Ultra FICO, FICO XD, the difference in the number of scorable people between FICO and Vantage score drops to about 10 million people. But what matters is the credit eligible individual Those are the people that can actually receive loans. And that difference falls to zero to two million people. So although Vantage Score launched and has tried to differentiate itself based on this purported advantage of being able to score more individuals, that has been largely nullified by the recent rollout of the FICO 10 suite. FICO is dominant in admission -critical applications, such as originations and monitoring of consumer loans.

20:25Vantage score has significant share in areas like marketing and lead generation, where you have the cheapest scores, where it's not as critical that the decision is right. Importantly, in 2021, a major company in the industry, Synchrony, which is the largest provider of private label credit cards in the US, switched advantage score from FICO after a three -year transition period. And at the time, this caused quite a stir. But after the implementation process, analysts estimate that Synchrony will only save $2 to $3 million a year on a revenue base of $18 billion. What they lost by switching away from FICO scores was the ability to, for example, securitize their debt at attractive interest rates.

21:11So given the limited risk reward of switching, I see very few companies following Synchrony's path. I would love to peel back the onion a bit on their inability to securitize their originations by removing the FICO score. How exactly did that play out? And were they aware of that when they made the decision to switch? They did ultimately have a debt offering that used Vantage score. It's very analogous to the rating agency business of Moody's and S &P Global. 90 % of all the debt in the world is rated with Moody's and S &P. If an offering uses another provider, usually the issuer ends up paying about 50 basis points more in their interest rate.

21:51So there's really no way to undercut S &P and Moody's. We could create our own credit rating agency to compete with Moody's and S &P and offer our services for free, and it still wouldn't make sense to use them. So the similar concept within securitization is that FICO's are the accepted standard for risk assessment. And if you go to Vantage Score or other methodologies, although there isn't a definite scientific number of what you will be paying additional interest, it will be viewed as a lower quality offering and there will be additional interest to pay. I can't quantify it, but probably in the range of zero to 50 basis points.

22:26It's eminently rare to find a business that benefits from scale economics, network effects, very high switching costs, it appears, all at the same time. And it's quite remarkable how they established themselves. I guess vantage scores is one competitive threat. I should probably ask the obligatory question around artificial intelligence, just given how incredibly topical it's become and how that could impact the competitive advantages the business have. How are they thinking about potential competitive risks in that way? Score predictability is measured using the Gini coefficient. And the Gini coefficient is a ratio that represents how close the model is to being a perfect model and how far it is from being a random model.

23:09The perfect model would have a coefficient of one and a random model would have a coefficient of zero. So if you look at this coefficient for some of the recent scores, Vantage score 3 .0, 4 .0, FICO 8, FICO 9, the coefficient for all those scores is around 0 .70. And that coefficient has not improved in essentially 15 years. So we have reached what is close to peak predictability for consumer credit loan decisions. Now, Fair Isaac claims that FICO 10 may be about 10 % better than the previous scoring models. FICO 8 was introduced in 2009. FICO 9 was introduced in 2014. Interestingly, the older scores are still heavily used by banks and financial institutions.

23:55As recently as 2021, the majority of FICO scores were still based on an old model FICO 8. There has been a concern that some new companies, Upstart is a good example. These companies are applying their own algorithms in addition to using FICO scores. So what's interesting is that Upstart is not replacing FICO scores. They actually use FICO scores in their methodologies, but they're layering on top their own algorithms, their own data. Although it's been spun as something new and exciting, banks have been doing this for decades, right? Banks have always had their own proprietary data, their own client files, their own algorithms.

24:36They've always been adding to the FICO scores, never relying on FICO scores alone. But it's an interesting question on whether AI can improve predictability in the future, but to do it without potential discrimination. That's quite important. And that's one of the advantages of something like FICO score is on race, gender, marital status, where you live, the type of job you have. FICO scores can be objective, whereas an AI program may not. So it does come with risks. You mentioned the example of Synchrony potentially looking to switch in their trials and tribulations. My understanding is that there's a transition at Freddie and Fannie as well that potentially impacts the business.

25:20I'd love to unpack that a bit with you as well. There was an overhang on Fair Isaac's stock for a couple of years as we were awaiting the decision from the FHFA, the Federal Home Finance Agency. And they looked at the situation, felt that Fair Isaac had a monopoly on scoring of mortgage originations that went through Fannie and Freddie. They decided that they were going to have a review process. And investors were quite worried that this could affect Fair Isaac's dominance in mortgages. What ultimately came out as a decision was a bit of a surprise to most people. The FHFA decided that they were going to require the use of both FICO 10T and Vantage Score 4 .0.

26:03So it doesn't affect the use of FICO scoring. It now requires the use of an additional score that is owned by the three credit bureaus. But the credit bureaus that were pushing for Vantage Score here may now be regretting that fight. Because what happened was that FHFA reduced the requirement of credit bureau reports needed for mortgages that go through Fannie and Freddie. So it used to be the case that you would need three credit bureau reports for these mortgages. And what the recent decision has decided is that they only need two slots or two reports. So what it has effectively done is reduce the revenue pie of mortgages by a third.

26:44And now the three credit bureaus have to compete with each other on every mortgage for the two slots. And so the minuscule incremental revenue that they gain by getting Vantage score inserted into the mortgage process, that small benefit, I think, is heavily outweighed by the lost revenue from going from the three bureau reports to the two credit reports for the mortgages. One part of the business we haven't spent a ton of time on yet is the B2C side of the business. And it's rare that you have such high brand awareness for something that's financially derivative and complex and kind of behind the scenes.

27:23But I presume that the non -aided brand awareness of a FICO score is actually incredibly high for a financial product. And they've been able to leverage that brand to create a very healthy consumer business. Can you just explain how that business works relative to the Scores business and what's different and exciting about it? The B2C part of the Scores business is about a third of the revenue of the Scores business. And the B2C Scores part of the business is composed of two major pieces, MyFyco and partnership licenses. The former MyFyco, we estimate, represents about 40 % to 45 % of the total B2C business.