Bank OZK sounds terrible on paper: an Arkansas-based regional bank with a large concentration not only to Commercial Real Estate, but specifically to the higher-risk construction loans. The research analyst at Citigroup covering the stock has not contributed to calm investor’s nerves, raising the alarm on a massive $915m construction loan that, in his view, is at risk of impairment.

However, we believe that these concerns are grossly overstated, and that they provide an attractive entry point to acquire one of the best-run and most differentiated financial institutions in the US. We will dig deeper throughout this write-up, but for now we will just present four facts:

       I.          In its 27 years as a publicly traded entity, Bank OZK has compounded at 17% p.a., vs 8% p.a. for the S&P 500.

      II.          Current CEO George Gleason took control of the bank in 1979 when he was 25. He is now 70, working 10 to 12h every day, and has stated his determination to stay for another 10 years.

    III.          Despite its focus on construction lending, Bank OZK has averaged, during the past 45 years, 63% less loan losses than the average US financial institution. Also, the bank has not lost money in any single year, not even during the GFC.

   IV.          Its current balance sheet would be able to absorb losses accounting for 22x their current level of non-performing-assets, or 6x the peak level of loan losses experienced during the GFC.

At 1.1x Tangible Book Value and 6.8x 2024 Earnings, we believe Bank OZK is a safe bank to own, and it is likely to outperform the S&P 500 over the coming decade.

Contents

1. Bank OZK’s fit with Red Beryl’s Investment Strategy

2. Opportunity Overview

3. CEO Referencing

4. Overview of Commercial Real Estate Lending

5. Bank OZK’s Real Estate Solutions Group

6. Negative Report from Citigroup

7. Diversification efforts

8. Risks

You can find this write-up in PDF here → https://bit.ly/RedBeryl_BankOZK

1. Bank OZK’s fit with Red Beryl’s Investment Strategy

In our first write-up since launching Red Beryl, we covered Nubank, the largest digital bank outside Asia. For all the reasons presented, we think Nubank is a uniquely compelling opportunity, but being based in Brazil and with a $60Bn market cap, it is not a good reflection of what we expect the average Red Beryl investment to look like.

Given we do not have the burden of managing a large amount of capital, we spend most of our time researching smaller companies, often in the low hundreds or even tens of millions in market cap. We believe that fewer eyeballs focusing on this segment are likely to result in a more fertile hunting ground for our efforts.

However, having now been live for 3 months, we must admit that, so far, we have failed miserably at finding such opportunities. We did come across a number of smaller businesses trading at optically low valuations, but in all cases where we spent proper diligence time, we reached the conclusion that the low valuation was warranted, in most cases, due to a lack of moat vs larger players. Perhaps this is the nature of small cap investing after all, but so far we have not managed to gain comfort on any of our ideas. We will keep trying.

Bank OZK is, instead, the product of a different exercise. Given strong negative sentiment on Regional Banks and Commercial Real Estate, we have spent time on the space, trying to find the one high-quality bank whose valuation is irrationally brought down due to industry headlines, as opposed to bottom-up research. With this renewed approach, Bank OZK was not hard to find: it trades as the bank in the S&P Total Market Index with the lowest 2024 P/E valuation. What follows is the result of a diligence exercise that has led us to believe that, contrary to what its valuation might suggest, Bank OZK is one of the best-run banks in the country, with a unique moat, sound lending principles and a healthy loan book.

2. Opportunity Overview

Bank OZK is the modern name for Bank of Ozark, which began as a one-branch community bank in Jasper, Arkansas in 1903. In 1937 it opened a second location in Ozark, Arkansas, and operated as a two-branch community bank until 1979.

In 1979, current CEO George Gleason, back then 25 years old, took control of the two-branch bank, financing it with $10k of his own funds and borrowing $3.6m from a client of the law firm he used to work for. When he took over, Bank of Ozark had 28 employees and $28m in total assets.

The company expanded as a community bank in Arkansas, and in 1997, with 11 branches and $287m in total assets, the company IPOed on Nasdaq. By 2005, having reached 54 branches in Arkansas, Bank of Ozark started an expansion into Texas, opening its first 3 branches in the state, and marking the first step of Bank of Ozark’s regional expansion. Recently rebranded as Bank OZK, the company today has $34Bn in total assets, making it the 55^th largest bank in the USA. Its market cap is $4.7Bn.

On the deposit side, Bank OZK is not particularly differentiated. It operates as one more of the multiple regional banks in the US, raising deposits from a network of 228 retail branches in the Southeast. Its cost of funds is in line with industry average: if anything, given Bank OZK’s above-average rate of growth, it oftens needs to offer slightly higher remuneration than average to attract new deposits with which to fund the increase in outstanding loans.

Instead, where Bank OZK differentiates is on the asset side, using those deposits to fund four different lending operations, the largest of which is unique across the US:

       I.          The Real Estate Solutions Groups (RESG) is the bank’s crown jewel. It accounts for 65% of the loan book, and it is a highly sophisticated nationwide operation focusing on construction loans of very large and highly complex commercial real estate projects. It has best-in-class return on assets, and has performed with a very low level of loan losses throughout its history due to (i) funding well-capitalized developers with very strong track records (ii) engaging only in high-quality Real Estate projects that Bank OZK thoroughly diligences and (iii) offering very low loan-to-value ratios of around 40%. We will deep-dive into this unit later in this report.

      II.          The RV and Marine segment offers, through a network of specialized dealers, loans to high-net-worth individuals for the purchase of different types of recreational vehicles, such as boats, campervans and motorhomes. They focus primarily on customers that have already owned two or more RVs in the past, so that they understand the financial burden it represents.

    III.          The Community Banking segment offers primarily loans to small and middle-market businesses in the states where Bank OZK has a physical presence.

   IV.          The Corporate segment offers primarily Asset Backed Loans (ABLs, secured by either Inventory or Accounts Receivable) and Equipment Finance loans to large corporations nationwide.

The Real Estate Solutions Group (primarily) and the RV and Marine division (to a lower extent) are unique and differentiated, with higher barriers to entry than most other banking products. The Community Bank and Corporate segments, on the other hand, are much more commoditized and not differentiated from what other regional banks regularly offer.

Given Bank OZK’s outsized exposure to construction loans, which are usually considered riskier, it will likely come to no surprise to the reader that Bank OZK obtains a higher average interest rate on its loan book than the average US financial institution. This spread has averaged 110bps over L20Y.

However, despite Bank OZK’s loan book commanding higher yields and apparently entailing more risk, the company has an exceptional track record of credit quality. When comparing loan losses across its entire loan book, the company has outperformed the average of all other banks in the US in every single year since IPO, averaging 63% less loan charge-offs than the industry.

The result of being able to offer loans with higher yields and fewer losses than the industry, combined with an adequate control of the cost base, results in Bank OZK having a much higher return on assets and much lower cost-to-income ratio than the industry, pointing to a high-quality banking operation.

Given the preceding discussion, the reader is likely to have assumed that the Return on Tangible Equity of Bank OZK would be best-in-class. And indeed, between 2002 and 2011 it far outstripped that of the average US banking industry, averaging ~20% RoTE in the decade that included the GFC.

However, in the last 10 years, RoTE has been meaningfully lower, averaging only 14%, even if Return on Tangible Assets has been roughly constant through the cycle (Figure 6). The explanation to the falling RoTE can be found on a steep reduction on the leverage ratio of the bank (how much assets a dollar of equity supports). As can been seen in Figure 8, pre-GFC the bank held 15x as many Assets as it held Equity, with the ratio falling to 7x for the L5Y.

The reason for the strong deleveraging of the bank is primarily attributable to regulation:

       I.          The Basel II and Basel III accords generally increased the capital requirements for banks.

      II.          Bank OZK has evolved into a model whereby the construction loans it offers are drawn progressively as the construction evolves. This provides additional safety for the bank, given strict covenants around completing the successive construction phases on-time and according to budget to unlock the next loan tranche. However, for regulatory purposes, having those undrawn commitments results in an increase in risk-weighted assets and therefore the requirement to hold additional capital, even if the undrawn portion of the loan does not provide interest income.

As per Figure 6, Bank OZK has about twice the Return on Asset as the average US Bank. However, as per Figure 8, it has roughly half the leverage ratio. As a result, the RoTE comes in at a roughly similar level as all other US Banks (120 bps lower for the L6Y, to be precise).

We believe that the current regulatory framework over-penalizes Bank OZK for having large undrawn commitments, which are a unique characteristic of its business model. As discussed in Section 5, this practice de-risks the loan book, as further evidenced by having 63% fewer charge-offs than the industry for the L20Y (Figure 6).

However, shareholder returns are seldomly affected by fairness considerations, and there is no doubt that a lower RoTE translates into softer long-term performance. There are only 2 partial mitigants:

       I.          Bank OZK has historically distributed ~50% of the earnings as either dividends or share buy-backs, and therefore the lower RoTE applies only to the undistributed portion of the earnings.

      II.          In the years 2018, 2019 and 2021 market participants took, in the CEO’s view, excessive risk when it comes to CRE loans. Hence, Bank OZK pulled back, reducing leverage and origination volume to avoid taking undue risk. The leverage ratio on those years fell to as low as 6.5x. The excessive risk-taking translated into the CRE crisis the US is now experiencing.

The ongoing stress in CRE loans has resulted in reduced competition since 2022, leading to more attractive loan terms and Bank OZK increasing originations and leverage (above 8x today). As a result, in the near future, RoTE will likely be stronger than it has been in L6Y (17% in 2023 vs 13.5% in L6Y). In the long-term however, periods of excessive risk taking and, hence, lower opportunities for Bank OZK will almost certainly re-occur.

All things considered, our investment thesis, which we will attempt to prove over the coming sections, is the following:

       I.          Bank OZK is a safe bank to own, despite its concentration in commercial construction loans, given (i) lower leverage and two-thirds fewer loan losses than the average banking institution, (ii) the specific type of construction loans it offers, (iii) a bright owner-CEO that has a 45 year track record and that intends to stay in the business for another 10 years and (iv) no interest rate risk and (v) a deposit base with a very high insurance ratio.

Also, note that its current balance sheet would be able to absorb losses accounting for 22x their current level of non-performing-assets, or 6x the peak level of loan losses experienced during the GFC (Figure 23).

      II.          Through the cycle, Bank OZK will continue to earn an average of ~13-15% RoTE, which, combined with a current valuation of 1.1x TBV, yields a base return of the same magnitude.

As discussed in Section 5, in our view, Bank OZK has a strong moat and engages in a far more differentiated activity than most other banks. This does not lead to better RoTE than industry average due to regulatory capital constraints, but it does provide us with a higher level of comfort that Bank OZK “deserves” to earn that RoTE and will be able to defend it in the future, which supports point (II) of our investment thesis above.

    III.          The valuation is currently depressed due to the ensuing CRE crisis and a highly negative (and unwarranted) recent equity research report from Citigroup. A re-rating to close the valuation gap to either historical levels or the peer group provides a source of upside. For example, an undemanding increase from 6.8x to 9x Earnings in 5 years would boost shareholder returns from 13-15% to ~20%.

With the S&P500 trading at 23x 2024 Earnings and a L20Y revenue-per-share CAGR of 5% p.a., if point (I) above is true, we believe Bank OZK is likely to outperform the index in the long-term.

3. CEO Referencing

Banking is a highly execution-driven business, where poor management can introduce an undue amount of risk. This is especially true in the case of Bank OZK, given its concentration towards construction commercial real estate. We draw great comfort from the fact that current CEO George Gleason has the demonstrable track-record outlined in Section 2, which he started at the very young age of 25:

“I arrived at the bank the morning after closing the transaction. I had called a meeting for my new staff at 7:30, to introduce myself and meet them and, and I gave them a little speech. […] Afterwards, I went to my office and I was thinking, gosh, I wonder what the Chairman CEO of a bank does at 8:30 on the morning?” – George Gleason, Bank OZK CEO

We have not been able to meet George Gleason directly, but we have referenced him through other stakeholders. He refers as a highly bright individual:

“George was the most intelligent human that I have encountered in my career, he is absolutely unbelievable. Photographic memory, incredibly intelligent […].  George worked harder than anyone at the bank and was more driven than anyone at the bank. That culture migrated its way down to the lower levels, it was a culture of performance.” – Former Chief Credit Officer, Bank OZK

“Attending a meeting with George was intimidating. You knew that even if you had spent hours preparing, he was going to ask the one relevant question you had not thought about.” – Former SVP of Construction Lending, Bank OZK

“Long ago when I was on the sell-side, not long after the financial crisis I spent a day with Bank OZK’s CEO George Gleason meeting clients. To this day he remains among the most impressive managers I have met. In listening to perhaps 200 questions asked of him all day he was different in how he tackled problems and opportunities. Gleason is detail oriented, holds a higher standard than most bankers, and he thinks. I recall when he described his 45% efficiency ratio (a top quartile number) as “terrible” in a call years ago. Most bankers would not have the proper idea of where the components of the efficiency ratio should be to form that opinion, much less be willing to describe that level as bad.” – Sam Haskell, former Research Analyst

George owns ~5% of Bank OZK, a stake worth ~$250m. He has had a number of liquidity events, but his stake in Bank OZK is likely to still constitute a large percentage of his net worth (he owns roughly the same number of shares as he did back in 2005). He has invariably acted in a highly shareholder-friendly manner. For example:

       I.          The bank has been highly rational about capital issuance and share repurchases. Between 2012 and 2018, with the bank trading at north of 3x TBV, they issued a large amount of capital to grow and acquire distressed banks in highly accretive transactions. On the other hand, in L3Y, with the bank trading at depressed valuations, they have implemented frequent buy-back programs (retired 13% of outstanding shares since 2021).

      II.          As discussed in Section 2, the bank de-levered in periods of excessive risk taking (prior to Covid and in 2021) and leaned in during periods of uncertainty (Covid and after Ukraine war).

    III.          The level of disclosure is incredibly granular and one of the best across all US Banks, as highlighted by the large amount of data the reader will find in Section 5.

George is currently 70 years old, but he has no intention to retire. He has publicly stated his goal to stay with the bank for at least another 10 years, and our conversations with insiders seem to confirm it:

“The bank is his number one priority. He loves it, he lives it. I told people as a joke, the worst day for George is a bank holiday because he can't work. And he just loves it. It's his baby. It rules his life. As he's gotten older, what I saw him do is make sure he eats incredibly healthy. He works out all the time. He's got a full gym in his house. He is regularly going to doctor appointments. So he's trying to look at it and say, really and truly, I love it. And I want to do this for another 10 years, until I'm 80 if I can.” – Former Chief Credit Officer, Bank OZK

“George is still working 10-12h every day. There is not a loan above $10m that he does not personally approve. He covers a lot of ground and knows everything that is happening in the bank, and I don’t see it changing anytime soon. He sees the bank as his life project” – Former SVP of Construction Lending, Bank OZK

“I'm trying to advance more and more team members to higher levels so that when I retire in 15, or 20 or 30 years, there is a strong leadership in this company.” – George Gleason, Bank OZK CEO, in a 2022 interview.

Finally, were he to retire, we believe it is unlikely to substantially affect the investment in the first 5-10 years after retirement:

       I.          George is likely to retain a strong stake in the bank, and therefore ensure the transition to a new leader is smooth. President Brannon Hamblen would be a likely candidate, and he refers well based on our industry conversations:

“Brannon is very hard working and really smart, it is the only way someone can work directly for George. He might not be the most growth-oriented guy, but he is incredibly switched on towards risk. He is extremely conservative, and I think he is very unlikely to do anything dumb with the bank, if he is appointed CEO one day.” – Former SVP of Construction Lending, Bank OZK

      II.          A new leader will not change the existing loan book overnight, and given average loan duration of 3-5 years, it will take time for newly originated loans to become a majority of the loan book.

    III.          In the downside case where (I) George becomes incapacitated, (II) a poor manager was appointed and (III) that poor manager changed the origination process overnight, we believe that as investors we would likely realize. As discussed, the level of reporting of the bank is incredibly granular, and an increase in LTV or other poor practice would likely be observed. If the new manager reduced the reporting quality, that would be a very strong red flag and likely compel us to divest.

All things considered, we like the leadership in place at Bank OZK and we believe that for the coming 10-20 years, mismanagement is highly unlikely to occur at Bank OZK.

4. Overview of Commercial Real Estate Lending

4.1. Overview

Before proceeding to deep dive into Bank OZK’s lending activities in Commercial Real Estate (CRE), a general overview of the space is of relevance.

Loans backed by Commercial Real Estate are a wide category that includes basically all real estate that is not single family residential (i.e. houses owned by individuals). There are two main types of CRE-backed lending:

       I.          Owner-occupied: loans to companies to finance real estate that they need for the normal development of their business activity. For example, an office to accommodate their employees, a factory, a warehouse, or a retail premise to sell their products.

      II.          Non-owner-occupied: loans to investors that do not use the property themselves, but instead

lease it to a third party and capture rental income.

Owner-occupied CRE-backed loans are considered very safe, given they are usually full-recourse to the borrower. In order for the lender to impair a portion of the loan, the borrower must enter bankruptcy proceedings AND the financed collateral not be enough to repay the loan. This type of loan is not the traditional CRE lending that is usually referred to. They are instead classified as C&I (Commercial and Industrial), and are outside the scope of this report.

CRE lending traditionally refers instead to non-owner-occupied loans. They include all kinds of properties, the most common ones are Multifamily (more than 4 residential units built for rental income), Condos (more than 4 residential units built with the end goal of selling the individual units), Senior Housing (either independent living, assisted living, nursing homes or memory care facilities), Office, Life Sciences (same as office but with Lab space for pharma or biotech companies), Retail, Industrial and Hotels.

Based on the status of the building, there are three types of CRE loans:

       I.          Construction loans are backed by land and unfinished buildings, as well as guarantees from the developer. They usually carry low loan-to-cost ratios to ensure that the developer has sufficient equity invested in the deal to avoid losses for the lender. They are usually variable rate (e.g. SOFR + 350bps).

      II.          Bridge loans are backed by buildings that have already been completed but that for some reason have not stabilized yet, i.e. the building is not yet generating cash-flow at its full potential. Bridge loans are used for refurbishment, to fund the period between construction has finished but leasing has not yet occurred, or for acquisition of a property that has a high vacancy rate while the investor attempts to turn it around. Similar to construction loans, they are usually variable rate, and may include guarantees from the developer.

    III.          Permanent loans refer to the long-term form of financing offered to investors who own buildings that are fully stabilized (i.e. leases are in place) and, therefore, cash generative. They can either be fixed or variable rate (more often than not, they are fixed), and they usually are non-recourse to the investor.

4.2. Permanent Lending

Permanent loans on good quality CRE assets are commonly considered safe, and therefore they conform the majority of CRE loans extended by traditional banks. For example, a 70% LTV loan for a luxury apartment building in Miami that has a vacancy rate of 3% and a Debt Service Coverage ratio of 1.25x is unlikely to lead to loan losses in a normal scenario, given rent is likely to grow over time and a rapid increase in vacancy rate is unlikely, resulting in the developer being able to repay the loan with the rental income without the need of adding additional equity.

There are mainly two ways a lender can lose money on a permanent loan:

       I.          The rental yield declines because of higher vacancies and lower rents. This can happen either because a specific area loses appeal (e.g. Detroit), an economic change lowers the appetite for a certain form of real estate (e.g. retail buildings with the advance of on-line shopping, or office buildings with the advance of work-from-home) or in the midst of a downturn if some of the lessees default (e.g. tenants become unemployed, office-space renters go out of business, etc.)

      II.          Interest rates increase and the value of the building falls below the original LTV. Note that the value of a building is computed capitalizing the net rental income of the business with a “cap rate”, which is essentially the inverse of the P/E multiple. For example, a multifamily building generating $1m in net rental income is worth $25m if using a 4% cap rate ($1m divided by 4%), or only $17m if using a 6% cap rate ($1m divided by 6%). If a bank had lent $20m on the basis of a $25m valuation, and the building is now worth $17m, the loan is likely to suffer a loss because the investor has an economic incentive for walking away if either (i) the loan comes due for refinancing or (ii) the loan is variable rate and the rental yield no longer covers debt service.

The current scenario is one where the primary sources of distress for CRE permanent lending are materializing in an unusually abrupt fashion:

       I.          The demand for office changed almost overnight during Covid, with vacancy rates increasing meaningfully and lease prices dropping. Office is a highly prevalent form of CRE (i.e. it is a large market), with some banks having significant exposures.

      II.          Interest rates increased a lot in a very short period, which has led to stress even in categories where rental yields have not suffered, such as multifamily.

Office is the primary case study for struggling commercial real estate. Average peak-to-current declines reach ~30% in some markets, and in some specific office buildings, much larger declines of up to ~70% of pre-covid value have been observed.

4.3 Construction Lending

Construction loans are usually perceived to be riskier than permanent loans, given the quality of the collateral is lower (an unfinished or pre-stabilization building is harder to monetize upon foreclosure than a finished, stabilized one). Therefore, most banks usually limit their exposure to this type of lending.

The following are the different ways a lender can lose money on a construction loan:

       I.          Once built, the building generates less rental income than expected. This risk can materialize in all kinds of assets, even apparently safe ones like Multifamily.

“In 2021, many developers decided to build massive apartment blocks in Texas and Florida, betting that lower taxes and the work-from-home trend would lead to large scale migrations away from cities like New York. All of a sudden, you would have three or four massive buildings open next to each other in cities like Houston, and the demand would simply not be there on time.” – Director of Multifamily at Greystone

      II.          An increase in interest rates increases the cap rate and, therefore, the value of the building once built. This risk, however, is generally lower in construction than in permanent loans, given lower LTVs for the former.

    III.          There are cost overruns during construction, which leads to the developer not being able to finish the building. Construction loans almost invariably include a completion guarantee from the developer (i.e. the loan is recourse until the building is finished). However, if the developer incurs higher costs than expected across several projects and defaults as an entity, projects might not be completed, and construction lenders could experience losses.

The third point is usually diversified away by having exposure to different developers and projects (i.e. it is unlikely that all the developers a bank lends to will experience cost overruns at the same time). However, in the current scenario, two elements are increasing the risk of widespread cost overruns:

       I.          Construction lending is almost always variable rate, and therefore developers are experiencing higher interest costs than they had budgeted for.

      II.          Construction costs are generally experiencing widespread increases due to inflation.

There is no question that we are leaving in a tough environment for CRE. In certain non-cyclical categories (e.g. Life Sciences, Seniors Housing), even more so than during GFC, given in ’08-‘09 interest rates (and therefore cap rates) declined rapidly and alleviated valuation and debt-service-coverage pressures.

5. Bank OZK’s Real Estate Solutions Group

5.1. Overview

The RESG is the crown-jewel of Bank OZK. It represents 65% of the total funded loan book, and it specializes in offering construction loans for very large, Tier-1,  “skyline defining” commercial real estate projects. Every loan that they underwrite abides to three core principles:

       I.          The developer must have a very successful track record with that specific type of project, as well as a strong balance sheet.

      II.          The project must have very attractive supply-demand characteristics, with positive market trends, similar properties in the same area observing low vacancy rates and limited new capacity coming to market in the foreseeable future.

    III.          The LTV must be very low, usually not larger than ~40% LTV (i.e. using the forecasted value of the property in the denominator) or ~50% LTC (using the projected cost of construction).

Note that since loans are progressively funded as expenses occur, at any given time only ~50% of the committed amount is outstanding. These leads to significant undrawn commitments that have the effect of requiring significant capital levels at the bank and reducing RoTE. See discussion in Section 2 for more information.

An example of a project Bank OZK lent to in 2022 is a 52-story luxury condo building in 50 West 66th Street, Manhattan. Located right in front of central park, the residential skyscraper was developed by Extell Development, which has successfully developed more than 10 skyscrapers in Manhattan over L15Y. Bank OZK signed a $620m loan for the project, which if fully drawn would be ~50% of the projected cost of the development or ~40% of the total value of the building. However, the actual drawn amount was far lower, given (i) the loan was only drawn progressively as the construction evolved and (ii) Bank OZK had to be repaid first as condos were being sold to final customers before the building was completed. In June 2024, when Bank OZK’s loan was repaid in full, the outstanding amount was only $340m, vs a building value of near $1.5Bn, providing a very large margin of safety for a highly demanded piece of real estate.

Other examples of projects Bank OZK has lent to are:

       I.          Waldorf Astoria Tower in Miami, which, once completed, will be the tallest building in Florida. It will have 100 stories of Condo and Hotel space. The building is projected to finish construction in 2028, but 90% of the condos are already pre-sold. The building is being developed by PMG, a developer with a portfolio of 150 luxury real estate assets mostly in New York and Miami.

      II.          Ritz-Carlton Hotel & Residences near Maddison Sq in Manhattan, which will include 250 hotel rooms and 16 residences. The project is being developed by Flag Luxury, a Real Estate investment group with $6Bn AUM and that has developed 6 other Ritz-Carlton-branded assets.

    III.          Brucker Boulevard Warehouse in the Bronx, a last-mile logistics facility with 106 loading docks and 1.1M sqft internal space. Amazon has already leased half of the available space.

As the reader can appreciate, the examples above are likely to become high quality real estate assets with high expected collateral value in case of default. Together with experienced developers and low LTVs, we believe this practice constitutes a sound form of lending.

“We have several loans with a committed balance above $500m. This is as much as 10-15% of our equity. I understand that other banks might not feel comfortable with this level of exposure, and I understand the reasoning behind diversification. But the reality is that the larger the project, the more experienced and better capitalized the sponsor is, and you can do a transaction like that at a much lower leverage than a $50m transaction, given lower competition. Also, the collateral is likely to be of much higher quality.” – George Gleason, Bank OZK CEO  

“Even with the best sponsors, you see them sometimes struggle and make mistakes. This is why it is paramount to have a great asset as ultimate collateral and a very low LTV.” – George Gleason, Bank OZK CEO 

“I would love to be able to participate in the permanent financing after construction is complete, but LTVs are too high. If you offer a 70-80% LTV and something goes wrong, you can get into a lot of trouble. Also, I would have to offer fixed rate loans with a tight pricing.” – George Gleason, Bank OZK CEO 

5.2. Terms of lending

Based on our research, Bank OZK is highly disciplined in their origination, and does not compromise on the following terms and conditions:

       I.          Bank OZK is always the most senior member of the capital structure.

      II.          Loans are short-term, and need to be repaid with a maximum of 3-5 years, in order to limit the impact of longer-term changes in supply/demand.

    III.          The loan must be variable rate, so that Bank OZK does not incur interest rate risk.

   IV.          The developer must purchase an interest rate hedge, so that he is protected from an increase in interest rates. This was already in place before rates rose.

     V.          The developer must provide a completion guarantee (i.e. the loan is full recourse to the developer until it is completed).

   VI.          The developer must abide by a carefully defined set of construction milestones. If a certain phase of the construction takes longer or is more expensive than expected, the developer is in breach of covenant and must negotiate a cure with Bank OZK, which entails introducing additional equity in the property.

  VII.          Bank OZK only starts funding the loan once the Developer’s equity is fully invested in the project, i.e. the first ~50% of construction expenses are borne by the developer, and only after that point does Bank OZK start lending to cover future costs. Exceptions to this rule occur exclusively in case a highly capitalized mezz lender (e.g. Blackstone, Brookfield) negotiates to fund its tranche pro-rata with Bank OZK, but even in that case the developer’s funds will have been deployed at inception.

VIII.          As discussed, LTVs are rarely above 40% and LTCs above 50%. More importantly, for non-residential projects that may be hard to lease upon completion (e.g. a large office, warehouse or life sciences project) a substantial portion of the funds is earmarked for “good news” funding, i.e. costs incurred once a lessee is found, such as lease commissions or building modifications requested by the incoming tenant. As a result, in such projects, LTC before lessees are found is rarely 50%, but more often in the 30-40% range.

Finally, Bank OZK screens as being on the more expensive end of the spectrum, ensuring that they are appropriately compensated for the risk taken:

       I.          On average, Bank OZK charges SOFR+350bps and a 1-2% origination fee, significantly more than the rates available on other types of CRE loans.

      II.          Loans have make-whole provisions that ensure a minimum total interest is paid throughout the life of the loan, protecting the bank from loss of revenue in case of early refinancing or low drawdowns.

    III.          Loans often have floor rates to protect the bank’s revenues in case interest rates decline. As of 1q24, 56% of the loans had a floor at SOFR=225bps or above (i.e. minimum interest of ~5.75% assuming 350 bps spread). The figure is currently trending upwards as old loans get repaid and new loans, originated in a higher-rates environment, have higher floors.

5.3. Key selling points

As described above, Bank OZK has far more restrictive loan terms than other banks. In order to close a deal, Bank OZK’s main selling points are:

       I.          Limited competition. For large construction loans ($300m+), Bank OZK comes in competition with very few other banks that are willing to that level of perceived risk. The only other bank known for commonly performing this kind of lending is Wells Fargo. Mezzanine funds are able to write similar checks, but at a much higher interest rate (SOFR + 600/700 bps), and therefore do not compete for senior financing with Bank OZK.

Note that this dynamic is more pronounced in times of market instability (e.g. Covid, today), while less so in periods of market exuberance (e.g. 2021).

      II.          Reliability of execution. Bank OZK screens as a professional organization that executes well.

“Bank OZK has a reputation for being quite burdensome with the information request, but extremely professional when it comes to timings and highly reliable. If you provide them with the right amount of access, and have a project that makes sense, they will get there on time and with no hiccups” – SVP of Capital Markets at Related

    III.          No loan syndication. Bank OZK does not syndicate the loans that it originates, which results in reduced diligence time, higher certainty of funds and lower fees.

   IV.          Accepting mezzanine financing junior to Bank OZK’s loan. Other banks refuse to let mezz lenders in the capital structure, with the primary argument that lower equity outlay reduces the alignment of interests between the developer and the lender. Bank OZK takes a different view, and lets Tier-1 mezz lenders (e.g. Blackstone, Brookfield) in the cap structure if the developer so desires.

The main argument for accepting mezz lending, beyond commercial considerations to win the deal, is that if the developer defaults on the equity, foreclosure of the mezz loan is a 45-90 days process, given the mezz lender does not technically foreclose on the real estate asset itself, but instead on the developers’ equity. Therefore mezz lenders can foreclose under the Uniform Commercial Code, as opposed to the judicial system. On the other hand, foreclosure by the senior lender is a foreclosure on the real estate asset, and requires a judicial process, which is a far more complex process that can take over a year.

Therefore, in a distressed scenario, a sophisticated mezz lender with strong Real Estate capabilities (e.g. Blackstone, Brookfield), will take control of the asset far sooner than Bank OZK would have been able to, and is likely to try to turn around the asset, potentially introducing additional capital and lowering the risk for Bank OZK. Also, upon foreclosure by the mezz lender, given change of control clauses, Bank OZK can further negotiate equity cures for their senior loan, which they often do:

“Upon foreclosure of a Dallas Office by the mezz lender, we negotiated a $16m loan curtailment from the mezz lender, without deteriorating our position in terms of collateral” – Q4 2020 earnings call

     V.          Offering non-recourse loans, that is, the developer can walk away if, upon completion, the building is worth less than the amount Bank OZK has lent. This is less of a differentiator than (IV), given other banks also offer non-recourse lending, albeit not all. The non-recourse provision always has 2 exceptions: (i) the completion guarantee, and (ii) carve-out guarantees or “bad-boy” guarantees, which basically limit the ability of the developer to diminish the value of the property in a near-bankruptcy scenario. For example, a bad-faith developer could use the long bankruptcy process to negotiate against Bank OZK and threaten to deteriorate the asset by allowing environmental contamination, not pay utilities or insurance, etc. Carve-out guarantees are designed to ensure developers act in good faith and do not have bargaining power during a long bankruptcy process.

5.4. Competitive moat and RoTE sustainability

In order to perform this type of lending, Bank OZK has built a unique organization that distinguishes it from other banks in two key areas:

       I.          Underwriting. Based on our conversations with industry participants, Bank OZK seems to perform a much deeper level DD than other banks, focusing on market supply/demand dynamics, potential construction pitfalls, etc. The larger average loan size allows Bank OZK to invest more diligence resources in every opportunity, but that would not be possible without the right in-house talent.

“I was very surprised when I first saw the underwriting memos from Bank OZK. They were 10-12 pages long and described the market dynamics for the asset in a lot of detail. In all other banks I have seen they are usually 3-5 pages long templates. Also, at Bank OZK the underwriting is centralized, with a single loan committee. Most other banks have decentralized underwriting for every region or business unit.” – Former VP of Commercial Lending, Bank OZK

      II.          Servicing & Asset Management. Bank OZK has a very proactive team following the development of each loan after origination. The bank constantly monitors whether construction is going according to budget and schedule, and only then do they approve further loan drawdowns. Also, they have the contractual right to require the developer to inject more equity, if a cost overrun or previously unidentified deficiency appears.

“Of course, large projects can go bad if the developer wastes money, plans poorly or tenant demand drops sharply. Bank OZK can mitigate poor execution by bullying the developer, which they are known for doing and currently may do quite actively, exchanging capital injections for time or concessions.” – Sam Haskell, former Research Analyst

Also, this type of lending activity requires Bank OZK to have the ability to take over foreclosed assets, which unlike in permanent lending, will be unfinished. Therefore, the bank might need to act as developer in a distressed scenario, and unlike other banks, Bank OZK does have that ability.

“Not every loan works, and you need to have the capability and the bandwidth to take an asset, finish it, manage it, and take it to market. If a sponsor is not performing, we have no reservations in operating the assets ourselves. In the past, we have operated hotels, fitness centres, built entire assets on land lots we owned, etc. It is not our preference, but if you are not willing to do that, then the borrowers can push you around if there is a problem, and we will not allow that.” – George Gleason, Bank OZK CEO

“Bank OZK has simply the best asset management team in country.” – Former SVP of Construction Lending, Bank OZK

In principle, nothing prevents other banks from replicating this model and building strong underwriting and servicing teams. However, we believe that it would not be a compelling decision for other banks because the large amount of undrawn commitments that this lending model generates requires the bank to be under-levered. As discussed in Section 2, Bank OZK, despite having almost 2x the Return on Assets as other banks, achieves a RoTE in line with industry average. The perceived level of risk from engaging in this activity, and the lack of remuneration for developing the required skillset, make an expansion towards this model unappealing for other banks.

“Almost no other banks want to do this because it is labour intensive and seems very risky. What lender making $225k, or regulator, wants to put their job on the line for a $60 million loan? Gleason takes advantage of this psychology.” – Sam Haskell, former Research Analyst

In summary, in our view:

       I.          Bank OZK has a strong moat and engages in a far more differentiated activity than most other banking business lines.

      II.          This does not lead to better RoTE than industry average due to regulatory capital constraints, but…

    III.          … it does provide us with a higher level of comfort that Bank OZK “deserves” to earn that RoTE and will be able to defend it in the future, which supports point (II) of our investment thesis.

Finally, note that we have not included mezzanine lenders as potential future competitors for Bank OZK. We believe that the fee structure and leverage capacity of Private Credit funds requires them to pursue higher risk opportunities at higher interest rates (above SOFR+500). Bank OZK will always be able to be more levered and have lower cost of funds than mezz lenders, and we believe the argument of overall low RoTE applies here as well. Based on our industry conversations, we have not perceived any ongoing trend around mezz lenders migrating towards the safer tranches of the capital structure.

5.5. Health of the current loan book

Bank OZK has one of the best disclosures we have seen for publicly-listed US banks, providing us with a higher-than-usual level of comfort on the health of the current loan book. Not only are Bank OZK’s lending practices sound in theory, but we also believe that, as of today, the existing loan book is in good standing.

Starting with non-performing loans, Bank OZK is not observing the upwards trajectory experienced across the industry, which has seen CRE non-performing loans increase by 74% in just 5 quarters.

Also, we draw comfort from the fact that the majority of the loan book has been originated recently, either during or after Covid. That reduces the issue of having pre-covid office exposure, and increases the likelihood that the market dynamics upon which the loans were underwritten remain relevant and have not changed.

Finally, regarding the LTV of all outstanding loans, Bank OZK publishes the chart above. Note that:

       I.          In conjunction with Figure 15, this graph provides a key insight: not only is the average LTV low, but also very few individual loans have high LTVs. Those that do are a consequence of a deterioration in a specific asset, they are properly provisioned for, and are addressed in detailed asset-by-asset explanations provided by management every quarter.

      II.          More importantly, while other banks often report LTV at origination, Bank OZK reports LTV using the latest appraised value. This is an important distinction, because an office loan originated in 2018, might have had an origination LTV of 50%, but if the value of the building has declined by 50%, it has now an LTV of 100%.

This bears the natural question around how often does Bank OZK re-appraise its loans (it is a time consuming and expensive process, and it would be unfeasible to re-apprise the 300+ outstanding RESG loans every quarter). According to our conversations with the industry, the company seems to re-appraise individual assets in case of loan extensions or modifications, if the credit quality is perceived to deteriorate and, also, as part of a periodic randomized sampling process.

As outsiders, we cannot be certain that all assets whose value has declined since origination have been recently re-appraised. However, the company discloses the following, which provides us with certain comfort:

“During the twelve months ended March 31, 2024, RESG obtained updated appraisals on 75 loans with a total commitment of $7.37 billion and appraisals on 67 newly originated loans with a total commitment of $6.99 billion, resulting in approximately 40% of RESG’s total commitments having appraisals within the last twelve months. Of the 75 loans reappraised in the last twelve months, LTVs were little changed (plus or minus 10%) for 59 loans, LTV decreased more than 10% for 2 loans and LTVs increased more than 10% for 14 loans.” – Bank OZK Mar-24 SEC Filing

6. Negative Report from Citigroup

On May 29^th 2024, Citigroup downgraded Bank OZK to “Sell” due to concerns on its largest outstanding loan. The loan is $915m in headline committed value, which represents about 21% of Bank OZK’s total tangible equity. Share price declined by ~15% upon publication.

The loan is for the construction of the “Research and Development District”, in Downtown San Diego. It is a 5-building project across 8 acres of land, with ~90% of the space designated for Life Sciences and the reminder for retail. Life Sciences facilities are designed to accommodate R&D activities for biotechnology and pharmaceutical companies, which require building structures that differ significantly from traditional office (due to need for lab space). Life Science real estate commands much higher rents than office (~70% more in San Diego).

The building is being developed by IQHQ, a group founded by Alan Gold in 2020 that has raised $2.5Bn in equity and is developing 3 major Life Sciences projects: the RaDD in San Diego and two more in Boston - Cambridge. Alan Gold has a long-standing track-record in the Life Sciences sector, having served as the CEO of BioMed Realty from inception in 2004 until its sale to Blackstone in 2016.

The loan was originated in August 2022, and in its report Citigroup highlights that the RaDD is likely a struggling project, with the main arguments being:

       I.          No leases have yet been signed, despite RaDD having been in development for more than 4 years and being projected to come to market over the next few months.

      II.          RaDD is based in Downtown San Diego, which is an unproven area for Life Sciences, given most pharma companies historically have chosen other areas in San Diego (Torrey Pines, Sorrento Mesa and Sorrento Valley) to stablish their research campuses.

    III.          Vacancy rate in Life Sciences in San Diego is increasing significantly, and additional new construction will result in an over-supply over coming years.

Note that Citigroup’s report focuses exclusively on the supply/demand dynamics of Life Sciences in San Diego, and does not evaluate the capital structure of the project (i.e. whether potential losses will affect only the equity portion, or if they may permeate towards the debt).

An analysis of available data for the Life Sciences Real Estate market yields the following conclusions:

       I.          San Diego is the 3^rd largest Life Sciences Market in the US, after Boston - Cambridge and San Francisco.

      II.          Vacancy rates have indeed increased sharply in San Diego during L18M, driven by lower VC funding for Biotech companies, as well as new construction coming to market.

    III.          However, this has been a widespread national occurrence, and San Diego does not screen as a city where either (i) vacancy rate is particularly high or (ii) an unusual amount of new capacity is under construction.

   IV.          Asking rents in San Diego have not yet declined, with 0% YoY change in 1q24.

     V.          The employed life sciences population in San Diego has grown at 3.5% CAGR over L11Y, with a departure from long-term trend in ’21-’22 due to unusually high Biotech VC funding, followed by a recent reversal to trend. Note that if this growth continues, a potential Real Estate oversupply of ~15% would be corrected in ~4 years.

Given the above, we agree with Citigroup’s believe that the RaDD might struggle to find tenants in the short term. However, based on Bank OZK’s additional disclosure on this loan, we believe that the risk to the debt is minimal due to the project funding structure:

       I.          Even if the headline commitment for the loan is $915m, only $555m have been fully funded, with the reminder earmarked for “good news funding” (i.e. tenant improvements and leasing commissions). See our discussion on Bank OZK’s lending terms in Section 5.

      II.          The developer has already invested ~$950m of their own equity in the project.

    III.          In order for Bank OZK to impair a portion of its loan, the value of the building should fall by ~75% vs the origination appraisal performed in July 2022.

Value declines of ~75% in prime real estate assets are an extremely rare occurrence and can be found almost exclusively on some very specific office buildings (see Section 4). We remind the reader that those value declines are driven by two highly unusual events: (i) a sharp decline in office demand given Covid and work-from-home trends and (ii) a rapid increase in interest rates that translated into higher capitalization rates.

The appraisal of the RaDD loan was made in July 2022, a time when WFH trends had already played out, US CPI was peaking at 9% p.a., and the Fed had already implemented three 50bps rate hikes in the preceding 3 months. We can imagine a world where the equity in this deal suffers some losses, but we believe that a decline of ~75% in the value of the asset is highly unlikely. After all, this is a high-quality asset, built by an experienced Life Sciences developer, in a Top-3 city in the country, with underlying Life Sciences employment growth, and at an enjoyable location in Downtown’s waterfront.

We therefore disagree with Citigroup’s views on the safety of this loan, and point out that, even if the funded portion of the loan was fully impaired (i.e. the recovery value of the asset was zero), that would only consume ~2 quarters of Bank OZK’s pre-tax pre-provision earnings (Figure 23). Certainly not an outcome we would cheer, but definitely not an existential risk for the bank.

7. Diversification efforts

Bank OZK ranks #1 in terms of exposure to Commercial Real Estate among all large, listed banks in the US. For the reasons covered above, we believe that in the specific case of Bank OZK this does not represent an undue risk. However, the regulator does not share our view: in the L10Y, Bank OZK has come under significant regulatory pressure to diversify away from Commercial Real Estate. George Gleason disagrees with this principle, but understands that an unfavourable relationship with the regulator is unadvisable.

Let's say, I'm in Arkansas and one of my commercial bankers from Dallas calls and says, "Hey, we've got a $10 million loan opportunity for a church in Dallas. And in order to close it, I'm going to have to do an 80% loan-to-value loan, no guarantors, a 10-year fixed rate and I can't put any covenants on, because everybody in town wants this business”

George would look at me and say, "Hey, let me ask you a question, why in the world, would I set aside capital to go put that loan on the books when I can do another $10 million loan in RESG at 55% of cost, get a 1% origination fee, at SOFR+350, with a make-whole provision for interest, and then get an exit fee" – Former Chief Credit Officer, Bank OZK

As opposed to rolling out commoditized community banking services, Bank OZK’s reaction has been to try and develop certain specific niches where they could obtain better risk-adjusted returns. To do it, they focus on acquiring highly-specialized talent within a specific vertical, and giving them freedom to grow their business inside the bank.

While in other regional banks compensation is often defined by seniority bands and other HR rubrics, Bank OZK looks exclusively at the cost-to-income ratio for the loan producers they hire, aiming to stay at ~20%. For example, a banker able to develop a $50m loan book with a 4% net interest margin would be generating $2m of Net Interest income for the bank and could be compensated with up to $400k in gross compensation.

The largest niche Bank OZK has created is the RV and Marine lending business, which currently accounts for 11% of all loans, and which they have grown out of a small M&A target in 2016. The key principles of operation of this division are:

       I.          Loans are originated indirectly through a network of ~1,400 dealers nationwide, which offer their customers a Bank OZK loan to finance the purchase of a boat, campervan or motorhome.

      II.          Customers must have a very high credit score (average of ~790 FICO).

    III.          Customers must have experience in having owned at least two similar vehicles in the past, so as to ensure that they understand and are able to withstand the financial commitment it represents.

   IV.          Loans are full recourse to the owners. The purchased vehicle acts as collateral, but given the depreciating nature of these assets and high LTVs, losses are likely to occur in case of default.

Credit Quality in the RV and Marine division has historically been outstanding, and no deterioration is being experienced in the current cycle. On the other hand, given the fixed-rate nature of this product, yields are increasing slower than SOFR. The portfolio reprices as old loans mature and are replaced by new loans originated in a higher interest rate environment.

Bank OZK is currently also ramping up efforts in its Corporate division. This division currently accounts for ~6% of all assets, with the majority allocated to Asset Backed Loans. The key principles governing Bank OZK’s ABLs are:

       I.          They are credit lines backed by existing account receivables and inventory.

      II.          Loan commitments average $30-300m, and around ~40-50% are usually drawn at any given time.

    III.          Loans are variable rate, with pricing 200-300bps above SOFR.

   IV.          The advance rate is 85-90% of eligible receivables and 85% of the net liquidation value of eligible inventory. Only current receivables (i.e. not aging) and easily disposable inventory are accepted as collateral.

ABLs are considered a safe form of lending, given losses occur only in the highly unlikely scenario where a large corporate enters bankruptcy AND the receivables or inventories that have been lent against are insufficient collateral. As a result, there is significant competition within this segment and differentiated RoTEs are unlikely. We remind the reader, however, that our investment thesis revolves around Bank OZK achieving RoTEs in line with the banking industry, and there is no apparent reason why growth in this segment would invalidate our assumption.

Finally, the Community Banking division, which accounts for 18% of all loans, consists mostly of generalist small and middle-market business loans originated in the states where Bank OZK has physical branches, and as part of their comprehensive deposit strategy. Included in this division are also smaller niche lines of national reach, such as poultry financing, homebuilder loans, business aviation or affordable housing.

Going forward, Bank OZK believes that growth in the business lines above will result in RESG becoming ~50% of total business, from 65% today. Should it occur, RoTE is unlikely to fundamentally change, but we believe that a valuation re-rating would become a more likely source of upside, given widespread market concerns around Bank OZK’s current concentration in Commercial Real Estate. Concerns that, as discussed, we do not share and that we believe provide an attractive entry opportunity.

8. Risks

The main risks that we believe an investment in Bank OZK entails are:

       I.         CEO retirement

      II.        Credit risk

    III.        Sensitivity to interest rates

    IV.          Run-on-the-bank risk

   V.          Regulatory Risk

As for CEO Retirement, we refer the reader to Section 3.

Regarding Credit Risk, a lot has been written in this report about the soundness of the lending practices and the health of the current loan book (Sections 5, 6 and 7). As a final remark, the analysis below on the loss absorption capacity shows that Bank OZK would be able to absorb losses accounting for ~10% of all loans outstanding, which equates to 22x their current level of non-performing-assets, or 6x the peak level of charge-offs experienced during the GFC.

Regarding Interest Rate Sensitivity, Bank OZK does not have a held-to-maturity security portfolio (i.e. carried at cost), accounting for all their securities as available-for-sale (i.e. carried at current fair value). Also, 75% of their earning assets are variable-rate, which paired with a sticky retail deposit base, has led to a faster repricing of the assets than the liabilities over the last 2 years, and hence expanding Net Interest Margins. To avoid the opposite effect in case of a decline in interest rates, interest rate floors are being introduced on all newly originated loans. As of 1q24, 56% of the loans had a floor at SOFR=225bps or above (i.e. minimum interest of ~5.75% assuming 350 bps spread).

The main risk regarding interest rates, in our view, is that a decline in rates would lead to a higher level of repayments in outstanding construction loans, given developers would be more likely to find attractive permanent loan offers with which to refinance Bank OZK’s financings. That would be a short-term stream of exit and make-whole fees for Bank OZK, but could put pressure on the RESG outstanding balance and result in a temporary deleveraging of the bank until new originations could replenish the repaid loans, as happened during 2021 (Figure 8).

We believe this tension between origination and repayment rates is inherent to the Bank OZK’s business model, given the short-term nature of the loans and the developers’ natural incentive to refinance in good times. As such, Bank OZK is likely to periodically experience periods of deleveraging, and this is the main reason why we forecast forward through-the-cycle RoTE in the range 13-15% as opposed to the current ~17%.

As for risk of a bank run, we note that sound financial institutions rarely experience them, although exceptions may occur (e.g. small bank run experienced by Western Alliance in 1q23).

We note, too, that bank runs are not driven by analysis but by collective perception, which would put Bank OZK at an increased risk: after all, the description “regional bank with an outsized exposure to commercial real estate” would be unlikely to calm wary depositors if a bank run started to unfold.

Conscious of this risk, the bank has pursued a deposit strategy focused on maximizing the percentage of deposits that are insured (i.e. those deposits that are under $250k and therefore insured by the FDIC). In the 2023 Regional Banking crisis, insured deposits across the industry proved highly sticky, given the American layman seems to understand that the FDIC (i.e. “the government”) will protect his funds if his bank fails.

Out of the Top 50 banks in the US, Bank OZK is #3 in share of insured deposits, having 81% of their deposits either insured or collateralized, and featuring a very low $44k average customer deposit, pointing to the truly retail nature of their deposit base. The proof of the quality of the deposit base can be found on the behaviour following SVB collapse, a time when Bank OZK managed to substantially increase core deposits while most Regional Banks did not.

Finally, Regulatory Risk is a concern, and can arise from two main sources:

       I.          As discussed in Section 7, the regulator has historically pressured Bank OZK into diversifying away from RESG. Until now, Bank OZK has been able to continue to hold ~65% of their loan book in CRE, but, should regulatory pressure increase, the bank would likely need to diversify faster into other business lines.

      II.          New regulatory frameworks might increase the capital requirements for all banks, as has happened with Basel II and Basel III. If that were to occur, it is unclear if Bank OZK would be able to translate the additional capital requirements into higher pricing, and therefore, RoTE might decline.

This is a risk that, although not existential, might reduce the long-term returns from our investment by a few percentual points. At 6.8x 2024 Earnings, however, we feel adequately compensated for this risk.

DISCLAIMER

THE INFORMATION CONTAINED ON THIS REPORT IS NOT AND SHOULD NOT BE CONSTRUED AS INVESTMENT ADVICE, AND DOES NOT PURPORT TO BE AND DOES NOT EXPRESS ANY OPINION AS TO THE PRICE AT WHICH THE SECURITIES OF ANY COMPANY MAY TRADE AT ANY TIME.  THE INFORMATION AND OPINIONS PROVIDED HEREIN SHOULD NOT BE TAKEN AS SPECIFIC ADVICE ON THE MERITS OF ANY INVESTMENT DECISION.  INVESTORS SHOULD MAKE THEIR OWN DECISIONS REGARDING THE PROSPECTS OF ANY COMPANY DISCUSSED HEREIN BASED ON SUCH INVESTORS’ OWN REVIEW OF PUBLICLY AVAILABLE INFORMATION AND SHOULD NOT RELY ON THE INFORMATION CONTAINED HEREIN.

The information contained on this website has been prepared based on publicly available information and proprietary research.  The author does not guarantee the accuracy or completeness of the information provided in this document.  All statements and expressions herein are the sole opinion of the author and are subject to change without notice.

Any projections, market outlooks or estimates herein are forward looking statements and are based upon certain assumptions and should not be construed to be indicative of the actual events that will occur.  Other events that were not taken into account may occur and may significantly affect the returns or performance of the securities discussed herein.  Except where otherwise indicated, the information provided herein is based on matters as they exist as of the date of preparation and not as of any future date, and the author undertakes no obligation to correct, update or revise the information in this document or to otherwise provide any additional materials.

The author, the author’s affiliates, and clients of the author’s affiliates may currently have long or short positions in the securities of certain of the companies mentioned herein, or may have such a position in the future (and therefore may profit from fluctuations in the trading price of the securities).  to the extent such persons do have such positions, there is no guarantee that such persons will maintain such positions.

Neither the author nor any of its affiliates accepts any liability whatsoever for any direct or consequential loss howsoever arising, directly or indirectly, from any use of the information contained herein. In addition, nothing presented herein shall constitute an offer to sell or the solicitation of any offer to buy any security.

Red Beryl is not a regulated institution.

Nu Holdings is the bank-holding company for Nubank, the largest neo-bank outside Asia with 100m customers. Nubank is based in Brazil, where it continues to grow at ~60% p.a., and has recently entered Mexico and Colombia, where it is accelerating from a low base.

27% of the adult population in Brazil currently use Nubank as their primary bank account. In 2023 it turned profitable, currently achieving 60% RoE in Brazil. With a $57Bn market cap, the company is trading at 24x Run Rate Earnings for Q4-2023 (after share-based comp and exceptionals).

We believe Nubank is no longer a Fintech, but has become a full-service digital bank. It has a strong moat in the form of low cost to serve and proficient credit underwriting, and, in our view, a long runway for growth. Also, we believe that risks are lower than the term “neo-bank in Brazil” evokes, and we therefore recommend the investment.

Contents

1. Nubank’s fit with Red Beryl’s Investment Strategy

2. Opportunity Overview

3. Banking Industry in Brazil

4. Nubank’s Runway for Growth

5. Deep-Dive into the Mexican Opporunity

6. RoE, Profitability, Valuation and Long-Term Potential

7. Compeitive Position and Moat

8. Risks

9. Conclusion

1. Nubank’s fit with Red Beryl’s Investment Strategy

Nubank is an important company for Red Beryl. It is one of the largest holdings in our portfolio and our first write-up ever. It configures, nonetheless, a severe deviation from our intended investment strategy.

Red Beryl is a small investment partnership and, as such, we have a key advantage over other money managers in that we can invest in illiquid stocks. We spend most of our energy evaluating companies with little float, slim trading volumes and market caps in the hundreds of millions.

Also, we have no particular interest in Emerging Markets. We understand that equities are often optically cheaper in these realms, but we feel incapable of adequately determining which is the appropriate discount we should require for this type of risk, and we often find that the discount offered by the market is not necessarily far from our initial intuition.

For all these reasons, Nubank is exactly the kind of company we would usually avoid spending time on. It has a market capitalization of $57Bn, an average daily trading volume of more than $100m, it is based in Brazil, it is by no means hidden (even Berkshire Hathaway invested in its IPO, with Bill Ackman later publicly quizzing Berkshire’s Todd Combs for his views on the stock), and trades at a hefty 24x Q4-2023 Run Rate Earnings.

However, we believe that it is a company operating within an incredibly unique set of circumstances that make it a true long-term compounder with fantastic moats and significantly less risk than the term “neo-bank in Brazil” evokes. We do not have a good answer as to why “alpha” might exist in such a well-researched opportunity, and we recognize that in a negative macro environment the stock is likely to re-rate significantly more than others. However, we hold a strong belief that in 10 years Nubank will be worth a lot more than it is now and, as such, we have gladly made an investment, even if it fits none of the traditional attributes we are looking for.

2. Opportunity Overview

Having recently crossed the mark of 100 million customers, Nubank is the largest neo-bank outside Asia. It operates primarily in Brazil but has recently expanded into Mexico and Colombia. It was founded in 2013 by David Velez, a Colombian entrepreneur who, after receiving an MBA from Stanford, joined the Venture Capital firm Sequoia to cover Latin America. Two years later, he recognized the opportunity to establish a digital bank in Brazil, received seed funding from Sequoia and set up Nubank. The performance of the company has been exceptional since inception, and the company turned sharply profitable in 2023.

Nubank started by offering a fee-free credit card. In 2017, it obtained a banking license, and soon after launched a checking account. Together with personal loans, these 3 products are the source of more than 90% of the Revenues today. Nubank does not aim to be a point-solution like most Fintechs in the US and Europe, but instead is on a journey to build a full-service bank offering able to compete head on with incumbent banks.

To this date, Nubank has achieved widespread adoption in Brazil, with 53% of the adult population currently signed up as a customer, 44% an Active customer and 27% using Nubank as their primary bank account.

In our view, there are 4 key elements that have driven Nubank’s success:

       I.          Structure of the Banking industry in Brazil: with the top 5 banks holding over 80% market share, consumers in Brazil used to be poorly served. Abusive fees were highly prevalent and interest rates on debt were sky high (400%+ p.a. on revolving credit card and 150%+ p.a. on personal loans). As a result, Brazilian banks achieved very high RoEs (over 20%), even if doing little to increase operational efficiency.

Nubank’s initial product was simply a fee-free credit card with an easy-to-use app and great customer support. It was met with exceptional enthusiasm by customers, who were used to queuing for hours in heavily armoured bank branches to obtain a credit card, which would have at least ~$50-100 in annual fees.

      II.          Lower Cost to Serve: traditional banks in Brazil operate with large amounts of overhead, managing thousands of branches and workforces in the region of 80-100k employees each. Nubank, with roughly the same number of customers today as the largest traditional banks, operates with only 8k employees. Nubank passes those lower costs to its customers, resulting in lower fees and interest rates, and hence attracting more customers and higher usage.

    III.          Ability to Underwrite Credit: Nubank is a fully licensed bank, with the ability to accept deposits and use them to extend loans. Originally, the main source of Revenue was interchange fees (~1.2% of the Purchase Volume) but since inception Nubank has understood that banking in Latin America revolves around extending credit. Interest on Loans represents a much larger profit pool than fees (see Figure 2), and offering credit products is the only way to become the primary bank account for a majority of the population.

Underwriting credit is, however, notoriously difficult, with several start-ups having recently tried and failed (a recent example in Brazil is Stone Co). Over the last 10 years, Nubank has built a highly sophisticated credit operation inspired by Capital One, which in the 1990s developed a new way of underwriting consumer credit in the US that consisted in performing complex data analyses to determine the repayment likelihood for each customer. Interestingly, Nubank’s current  COO is the former Head of Card Analytics and Infrastructure at Capital One.

As a result, in the last decade Nubank has amassed a ~$18Bn Consumer Credit Portfolio spread across ~40m customers, providing an invaluable amount of proprietary data on the performance of different credit cohorts across the cycle. That serves as a significant moat against new start-ups, and reduces the once meaningful gap vs traditional banks.

   IV.          Overall great execution: across this report we will present several examples of Nubank’s ability to ramp-up new products and geographies, recruit a top tier US-centric management team and achieve high NPS across its products. Overall, Nubank’s execution to date is impeccable, with no major missteps, and far outcompeting similar start-ups.

The reader might be surprised that “great Customer Experience” does not feature more prominently as one of our key reasons for explaining Nubank’s success. When looking at neo-banks in the US and Europe, an intuitive and easy to use app is often one of the main selling points. In our view, that explains the lack of traction of neo-banks in these geographies (we urge the reader to try to find a digital bank in the US or Europe meaningfully taking share in core banking products, such as deposits, loans or credit cards). Start-ups like Chime or Revolut have great mobile apps, but they are not comparable to Nubank, given their very shallow offerings and inability to displace traditional banks as primary banking providers.

In contrast, in Brazil 27% of the adult population uses Nubank as their primary bank. We believe that this success is not primarily driven by having a great mobile app, which they have. Instead, what Nubank has achieved in Brazil is driven by charging lower fees, lower interest rates and sustainably providing credit to customers that other banks would not underwrite.

“In Brazil, when you need a credit card, you just go to all the main banks and ask for one. Usually only one or two approve you, so those are the ones that you use, ideally the one with the lower fees. It has little to do with anything else” – Former Nubank Director of Corp. Dev.

To this end, Nubank has mastered the low-and-grow approach to credit limits pioneered by Capital One, which revolves around on-boarding unknown customers by offering a very low credit limit to start with (often as low as $10). Progressively over time, as customers show their willingness and ability to repay debt, credit limits are expanded. The result is the cohort behaviour shown on RHS of Figure 3, where customers’ usage, and thus ARPAC, increases systematically over time.

Among the big banks in Brazil, Nubank is the best positioned to implement a Low-and-Grow approach, given the lower cost to serve. As the reader can imagine, serving a customer with a $10 credit limit is hard to do profitably from a branch. This has historically resulted with a moderate skew for Nubank towards lower income customers than traditional banks. However, with 53% of the current adult population being a customer of Nubank, its customer distribution is slowly converging to other banks’, even if Nubank’s Share of Wallet is still lower for high-income individuals (more on this later).

Nubank’s success has left its founder, David Velez, with a fortune worth more than $12Bn. Alignment of interests is strong, with the majority of his wealth fully invested in Nubank (he owns ~20% of the company). Share Based Compensation is minor and fully expensed into the presented Net Income figures and, as a further proof of good corporate behaviour, in Q4-2022 David Velez voluntarily surrendered (with nothing in exchange) a Long-Term Incentive Package that was awarded to him at IPO.

Throughout this write-up, we will expand on each of the five pillars of our investment thesis:

       I.          Nubank has a very strong moat and competitive positioning.

      II.          The runway for growth is long, with end-state potential earnings of $4/share (vs spot price at $11.76/share).

    III.          RoE is already as high as 60% in the Brazilian subsidiary (other Geos still loss making).

   IV.          Valuation is more attractive than headline figure suggests, trading at 24x Q4 2023 Run Rate Earnings, when adjusting for the accounting impact of IFRS 9.

     V.          Risks are much lower than initially apparent, particularly macroeconomic and credit risk.

Note that we have chosen to present valuation on an Earnings Multiple, not on a Book Value Multiple. At $11.76/share Nubank trades at a very high BV multiple of ~9x. However, with a RoE of 60% in Brazil, we believe that Book Value becomes meaningless. For context, below we compare the RoE of Nubank vs that of well-known US tech companies. We believe the reader is likely to agree that valuing Amazon on a Book Value basis is inadequate, and we argue that the same is true also for Nubank.

3. Banking Industry in Brazil

Brazil has well-developed financial markets, with a functioning regulatory framework and an increasing adoption of credit. Over the last 15 years, household debt to GDP has more than doubled, leading to significant market growth, with a long runway ahead if the gap vs developed markets is to be progressively reduced.

The market has historically been very concentrated on the Top 5 banks (Itau, Banco Bradesco, Banco do Brasil, Banco Santander and Caixa Federal). They used to hold 80%+ of the total market, both in Consumer and Commercial. As a result of the lack of competition and limited capital availability, the Brazilian banking industry is among the most profitable ones in the world.

However, powered by the advance of smartphones, a number of digital disruptors led by Nubank have emerged and started to capture market share on the consumer side. The commercial side remains undisrupted.

As seen on Figure 8, on the consumer side, the main types of loans are credit card, personal loans and payroll loans. The reader is likely to be surprised by two factors. First, that interest rates are obnoxious by US or European standards, which is the result of a stark lack of competition combined with a high level of delinquencies (more on this later). Second, that Mortgage is a much smaller asset class, which is due to the fact that banks have historically struggled in court to enforce foreclosures. Also, given the very high interest rates on other loans, Mortgage holds an even smaller weight in the total revenue pools.

To set the scene, and given significant differences vs the US and Europe, a detailed description of the structure of the three main types of consumer loans feels appropriate.

Credit card is the main form of financing for the average Brazilian consumer. When a consumer purchases an item, it has three ways of paying with the credit card:

       I.          Without instalments: if the consumer selects this option, the purchase will be completed instantly and debited from the consumer’s account (on average) 28 days later without any cost to the consumer. The bank receives an average interchange fee accounting for ~1.2% of the purchase value, which is discounted from the amount paid to the retailer (same as in the US and Europe).

Unlike in the US and Europe, however, the retailer does not get paid instantly for the transaction. Instead, the funds are transferred by the bank to the retailer 30 days later. Therefore, given the bank is paid 28 days after the transaction but only pays the retailer after 30 days, a basic credit card transaction in Brazil generates 2 days’ worth of funds for the bank, as opposed to consuming funds like it does in the US and Europe (and in Mexico, more on that later).

If when the bill is due after 28 days the consumer does not satisfy it, the balance becomes “revolving”. Revolving balances can last, by law, only for 1 month, after which they need to be automatically converted into instalments. The revolving balance is notorious for having high NPLs and a very high average interest rate of 476%, as per Dec-23 Brazilian Central Bank data. The interest rate after the revolving balance is converted to instalments is lower, with an average of 192% p.a.

      II.          With interest-free instalments: in Brazil it is very common that retailers offer interest-free instalments. In this case, at check-out, the customer is presented with the option and, if selected, the purchase value is spread across 2 to 12 payments and debited periodically from the customer account over the following months.

In this case, the bank is remunerated with the same ~1.2% interchange fee as in the no-instalments option, and receives no interest, given the retailer is paid in instalments following the same pattern that the funds are debited from the consumer’s account (e.g. if the consumer pays 1/12^th of the purchase value each month over 12 months, the retailer receives the funds following the same timeline).

However, the risk of default stays with the bank. Following the example above, if the customer defaults after 6 months, the bank is still liable for paying the retailer on time for the remaining instalments. This is a negative dynamic for banks, given they take higher credit risk than in the no-instalments option (the customer needs to stay current for longer) without being appropriately compensated for it (the interchange fee is the same and there is no interest). The high prevalence of the interest-free instalments option in the Brazilian credit card industry is often used as an argument by banks for the relatively higher interest rates in the interest-paying portion of the credit card portfolios.

    III.          With interest-paying instalments: if the retailer does not provide an interest-free instalment option, the consumer can choose to divide the purchase across several instalments with their bank. In that case, the bank pays the retailer 30 days after the purchase and debits the customer accounts over the coming months, receiving the interchange fee and interest for financing the purchase.

The chart above deserves three main comments. First, note that the credit card in Brazil is primarily a tool for financing as opposed to transacting, unlike in the US and Europe, where the majority of the balance is current (vs just 35% for Nubank). Second, note that Nubank presents a much lower share of revolving balance than the market (6% vs 12%). This is by design, since revolving introduces very high interest rates, low NPS and high NPLs. As such, Nubank actively tries to quickly structure its customers’ revolving balances into instalment products. Finally, note that as the credit cohorts mature, Nubank is expanding the interest-earning portion of their balance (29% in 2023 vs 9% in 2020).

Switching to the second asset class, personal loans have the same structure as in the US or Europe, where cash is advanced in exchange for interest payments. Interest rates are substantially higher though, with the market averaging 155% p.a. and Nubank averaging 114% p.a, as of Dec-23 Central Bank data. As an anecdote, figures are stated to the Brazilian consumer as interest per month (e.g. 6.5% p.m.), rendering the figure apparently more palatable.

Payroll loans, on the other hand, are a unique feature of certain Latin American markets, such as Brazil. In this case, banks extend credit to individuals and, through IT integration with government systems, the principal and interest payments are diverted automatically from the borrower’s government-sponsored monthly income. Naturally, this form of financing is available only to those perceiving recurring payments form social security (retirees and disabled individuals) or employed at a public entity (federal or state governments, municipalities, etc.).

These loans have a much lower interest rate (capped at 1.7-2% p.m. by law). Default rates are much lower and, importantly, stable and uncorrelated from the broader macroeconomic environment. Default on payroll loans occurs mostly under three scenarios:

       I.          A public sector entity (e.g. a small municipality) is no longer able to pay salaries. Note that banks usually limit lending to employees of public entities with doubtful creditworthiness.

      II.          Another financial obligation (e.g. taxes, child support or a lawsuit) gains seniority over the credit payment.

    III.          The borrower deceases and there are not enough assets left to settle the loan.

Given the lower interest rates and longer maturity of these loans (2-3 years vs ~6 months for personal loans), the only way to fund them profitably is by having access to a strong deposit franchise that is both low-cost and sticky. Since building a deposit franchise is difficult and time-consuming, traditional banks have seen little disruption in this market and retain a very high market share. Note that Nubank has, over the last decade, built a meaningful deposit franchise, having recently changed its remuneration policy to narrow the gap vs incumbents (with no impact to churn metrics). As such, Nubank recently launched a payroll lending product that is rapidly accelerating (more on this later).

A Banking overview would not be complete without a brief description of the dynamics around Deposits. Deposits in Brazil are similar to the US or Europe, with checking accounts that are (often) unremunerated, complemented with remunerated savings accounts and term deposits. The main singularity for Brazil is the fact that the Central Bank interest rate is much more volatile than in the US or Europe and often much higher. In result, the industry has evolved to pricing deposits not at a fixed rate but as a stated percentage of CDI (the Brazilian equivalent to SOFR). A typical deposit in Brazil will not pay “11% p.a.” but instead “100% of CDI”. This results in Deposits repricing instantly with changes in the interest rate environment, as opposed to the lag experienced in other markets. However, unlike the US, the very high Net Interest margins in Brazil and the relatively short asset duration have historically prevented major funding crises.

Finally, to conclude this section, we would like to include a few remarks on the Brazilian Banking Regulator, the BCB. Conscious of the need to reform the banking sector, given abusive fees and very high interest rates, the Brazilian Central Bank has been very open to fintech as a way to promote competition. From a regulatory perspective, though receiving no direct support, Nubank has found it easier to obtain the necessary licenses and permits in Brazil than other startups in similar countries, such as Mexico (more on that later).

An example of this benign regulatory environment occurred in 2017, when the Brazilian Central Bank blocked a proposal from the Federal Government to change the deadline for credit card issuers to settle transactions with retailers from 28 days to 2 days (see discussion on credit card above). That proposal would have aligned Brazil with international standards and protected retailers. However, the existing lag allowed Nubank and other start-ups, which started as credit card issuers without a banking license (and hence without the ability to accept deposits), to scale without a massive need for equity funding. Changing the settlement time would have benefited incumbent banks, who had access to strong deposit franchises, and hindered competition, which was clearly against the BCB’s stated goals.

More recently, the BCB has launched two highly innovative and disruptive frameworks focused on increasing competition, both of which Nubank is navigating effectively: Pix and Open Finance.

Pix is an instant payment system that launched in 2020, is centrally managed by the BCB, and allows businesses and individuals to execute account-to-account payments in a free, fast, and easy manner. It has experienced explosive growth, cannibalizing primarily cash payments (52% of POS Payments in 2017 vs 22% in 2023, as per The Global Payments Report from Worldpay/FIS).

Initially Pix was mostly used for peer-to-peer transfers, but over time, businesses are allowing payment using Pix (payment can easily be executed by scanning a QR code). That bears the natural question on whether Pix will eventually disrupt debit and credit card payments. We admit that this is tough question to answer, but we offer a few thoughts:

       I.          Businesses prefer to be paid using Pix, given no interchange fee is involved.

      II.          However, we have argued before that credit cards in Brazil are a financing tool, not a transacting one, so Pix itself should not be able to disrupt credit cards, but…

    III.          ...financing for Pix is a reality and Nubank currently allows customers to use their credit card limit to finance their Pix transactions in exchange for interest payments. Nubank has been the first major bank to launch this product, allowing it to increase the interest-earning instalments portion of their credit card receivables (23% for Nubank vs 11% for the Market, as shown in Figure 9).

   IV.          The above argument does not apply to debit cards, which are clearly at risk of disruption by Pix.

We do not know if Pix adoption will reduce debit and credit card adoption. If it reduces debit card adoption it will impact Nubank’s debit-card-related interchange fees, which account for ~3.5% of their total Revenues. If it reduces credit card adoption, we believe there should be no impact on Nubank’s profitability, given (i) much clearer leadership on Pix financing than generally on credit card and (ii) there is no reason why Pix financing should overall be a less profitable product than credit card. Sure, it does not include interchange fees, but we remind the reader that 71% of Nubank’s credit card balance is interest-free (Figure 9), vs none for Pix financing.

Overall we believe Pix is a great case study of the financial innovation happening in Brazil and how Nubank is navigating it. First, it reduced cash usage, which is positive for a player with no branches or native ATMs like Nubank. Second, it allowed Nubank to differentiate by launching Pix financing faster than incumbents. And finally, it showcases Nubank’s great execution by observing Nubank’s market share of Pix keys, which is significantly ahead that of all other incumbent banks.

In order to execute a Pix transaction, an identifier of the destination account is needed. That identifier is called a Pix key. The user can choose what that key looks like. It can be an email address, a phone number or an alias.

When Pix was released in 2020, there was a fierce competition among banks to capture Pix keys. Banks correctly anticipated that if customers associated their preferred Pix key (e.g. their phone number) with their bank account, that would drive increased usage of that bank account.

In 2020 Nubank counted only ~15% of the adult population in Brazil among their customers vs ~40-60% for each of the large banks (individuals in Brazil often have multiple bank accounts). Hence, gaining market share of Pix keys was, in principle, much harder for Nubank. However, after designing the best sign-up experience in the industry, Nubank managed to capture more keys than any of the incumbents, using the Pix opportunity to accelerate growth.

“We bet big on Pix, and we sweat every pixel of the sign-up flow, because customers were going to decide who would get their Pix keys. We drove our team crazy to make it as simple as we possibly could. That has been a key driver of our recent growth.” - Jag Duggal, Current CPO at Nubank.

Finally, the dynamics with Open Finance are similar to those around Pix. The Brazilian Central Bank launched in 2021 a regulatory framework by which customers could transfer their data from one financial institution to another. The goal was to increase competition for the incumbent banks, who held the very valuable historical transaction data that is used for underwriting credit. Having access to that data allows newcomers, like Nubank, to offer credit with better terms to those customers.

Getting customers to approve the transfer of data and incorporate it into the underwriting flow is, however, not easy. As with Pix, Nubank has executed better than the incumbent banks and, as can be seen below, leads by a wide margin the number of successful incoming data transfers.

Note, however, that the set of data available through Open Finance is limited and that we believe that underwriting credit is an incredibly complex endeavour. As such, based on conversations with industry participants, we believe Open Finance will not meaningfully erode Nubank’s moat around Credit underwriting vs newer Fintechs (e.g. Mercado Pago), even if it is definitely a negative element on that respect. More on that later.

4. Nubank’s Runway for Growth

A key element of our investment thesis is that Nubank has a very long runway for growth, very high visibility on that growth, and that little capital will be consumed due to high RoEs. In this section, we address the 7 growth drivers that we believe will drive Nubank’s top line over the next decade:

       I.          Increase in number of customers.

      II.          Increase in per-customer usage of credit card and personal loans.

    III.          Rollout of payroll loans.

   IV.          General market growth.

     V.          Increase in penetration within the high-income segment.

   VI.          Rollout of adjacent products.

  VII.          Geographical expansion.

Starting with the increase in the number of customers, Nubank has consistently been adding 4-5 million new customers in Brazil every quarter for the last 3 years. This growth avenue will eventually tap out, but we believe it is likely that some more quarters of customer growth in Brazil remain ahead. For reference, management has guided to 100 million active customers in Brazil alone (vs 73m active customers in Q4-23).

Given the low-and-grow approach we discussed above (see Figure 3), existing customers unlock higher credit limits over time, which results in the increase in usage for credit card and personal loans seen below. Together with net customer growth, this has been, and is expected to continue be, the main driver of Nubank’s growth in Brazil.

To calibrate the runway for increased usage, note that currently Nubank is the primary bank account for 27% of the Brazilian Population, but only captures 7-13% Market share for credit cards and personal loans. Management has stated that there is no reason why, in the long term, that market share should not converge towards principality, and intuitively we agree. If that happens, credit card revenues could ~2x and personal loans could ~4x.

As a second triangulation, see below a comparison for Monthly Average Revenue Per Active Customer (ARPAC) between Nubank and incumbents. As discussed in Figure 8, credit card and personal loans (together with payroll loans) are the main contributors to revenues. Nubank is currently at a fraction of incumbents’ ARPAC, therefore suggesting opportunity for growth. Management has indicated that they do not expect to achieve the same ARPAC as incumbents, given lower fees and lower share of wallet with high-income customers, but feel that $25-30/month should be achievable in the medium term. If true, that would support significant expansion in credit card and personal loans.

This leads to the third pillar of growth: rolling out payroll loans. Nubank has recently launched a payroll loan offering that, according to management “has been met with absolute enthusiasm”. Like all Nubank products, the distribution is entirely digital, which in the case of payroll loans is particularly relevant given the Brazilian market is still heavily reliant on brick-and-mortar brokers (locally called “Pastinhas”), which account for ~40% of the payroll loans originations.

We believe Nubank’s Payroll offering is likely to succeed because of the following reasons:

       I.          Nubank D2C distribution eliminates origination fees and, together with offering a lower interest rate than incumbents, results in an overall better proposition for customers.

      II.          In April 2024 Nubank started accepting portability of payroll loans and, leveraging Open Finance (see Figure 13), is targeting customers with high-interest balances at other financial institutions, proactively offering the opportunity to easily port the loan to Nubank and obtain a lower interest rate.

    III.          In an environment of declining interest rates (see RHS of Figure 11) the dynamic described above is particularly powerful, given payroll loans are fixed rate.

   IV.          ~40% of all payroll loans in Brazil are currently drawn by individuals that already have a Nubank bank account, as per management commentary. Therefore, this is primarily a cross-selling exercise to the existing customer base.

Payroll loans is a large, but less profitable asset class than credit card and personal loans. Doing a quick back of the envelope math, if Nubank achieves 10% market share, that would translate to a ~$13Bn portfolio, which at a ~18% annual interest rate, ~8% cost of deposits and ~2% Charge-off rate, yields ~$1Bn of pre-tax income, which is small when compared to Nubank’s $57Bn current market cap. However, payroll loans are a key product that Nubank was missing until recently, and we believe that offering it is likely to increase principality and add to the competitive moat vs other digital players.

Across credit cards, personal loans, and payroll loans, Nubank has a clear opportunity, in our view, to gain market share. However, it is important to note that all these products present a fundamental tailwind in the form of general market growth, the fourth pillar of growth. Total outstanding balances in Brazil have grown, in USD, at 8-15% p.a. over the L5Y. This reconciles with Figure 6, which shows the increasing household debt as a percentage of GDP and the still significant gap vs developed economies. Based on our conversations with industry participants, this growth is generally expected to continue over the next few years.

The fifth pillar of growth revolves around the opportunity to increase penetration within the high-income segment. According to management, Nubank already counts among its customers ~60% of the high-income population in Brazil. The figure reconciles with the LHS of Figure 4. However, to this date, Nubank has struggled to drive usage from this customer group, lagging incumbent banks in Share of Wallet.

I always use my credit card from Itau, as opposed to the one from Nubank. They give me a much higher credit limit and there are a lot more ways of using my cash-back points. On high-income, Nubank is clearly behind, even when comparing to much smaller start-ups like C6.        –Cofounder of Drip, a fintech startup in Brazil.

Over the last few quarters, Nubank has been developing their high-income offering, called Ultraviolet, which is free for customers that spend more than $1k per month on their credit card. They have finetuned their credit models to increase limits for this segment, implemented a cash-back offer, signed partnerships with companies like Rappi and HBO, and started adding functionalities like a low-cost multi-currency bank account designed for foreign travel. As a result, the purchase value on their Ultraviolet credit card has increased 2x over the last year to $1.1Bn in Q4-2023, but still constitutes only ~3.5% of Nubank’s total purchase volume.

We intuitively believe that Nubank will likely be able to develop, over time, a high-income offering comparable to that of incumbent banks. They could also choose to share part of the cost to serve differential vs incumbents in the form of better cash back or incentives. However, we admit that we are unsure about the role of Brand in this segment. Nubank is perceived in Brazil as a mass market brand, and we could see a world where high-income customers refuse to use Nubank for what the brand evokes, in a similar way that we would put only so much credibility on a theoretical Capital One’s bid to compete with American Express in the US. Our conversations with industry participants in Brazil have yielded different answers to this question, and our understanding of the behaviour of the Brazilian consumer is not deep enough so that we can fully underwrite Nubank’s ability to win in this sector. Therefore, we treat this growth opportunity as an upside.

The sixth pillar of growth revolves around Nubank rolling out adjacent products. In management’s view, consumers in Latin America tend to concentrate the purchase of all financial services in one provider, and Nubank is therefore striving to become a full-service platform. The main products they are currently rolling out are:

       I.          Insurance: their core offering to date is life insurance, which they offer through a partnership with Chubb. They are currently developing car insurance and home insurance products, which are in testing phase.

      II.          Investments: In 2020 Nubank acquired Easyinvest, a digital broker in Brazil. After struggling more than expected with the integration, Nubank currently offers a full-service investment platform that includes stocks, bonds, ETFs, mutual funds, etc.

    III.          Cryptocurrencies: In 2022 Nubank started offering the possibility to buy cryptocurrencies through their app. For now, they offer 15 cryptocurrencies, including USDC, a stablecoin pegged to the Dollar that has received significant interest in Latin America.

   IV.          Shopping: Nubank’s mobile app includes a tab where customers can purchase e-commerce products from more than 200 retailers. Nubank captures a price per lead and leverages its 90m customers in Brazil to obtain better offers for their customers. In 2023, Nu Shopping received 255m total visits.

     V.          BNPL: Nubank offers a BNPL product called NuPay, through which customers can easily check-out online and obtain additional financing options. Currently the product has 3.5m users and is accepted by 9% of the Top 150 retailers in Brazil, as well as apps like Uber or iFood.

   VI.          SME Accounts: Nubank counts among its customers millions of self-employed individuals and microentrepreneurs. To address their needs, Nubank has developed a “pessoa jurídica” offering that allows these individuals to set up a commercial bank account, receive a credit card, accept payments through the NFC reader of modern Smartphones, and manage their invoicing and tax workflows. To date, Nubank has 2.2m commercial users, and in March 2024 they announced an expansion into lending for this segment, by launching a Working Capital financing solution.

In general, the overall size-of-the-prize for these products is small. As of today, the fees associated with all the products above contribute about ~2% of total revenues (excl. SME credit card, which is not disclosed separately from consumer credit card). However, we believe that they are a useful tool to increase principality and, thus, the value will arise from a higher usage of Nubank’s main revenue engines, namely credit card and personal loans.

To this point, all the drivers presented revolve around Brazil. This brings us to the last pillar of growth: geographical expansion. In 2019 Nubank expanded into Mexico, and in 2020 into Colombia. In 2023, Nubank obtained banking licenses to be able to accept deposits in both markets. We believe that the market dynamics is both cases are such that Nubank could replicate the same strategy as in Brazil and achieve similar levels of success. Combined, Mexico and Colombia have the same GDP as Brazil, so over time they can become meaningful contributors to the P&L.

On several interviews, Nubank’s founder has repeated that more countries will come in the next 2-5 years. Interestingly, he has stated that in a 5-year horizon, he envisions Nubank expanding outside LatAm. We admit we are somewhat puzzled by this assertion.

When looking at the remaining countries in LatAm, we intuitively believe they are all too small to warrant management losing focus on the very large runway remaining in the 3 countries where Nubank already operates. Banking is an extremely local endeavour, with varying legislations, types of products and needs for licenses. Therefore, we like Nubank’s current strategy of going narrow but deep (as opposed to other players, like Revolut, who chose a global but point-solution approach).

When looking outside LatAm, the obvious candidate for expansion would be the US, potentially seeking to target its Hispanic population (65 million individual). However, the US is a very different market, with highly sophisticated competition and comparatively very low interest rates on personal credit. We are unsure how could Nubank build a successful operation in the US without significant RoE dilution. If not the US, we admit we cannot imagine Nubank travelling to India or other developing markets where, on another hand, their product would likely have a better fit.

With any other management team we would likely be concerned with these remarks. But given Nubank’s exceptional execution track record to date and the full alignment of interests between Shareholders and the Founder, we feel inclined to give David Velez the benefit of the doubt also on this point.

5. Deep-Dive into the Mexican Opportunity

Nubank entered Mexico in 2019 and followed the same roadmap it did in Brazil 5 years earlier. Given accepting deposits requires a banking license, and this is a time-consuming endeavour, they started with a credit card offering. That had the benefit of being able to start acquiring customers while the regulatory dialogue progressed and, more importantly, to start adapting the credit underwriting models to a new geography.

The process to train these models starts by giving credit to a large number of customers with different backgrounds, observing how the repayment rate unfolds over the next couple of years and, eventually, looking for correlations between repayment rates and customer characteristics. This initial survey results in a heavily loss-making cohort, where the priority is not to maximize risk-adjusted returns, but to extract as much data as possible to train the credit models.

Hence, until Q2 2022, Nubank focused on acquiring as many customers as possible. Given incumbent banks charge significant annual fees for a credit card (e.g. BBVA currently charges USD50/year for the entry-level card), by simply offering a free credit card Nubank was able to generate significant demand.

In that period, Nubank climbed to position #5 in Mexico by total number of cards issued. Given that this initial cohort was not intended to obtain risk-adjusted returns but to gather data, and in order to cap the initial investment, Nubank reduced the approval rate for new customers, while waiting for the original cohorts to age and the credit models to improve. That created the slowdown between Q2 2022 until Q2 2023 seen in Figure 21.

In Q2 2023 Nubank was finally granted a Mexican banking license. That allowed to onboard customers even if not approving them for a credit card, given Nubank could offer instead a bank account with an associated debit card. Nubank therefore resumed customer growth, even if not yet expanding the loan balance (see below).

Overtime, the NPLs of the original credit cohort started to decline, as aging non-performing loans where progressively charged-off, the models improved, and the performing customers were retained. As such, Management has indicated its willingness to reaccelerate loan growth, which can already be seen in the recently published Q1 2024 Mexican regulatory filing.

On the Deposit side, Nubank is pursuing a rather aggressive strategy. In November 2023, they announced they would remunerate deposits at 15% p.a. This compares with BBVA, Mexico’s largest bank, having an average cost of deposits of 3.7% p.a., as per Central Bank filings. This is a totally off-market rate, and the result can be seen in the explosive growth in deposits observed above, with deposits increasing 5-fold in Q4 2023 and 2-fold again in Q1 2024.

Nubank currently has $2.2Bn in Deposits and only $0.9Bn in Loans. Most of these loans respond to the current revolving balance of credit cards (and hence do not pay interest). However, the interest rate is high enough on the interest-paying portion of the credit card that the Net Interest Margin is still positive at 2%. The Mexican subsidiary, however, was overall loss making in Q1 2024, printing a $45m pre-tax loss (as per CNBV data) vs ~$400m positive net income from the Brazilian subsidiary (as per Jan-Feb BCB data).

Management has acknowledged that this is a large investment for Nubank. The path forward revolves around (i) expanding the loan book in Mexico to utilize those highly remunerated deposits and (ii) eventually decreasing the interest rate paid on deposits. Management believes that their offering is not yet mature enough in Mexico to be able to fulfil the entire needs of their customers, and hence they need to provide an alternative incentive. Once the offering is more mature, they intend to reduce the deposit remuneration, in line with what they did in Brazil in Q3 2022 (and which led to virtually no churn).

It is important to note that other digital banks are currently operating in Mexico. Klar, Stori and Uala are the 3 most prominent independent start-ups offering a credit card and a bank account. All of them have followed Nubank in offering remuneration rates around 15%. However, their offerings are gathering limited traction, as seen below. We believe it will be hard for independent start-ups to compete with Nubank because:

       I.          They lack Nubank’s financial muscle. Nubank has invested $1.4Bn to date in Mexico, and it is recycling a high proportion of the infrastructure and products developed for Brazil. In contrast, Klar and Stori have each raised less than $400m and need to build a new bank entirely from scratch.

      II.          Nubank has a decade of experience underwriting credit and has built one of the most sophisticated credit operations in the world. Being able to extend more credit to their customers is a key differentiation in Mexico, and based on our conversations with industry participants, Klar, Stori and Uala are at a much earlier point in their credit journey.

On the other hand, Mercado Libre, through their fintech arm Mercado Pago, is a formidable competitor. We will spend a lot of energy in Section 7 discussing about them, but for now it suffices to say that:

       I.          Mercado Pago launched in Mexico 3 years earlier than Nubank, and still has more active bank account customers than Nubank (~14m vs ~7m), even if the latter is closing in.

      II.          However, Mercado Pago does not have a banking license, and therefore cannot use deposits to fund loans. Hence, it can only offer a lower remuneration for Deposits (10% vs 15% for Nubank).

    III.          Mercado Pago historically struggled with credit underwriting across both Brazil and Mexico, and for now they act more as an electronic wallet than a full-service bank.

   IV.          However, Mercado Pago launched a credit card offering in February 2023, and they currently have 1m credit cards issued (vs more than 4m for Nubank).

All in all, Nubank is making a big investment in Mexico, but for now it seems to be paying off, with Nubank taking the leadership among emerging Fintechs.

Stepping back and focusing on the market, Mexico shares a lot of characteristics with Brazil. The Top-5 banks hold a very large proportion of the market, fees and interest rates on loans are elevated, bank RoEs are high, and customer service poor.

However, the Mexican regulator has historically been a lot slower and less sophisticated than the Brazilian one, and has done a worse job at incentivizing competition and the digitalization of the economy. Also, credit cards in Mexico have a different structure than in Brazil, with retailers being reimbursed in 3 days from purchase date, as opposed to 30 days. Therefore, credit cards in Mexico are capital intensive, and it is incredibly costly for start-ups without a banking license (and hence the ability to accept deposits) to ramp up a successful credit card operation in Mexico.

As a result, today Mexico presents a situation similar to that in Brazil 10 years ago. Cash is still the primary transaction method and credit is restricted to high-income individuals. However, according to our conversations with industry participants, as it happened in Brazil in the mid 2010s, the economy is experiencing a strong digitalization wave that we believe constitutes a great hunting ground for Nubank.

When it comes to underwriting credit cards, banks in Mexico are much more conservative than in Brazil. For example, Mexican banks for the most part only issue credit cards to customers that have kept their payroll with them for at least 2 years. Nubank, on the other side, has a much more sophisticated underwriting operation, and is able to provide credit cards to consumers that never had a credit card before. - Former Head of Credit and Analytics at Nubank Mexico.

6. RoE, Profitability, Valuation and Long-Term Potential

To this point, we have long discussed Nubank’s runway for growth. In this section, we will argue that this growth will be achieved with a very high RoE. In particular:

       I.          RoE in the Brazilian operation is already as high as 60%.

      II.          Underlying earnings are higher than reported earnings.

    III.          In the long-term, earnings have the potential to reach ~$4/share (vs spot price of $11.76/share).

Starting with Return on Equity, Nubank printed 23% of consolidated RoE in Q4 2023 (fully loaded, after share based compensation and exceptionals). However, this figure is severely diluted by having (i) $2.4Bn in cash at the holding company in Cayman Islands and (ii) capital trapped in the loss-making Mexican and Colombian subsidiaries.

Every semester, Nubank files with the Brazilian Central Bank detailed financials for the Brazilian subsidiary. Using those filings, we can compute the evolution of RoEs presented below. Given Q4 was significantly more profitable than Q3, we can safely assume that RoE of the Brazilian Subsidiary is currently ~60%.

With that we do not intend to say that it would be prudent to distribute to shareholders the entire cash balance at the holding company, and we do not disregard the fact that the Mexican and Colombian subsidiaries are currently loss making. But we believe that the fact that the Brazilian banking subsidiary is able to achieve 60% RoE while both (i) comfortably meeting the regulatory capital requirements and (ii) achieving growth rates of ~60% p.a., is indicative of the long-term earnings potential of Nubank.

Also, note that management has repeated in multiple occasions that they do not expect to raise additional capital to support growth.

On another note, regarding underlying earnings, we believe that the IFRS treatment of loan loss provisions results in reported earnings that underestimate the true earning power of growth companies.

Under IFRS 9, a provision for the expected losses during the entire life of the loan needs to be expensed when the loan is first granted. That creates a revenue and expense recognition mismatch, whereby revenue (i.e. interest) is recognized through the entire life of the loan, while the expenses (i.e. loan losses) are recognized entirely at inception. That mismatch averages out for companies that do not grow, but severely penalizes earnings for growth companies like Nubank.

Nubank tracks the change in provisions due to growth and we believe that, adjusted for a 40% Tax rate in Brazil, should be added to net income to reflect the true steady-state earnings power of Nubank.

With this adjustment, Valuation is currently at 24x Q4 2023 Run Rate Earnings, which is the multiple we have used internally to value the business.

Note that we have not incorporated this adjustment into the RoE calculation. The reason is that overprovisioning has the effect of reducing earnings, but also book value, and therefore we should adjust for this effect both in the numerator and denominator. It would be hard to do with the available information, and the overall answer (that Nubank has a very large RoE in Brazil) would not change.

Finally, regarding the long-term earnings potential of Nubank, we think there is value in doing a back-of-the-envelope calculation on how profitable Nubank can become. Important to note that the below are by no means conservative assumptions, and are more an exercise of sizing the prize, as opposed to a business case we are fully underwriting. Our assumptions are the following:

       I.          Nubank can achieve 100m active customers in Brazil (mgmt. guidance, they are currently at 73m and adding 3-4m every quarter).

      II.          Nubank can achieve ARPAC of $25/month (mgmt. guides to $25-30, represents discount vs incumbents as seen on Figure 16). Note that mature cohorts already exceed that number, as shown in Figure 3.

    III.          That would translate to revenues in Brazil of $30Bn/year. This figure compares with $126Bn Retail Financial Services TAM in Brazil in 2021, as per Mgmt. analysis. That would represent a 24% market share, or 3 p.p. less than the percentage of the Brazilian population that, already today, use Nubank as primary bank account.

   IV.          Net Income margin can become ~30%. That compares with 25% for Nubank in Q4 2023 if adjusting for the impact of growth on provisions, and we therefore believe upside potential exists if further operating leverage materializes, as shown in Figure 26.

     V.          This leads to earnings in Brazil of ~$9Bn or ~$2/share.

$2/share is what we consider the full-potential earnings in Brazil. Outside of Brazil, mgmt. has stated that they believe they can become as large as in Brazil. On a GDP basis, Colombia and Mexico combined are similar to Brazil (Figure 20). However, their Retail Financial Services TAM is currently much smaller than Brazil ($51Bn vs $126Bn, as per Mgmt. analysis in 2021). The thesis is that the market will expand very meaningfully as the ongoing digitalization wave in Mexico progresses (see Section 5).

If we take Mgmt. at face value, we therefore arrive to long-term potential earnings of ~$4/share. As discussed, we do not necessarily endorse this figure, but we believe it serves as a calibration of the earnings potential of Nubank. Note that this figure does not include any underlying market growth, which is meaningful in Brazil, or expansion to additional countries.

In terms of the time required to achieve that number, we refer the reader again to the RHS of Figure 3. Note that cohorts take ~6 years to reach their full potential. In Brazil, Nubank already has a vast number of customers at different stages of maturity, and we therefore believe that in another 5-7 years, the business should not be far from its full potential. In Mexico and Colombia, expansion started more recently and therefore Nubank will need time to first recruit the customers and then let them mature. Hence full potential is likely to be 8-10 years away.

7. Competitive Position and Moat

In this segment we will address how Nubank competes with each of 3 main groups of competitors:

       I.          Incumbent banks

      II.          Other digital banks

    III.          Mercado Libre, through their fintech arm Mercado Pago

Starting with incumbent banks, we have long discussed Nubank’s lower-cost-to-serve advantage, and their willingness to pass most of these savings to its customers, creating a flywheel with larger scale feeding lower cost to serve and vice versa.

This flywheel is particularly strong for the low-income segment, where average tickets and credit balances are much lower and, therefore, Nubank’s advance has made it very hard for incumbent banks to serve this customer group profitably:

The whole market is trying to reposition towards high-income. I think low-income is a challenge for us and for the entire market. I think it is a major challenge and we do not have all the answers, but we are asking ourselves the questions now. We will only be able to afford to have a profitable low-income operation if we drop our cost to serve to a very low level. A 10 or 20% reduction will not be enough, it needs to be much lower. We need to redesign the entire value proposition, and also our brand positioning. – Banco Santander Brazil, Q3 2023 Earnings call.

We just want to stress, once more, that we believe an equally important differentiator lays on Nubank’s level of sophistication in underwriting credit, with mgmt. claiming to achieve 15-25% lower NPLs  than traditional banks when underwriting comparable customers. Our conversations with industry participants seem to confirm this:

In Brazil, incumbent banks operate the business as they always have, hiring the same type of people from the same universities. Nubank brought a different kind since the beginning, when they set up the whole credit underwriting operation to look like Capital One. They brought a lot of talent form Capital One, and until today, for the past 10 years, the senior leadership of Nubank has always been Capital One formers. They brought something different to the market. – Former Credit Analyst in Brazil at Nubank.

We finally note that Nubank also has developed a highly regarded consumer brand, given it was the first player to provide radically lower fees and better terms of service for their customers, creating strong loyalty vs incumbents. The reader can imagine, for example, how customers in Mexico feel today towards Nubank after being offered 15% for their deposits, vs 3-4% historically by incumbent banks.

On the negative, Nubank faces a higher cost of deposits, but in Brazil the difference is narrowing as Nubank’s offering expands, as shown on Figure 11.

When competing with other digital banks, we believe that credit underwriting becomes the biggest moat. Setting up a successful credit operation is incredibly complex, and we encourage the reader to try to find a start-up in the US or Europe that has scaled a big consumer credit portfolio.

Also, Nubank has currently been underwriting credit for 10 years and has therefore lived through 3 credit tightening waves (the 2014 Brazilian economic crisis, Covid, and the credit tightening of 2022) while carrying an outstanding credit portfolio of billions of dollars spread across tens of millions of customers. That has provided a massive amount of data on consumer behaviour and stress tested its models. A newer digital bank, even if it was able to set up the credit models in the same way that Nubank did, would face a major disadvantage in terms of the amount of data it controls (note that the number of data points that can be transferred with Open Finance is relatively small and limited to certain categories).

Given most digital banks are not listed and no public information is available, the best KPI we have found to gauge their success is the growth in the number of Monthly Active Users of their mobile apps, as reported by Sensor Tower. As can be seen, all other start-ups with the exception of Mercado Pago, are performing poorly, and our conversations with industry participants point to the inability to offer higher credit limits to their customers as the main reason.

Finally, when it comes to Mercado Libre, we believe it is the most formidable competitor Nubank faces. Founded in 1999 in Argentina, Mercado Libre has evolved to become the largest e-commerce platform in Latin America, comfortably beating Amazon. Over time, Mercado Libre expanded into Fintech by launching Mercado Pago, which originally was focused on offering 3rd-party e-commerce sites the capabilities to accept online payments.

In recent years, Mercado Pago has entered the space of retail financial services, where it competes head on with Nubank in Mexico and Brazil. Through their mobile app, Mercado Pago offers a bank account, debit cards and credit cards. Also, they have a strong BNPL offering, mostly catered to providing credit to consumers who purchase products on Mercado Libre’s e-commerce site, but also open to 3rd-Party e-commerce vendors.

Mercado Pago has two obvious advantages when competing with Nubank as a bank account:

       I.          The opportunity to attract customers by bundling financial services with their e-commerce site (in the same way that, for example, Amazon Prime Video competes with Netflix).

      II.          Possessing customer data about their e-commerce behaviour, which Nubank does not possess, and which could be used to better underwrite credit.

On the other hand, Mercado Pago has three obvious disadvantages when competing with Nubank:

       I.          They do not possess a banking license in Brazil or Mexico, and therefore they cannot use deposits to fund loans.

      II.          They have historically struggled with underwriting credit, do not have a long history of cohort performance data, and are currently experiencing very high levels of NPLs.

    III.          Their offering in Brazil is a lot less developed than Nubank’s, given their focus in multiple verticals (e-commerce, payments and credit) and geos (Argentina, Brazil, Mexico, Chile, Colombia, etc). They lack, for example, Pix financing, payroll loans, an investment account, or a high-income credit card offering with cash-back.

As of today, there is no question, in our view, that Nubank is far ahead of Mercado Pago when it comes to providing end-to-end financial services to individuals, particularly in Brazil. Note that Mercado Libre “only” has 54 million e-commerce users spread across Mexico, Brazil and Argentina, which have a combined population of 390 million. For those that are not frequent customers of Mercado Libre, we believe that Nubank is likely to remain a superior proposition for the foreseeable future. For those that are frequent customers of Mercado Libre, on the other hand, we can imagine a scenario where the race gets tighter over time.

We chiefly respect Mercado Pago, and we have no doubt that Mercado Libre will grow over time, and most likely, eventually be able to underwrite credit proficiently. We believe that the market is large enough to accommodate both Nubank and Mercado Libre, and remind the reader that our long-term potential for Nubank assumes a market share of under 24% ($30Bn Revenues in Brazil, vs $126Bn Retail Financial Services TAM as per management analysis), leaving plenty of room for Mercado Pago to thrive.

8. Risks

We acknowledge that an investment of these characteristics calls for a higher-than-usual attention to the “Risks” section of the investment thesis. However, we believe that Nubank is a safer investment than it appears at first glance. In our view, the following are the main risks Nubank faces:

       I.        Macroeconomic risk

      II.       Geopolitical risk

    III.       Regulatory risk

   IV.         Credit risk

     V.        Liquidity and Run-on-the-Bank risk

Starting with macroeconomic risk, there is no doubt that LatAm is a volatile market. Brazil, in particular, is etched in the recent memory of several investors given the country lived, in 2014, one of the most devastating crises of its history, with Brazilian equities and leveraged buyouts severely underperforming their US counterparts in the period 2012-2017. The origin of this crisis is primarily attributed to poor economic policy during the government of Dilma Rousseff, coupled with the 2014 commodity price shock that negatively affected Brazil’s exports. Between 2014 and 2016, GDP fell by 8.1%, making it the second worst peak-to-trough recession on Brazil’s record.

Note that the 2014 crisis led to a strong currency devaluation. However, over time, due to significant inflation, even if the currency devaluated, Brazil GDP in USD has not fallen with the Real. On a longer time horizon, GDP in 2023 is 1.6x the pre-GFC figure, when compared in USD terms. In that same period, the Brazilian Real has depreciated by over 50%.

We argue that this is the prevailing expression of financial crises in Latin America. Over time, the affected country continues to produce a broadly similar amount of total output, and the strong devaluation of the currency leads to an offsetting amount of inflation. The USD figure, which is the best gauge of true economic output, is therefore likely to remain broadly stable.

We believe Argentina is a good case study for this phenomenon, with a GDP in USD that is volatile but stable over time, even if the Argentinian Peso depreciated by 95% in the last 5 years. 

The key question, however, is how banks perform within this economic backdrop. As can be seen, banks have performed historically particularly well in high inflation environments, even when observing results on a USD basis, which is an unintuitive result. The reason for that is twofold:

       I.          Bank growth is strongly linked to inflation. Interchange fees is the most obvious link, but also higher loans can be expected if the value of currency falls in relation to goods and services.

      II.          Banks in LatAm present high RoEs, and can hence grow with inflation without excessive capital consumption.

As discussed in Section 6, Nubank has the highest RoE among all major banks in Brazil (60% vs ~20% on average), and thus we believe it is primely positioned to weather a period of high inflation and currency devaluation.

Also, note that while Nubank is currently primarily focused on Brazil, it is currently scaling its Mexican operations and, if successful, it would provide a source of macroeconomic diversification. Mexico has been somewhat more stable than Brazil in recent years, and its performance below serves as one more example of how USD-based GDPs in Latin America are stable over time, even if their currencies lose value.

As a final note on Macroeconomy, given we have used the comparison with Argentina as an extreme example of hyperinflation, we feel it is appropriate to point that Brazil is structurally a much better economy, because:

       I.          The Argentinian government has historically borrowed in USD, as opposed to local currency, while the Brazilian one has very little foreign currency debt.

      II.          The Argentinian Central Bank has no foreign currency reserves, vs significant reserves at the Brazilian Central Bank.

Also, Brazil presents a currency tailwind because it is significantly expanding its offshore oil drilling operations, which creates a positive impact on the Trade Balance, which has turned positive in recent years.

By no means do we claim to be able to forecast the Brazilian macroeconomic environment, but with the above we just seek to dispel the notion that the country is “un-investable”, which is a view that many fellow investors have expressed to us in recent years.

Moving on to the next type of risk, we believe what we have called geopolitical risk is an unlikely one, but one that if it materialized could have significant consequences. Brazil has been a democracy since 1985. However, its current president, Luiz Inácio Lula da Silva, is a big proponent of the BRICS alliance (Brazil, Russia, China, India and South Africa). As such, he has publicly stated he plans to visit Russia in 2024, rejected to enforce the international arrest warrant pending on Vladimir Putin were he to visit Brazil for the G20 that will occur in November, and formed close ties with China, which is the biggest importer of Brazilian goods. There is no current prospect of Brazil facing sanctions from the US, but in an increasingly divided world, we would feel our funds would be less likely to be caught between international sanctions if they were invested in a country with a different set of strategic allies. We remind the reader, however, that Nubank has its primary listing in the NYSE.

A far more tangible one, we believe Nubank has a moderate level of regulatory risk. As we discussed in Section 2, the government and the BCB are both laser-focused on promoting competition and ensuring that consumers experience a lower level of interest rates and fees. In the last 2 years, at least three pieces of legislation have been passed which impacted banks’ profitability:

       I.          Implementation of an interest rate cap of 1.7% p.m. for certain types of payroll loans, in particular those backed by social security (as opposed to government salaries). That did not affect Nubank, given it was not yet offering payroll loans. Also, Nubank is currently offering a rate of 1.4% for this specific type of loan, well below the cap.

      II.          Reduction in the interchange fee for a certain type of debit cards, called prepaid cards. The change was implemented in April 2023, and the interchange fee was capped at 0.7%, down from ~1.2%. That did affect Nubank, erasing ~2% of its revenues overnight.

    III.          As of January 2024, a new rule came into effect that limited the total amount of interests to be paid on a credit card transaction to 100% of the original purchase value. As discussed, Nubank is very proactive in reducing the revolving portion of the credit card balance, which is the main cause of excessive rates (see discussion on Section 2). The impact on revenue will be negligible, with Nubank having published that the percentile 99th of credit card transactions that pay the most interest is at 40.5% of the original purchase price, far below the 100% cap.

We agree with the government approach to limiting what we believe are abusive practices. We recognize that some of the future pieces of legislation might be a headwind for Nubank’s profitability, but drive comfort from two thoughts:

       I.          Nubank’s model is predicated on offering the lowest fees and interest rates in the market, and will thus be one of the last banks to be affected.

      II.          Nubank’s RoE is currently at 60% and growing, vs traditional banks at ~20% and declining.

Both things considered, we believe that if more stringent legislation is passed, it will far earlier drive the incumbent banks to not earn their cost of capital than it will meaningfully affect Nubank’s profitability. We do not necessarily dislike a world where incumbents are forced to pullback and Nubank exchanges lower unit profits for higher scale, with the positive effect that this would have on Nubank’s competitive moat. Even if we consider the behaviour of the Regulator certainly a risk, we believe it is unlikely to become a severe headwind for the business.

On a different note, regarding credit risk, a lot has been written in this report about Nubank’s underwriting capabilities. Beyond the fact that, as discussed, their credit models have been stress-tested across 3 different credit tightening cycles, two elements provide additional comfort:

       I.          Nubank is very well provisioned, with a provision to 90+ NPLs coverage of 225%.

      II.          It has ~3.5x more capital than the regulatory minimum which, together with provisions, provides for a large loss absorption capacity.

As for liquidity risk, it is important to note that financial instability in Latin America has led to bank runs before (e.g. “Corralito” in Argentina in 2001). Brazil has not seen this dynamic to date, but was there to be widespread financial panic, we believe Nubank would be likely to be one of the last banks standing:

       I.          Loan-to-deposit ratio stands unusually low at 34%.

      II.          Cost of deposits for Nubank is ~80% of the Brazilian SOFR-equivalent, called CDI. A run on the bank that brought deposits to zero would force Nubank to, at most, borrow $8Bn (the amount of interest-earnings loans outstanding) at 100% of CDI, which would have an impact on profits of ~$40m per quarter vs borrowing at 80% of CDI. That compares to earnings of ~600m in Q4-2023, when adjusted for IFRS 9 impact of growth on provisions.

    III.          Nubank’s loans are very short term (6 months on average) and, if needed, Nubank could thus decrease its liquidity needs quickly.

For all these reasons, and strictly in terms of credit and liquidity risk, Nubank is the bank with the least amount of existential risk that we have encountered in a while.

9. Conclusion

We encourage the reader to think what the appropriate valuation would be for a US-based company with the following characteristics:

       I.          27% of the adult population use it as their primary banking provider.

      II.          Has 3x the RoE of the best incumbent banks, like Bank of America.

    III.          Continues to grow inside the US at ~60% per annum.

   IV.           It has found a second market somewhere with the same size and conditions as the US, and is ramping up quickly from a very low base.

We acknowledge the US is a much better economy than Brazil and Mexico, but at 24x Q4-2023 Run Rate Earnings (after SBC and exceptionals) we feel we are appropriately remunerated for the risk. We believe this is a truly unique situation and we are therefore proud owners of Nubank.