Following a 30% share price decline, Axos Financial is fully valued.
Axos Financial does not have a competitive advantage in the online banking industry.
Growth is likely to decelerate dramatically compared to historical rates.
Editor's note: Seeking Alpha is proud to welcome Maurice Spolan as a new contributor. It's easy to become a Seeking Alpha contributor and earn money for your best investment ideas. Active contributors also get free access to SA PREMIUM. Click here to find out more »
Axos Financial (AX) does not have a competitive advantage in the online banking industry. As a consequence, balance sheet growth is likely to decelerate meaningfully over the next decade while profit margins are at risk of long-term contraction. My intrinsic value estimate for Axos Financial is between $28 and $35 per share, suggesting that the stock is fully valued at its current $29-30 market price.
Axos Financial was founded in 1999 as the country's first online-only bank. Unencumbered by costly branches, Axos runs a very lean operation, consistently reporting a sub-forty percent efficiency ratio (non-interest expense/total income). This enables Axos to pay depositors savings rates much higher than what they can receive at traditional brick and mortar banks. Deposits fueled the balance sheet to 25% annualized growth between 2011 and 2018.
Axos is primarily a real estate lender, with 78.5% of the loan book allocated to mortgages across Single-Family Secured (54% of total loans), Multi-Family Secured (21% of total loans) and Commercial Secured (3.5% of total loans). An additional 17.5% of loans are in Commercial & Industrial channels. Axos' lending activity is concentrated in California, as over 72% of loans originate in the state.
Despite expensive deposits, Axos produces a strong net interest margin, between ~3.90% and 4.10%, by lending in riskier verticals such as non-conforming jumbo mortgages. Yet, Axos' underwriting appears judicious. Axos makes low LTV loans, around 55%, and reported minimal Nonperforming Assets over the prior five years, between 0.50% and 0.60% annually. Axos has merely three, one and 27 basis points of cumulative losses in its Single-Family Mortgage, Multi-Family and Commercial & Industrial origination lifetimes, respectively.
Between 2011 and 2018, Axos achieved consistent returns on equity around 17% and grew book value over 7-fold. Axos' share price increased more than 1000% in this period, reaching all-time-highs of $44 per share in June 2018. However, over the past 18 months, Axos' growth and profitability have deteriorated, as fiscal 2019 net income rose just 1.7% year-over-year and the efficiency ratio elevated above 50%. As a result, shares declined to around $30. The tumble can be attributed to four causes: Universal Digital Bank, a trading loss, subsequent investment in risk management, and an unprofitable acquisition.
In June 2018, Axos launched its Universal Digital Bank, an initiative to centralize all of the bank's offerings under a single platform. Building Universal Digital Bank necessitates considerable investments in technical infrastructure and hiring, as indicated by marked year-over-year increases in Salary, Data Processing & Internet, Depreciation/Amortization, Equipment, and General & Administrative.
Later, in October 2018, Axos made two acquisitions to create a Securities Business, purchasing COR Clearing, a clearinghouse for correspondent broker-dealers, and WiseBanyan, a robo-advisor with $150 assets under management. In March 2019, COR suffered a $15 million trading loss. Axos took measures to shore up risk management systems, but the one-time loss and risk investments damaged 2019 profitability. WiseBanyan, meanwhile, was effectively an expenditure to construct a retail investment product. CEO Greg Garrabrants calls it "an acqui-hire that cost 10x to build what we bought it for." On the same call, management reported that WiseBanyan was running at an annualized loss run-rate of $14-16 million, immediately destructive to earnings.
To summarize, Axos built a valuable franchise over the past decade by minimizing non-interest expenses through its branchless model and passing its cost savings onto the consumer through attractive deposit rates. This allowed the bank to gather deposits at an extraordinary pace as brick and mortar institutions could not compete with the low-cost structure. However, over the past 18 months, Axos' growth and non-interest expenses have deviated from historical results. What do Axos' long-run prospects look like? Has a buying opportunity been created?
Value Drivers: Asset Growth, Net Interest Margin, Efficiency Ratio
To answer these questions, I will examine the prospects for each of Axos' three value drivers: balance sheet growth, net interest margin and efficiency ratio. I will make 10-year forecasts of each driver to arrive at an intrinsic value. My forecasts will be founded on the underlying thesis that Axos does not have a competitive advantage in the online banking industry.
This perspective is based on the following observations. Axos' high-rate deposits have been widely replicated, both by online-only competitors and bifurcated divisions of physical institutions. As an increasing number of alternatives have come to market, Axos' growth has decelerated and its market share has declined. To compensate, Axos has turned to inorganic, less sustainable means of growth. These means of growth are not protected by Axos' low cost structure or its head start in the online banking space.
I believe that over the next decade, Axos will likely forfeit either its market share or its profit margins. Because many online banks pay higher deposit rates than Axos, Axos will need to hike its own rates, provide additional services that consumers value, or both, in order to keep pace with competitors. In any case, margins are injured. If Axos rebuffs the customer to preserve margins, its market share will fall.
At this point, my findings have been sufficiently summarized. The following section will allow the reader to see evidence for my assumptions and challenge my reasoning.
Long-Run Balance Sheet Growth: 7.30-9.70%
In the period between 2011 and 2016, Axos grew interest-bearing deposits and assets at an astounding annualized rate of 32.5% and 31.4%, respectively. During this period, Axos' depository rates fell from 1.85% in 2011 to 0.87% in 2015, rising only slightly to 0.91% in 2016. That Axos grew so torridly while lowering rates suggests that it was benefitting from a dearth of high-yield savings account alternatives.
Since 2016, however, Axos' results suggest that it has been contending with a proliferation of high-rate, online deposit competitors. Between 2016 and 2019, interest-bearing deposits and assets decelerated sharply to annualized growth rates of 11.4% and 13.7%, respectively, while interest-bearing deposit rates moved from 0.91% to 1.99%. This demonstrates that Axos was fighting harder to gain deposits via higher rates, yet accruing less growth as many high-rate substitutes emerged.
Because Axos' high rates have been commoditized, the bank sustained its high double-digit asset growth since 2016 through pathways beside organic deposit gathering. First, the business nearly tripled its deposit contribution from non-interest-bearing accounts through entering a partnership with H&R Block. Second, Axos made five deposit-related acquisitions: two pure deposit purchases and three whole business acquisitions with deposit-generative features. Last, Axos extended its balance sheet leverage across all reported capital ratios. The key point is that these means of growth, upon which Axos has become increasingly reliant, have no moat around them.
Source: Created by author
Axos' market share fell 2% between May 2018 and September 2019 after it increased 3.5x over the prior eight years. This is the result of slowing organic growth and intensifying competition. But this image also gleans two challenging characteristics of the industry for its inhabitants: minimal customer captivity and low barriers to entry.
That Axos more than tripled its market share of online deposits in eight years suggests that share changes hands freely in the industry. While Axos benefitted from this as it pioneered the online banking space, it is susceptible to losing the share it grabbed over the past decade. Additionally, Axos' exceptional return on equity throughout the period, with extremely small scale, indicates low barriers to entry. If an entrant in an industry can earn high returns while garnering only a small market share, a great number of businesses will be motivated to enter. This is what we have seen in online banking.
I believe that the most logical way to analyze a bank's growth runway is to estimate its ability to gather deposits. To this end, I will forecast total deposit growth, online's deposit share of total deposits, and Axos' share of online deposits. Only my estimate of Axos' share will change. The high end of my range represents Axos' decision to forgo margins in favor of growth, while the low end indicates the converse. My assumptions are detailed in the following paragraph.
Total deposits will grow 6% between 2019 and 2029, equal to the rate between 2010 and 2019. Online deposits will increase share 35 basis points annually, similar to its annualized share grab over the prior decade, settling at 12.2% of total deposits in 2029. In the event Axos sustains margins, market share will decline at the same pace as the period between May 2018 and September 2019, such that market share is 0.64% of online deposits in 2029. If Axos offers higher deposit rates and/or spends more on services, damaging margins, it will retain share at 0.80%. Finally, Axos' deposits will remain a stationary percentage of total assets (roughly 80%), and total leverage (equity-to-assets) will not change. These assumptions result in Axos 2029 total assets of $22.7 billion and $28.4 billion, a compounded annual growth rate of 7.3% and 9.7%, respectively.
Long-Run Net Interest Margin: 3.60-3.80%
Between 2015 and 2019, as Axos significantly increased deposit rates, it was able to expand net interest margin from 3.91% to 4.14%, through three tactics. First, it increased its allocation of higher-yielding Commercial & Industrial loans, from 5% of the loan book in 2015 to 17.5% in 2019. Second, the agreement with H&R Block is accretive to net interest margin by virtue of its non-interest-bearing deposits. Last, the five acquired sources of low-cost deposits brought down the cost of funding. Fundamentally, I believe that Axos' offsetting measures to its spread are unsustainable and subject to competitive pressures for four reasons.
Primarily, if Axos' spread declines over time as it competes on deposit rates, sustaining the net interest margin will become more challenging. The spread is the focal point of net interest margin and counteractions cannot resolve a collapsing long-term spread. In 2019, Axos spread declined to 3.66%, its lowest since 2013. Next, if Axos allocates more of its loans to Commercial & Industrial (which is riskier than low LTV real estate), eventually its credit quality is going to decline. If Axos maintains its net interest margin through this method, underwriting metrics should be monitored diligently. Third, Axos' low-cost deposit acquisition strategy depends on consistently successful capital allocation. As high-yield online deposits continue their share grab from total deposits, acquiring low-cost deposits will become more competitive. Finally, Axos agreement with H&R Block will be going on its fourth year in tax season 2020. Maturation and year-over-year decelerations may occur over time, which would dilute the partnership's positive impact on net interest margin.
Axos guides for a long-term net interest margin of 3.80-4.00% and has been remarkably adroit at landing in that range since 2013. Moreover, Axos refuses guidance on net interest margin's constituents, advising curious analysts to model based off of the target range. Net interest margin is the most opaque value driver, so I will opt for conservatism through the following estimates. If Axos maintains its share of the online deposit market, net interest margin will fall below the target range, to 3.60% in 2029. If Axos bleeds market share in favor of net interest margin, I will account for 2029 net interest margin at 3.80%. In both cases, I will model linear declines.
Long-Run Efficiency Ratio: 45-52%
In Q1 2020, Axos efficiency ratio was 53.6%, significantly higher than the range of 34-41% between 2011 and 2018. As previously discussed, Axos' spike in efficiency ratio is owed to investments in Universal Digital Bank and lackluster results in its acquired Securities Business. In order to make an estimate of Axos' long-run efficiency ratio, I will isolate the impacts from the Securities Business and Universal Digital Bank. The result will represent Axos' efficiency ratio when the Securities Business is functioning at expected profitability and growth expenditures in Universal Digital Bank have ceased.
In Q1 2020, the Securities Business ran at a 96% efficiency ratio. However, management's long-term guidance is for mid-seventies efficiency. Furthermore, the Securities Business' fee income, on an annualized basis, was down . If we increase Securities' fee income such that it is flat from the time of acquisition and operate the business at its long-run efficiency ratio target, Axos' consolidated efficiency ratio in Q1 2020 was 51.9%. This makes clear the extent to which Axos' heightened expenses are being driven by its investment in Universal Digital Bank, as the poor-performing securities business has only a modest impact on consolidated efficiency.
In order to estimate the efficiency ratio once investment in Universal Digital Bank ceases, I will adjust the expense items that rose most significantly in Q1 2020, assuming that such expenses represent the building costs of Universal Digital Bank. In order to complete this exercise, I will assume that such expenses are fully variable with sales. This will allow me to categorize a proportion of costs as "expenses" - those that will naturally recur, versus "investments" - those that will only appear so long as Universal Digital Bank is under construction. I will demonstrate the process through the following example. Between Q1 2019 and Q1 2020, total income grew from $102 million to $122 million, an increase of 19.6%. "Data Processing," however, expanded from $4.735 million to $7.81 million, a rise of 65%, or $3.075 million. When I grow Data Processing by 19.6%, thereby assuming it is a variable expense, the result is $5.66 million. The difference between $5.66 million and $7.81 million, or $2.15 million, is an "investment," whereas the increase from $4.735 million to $5.66 million, or $0.925 million, is an "expense."
I carried this process out through each notable line item and deducted the total "investment" quantity from Axos' Q1 2020 non-interest expenses. The resultant efficiency ratio, still incorporating normalized fee income and expenses in the Securities Business, is 45.6%. However, in Q1 2020, Axos slashed Advertising & Promotional spending by roughly 15% year over year, an irregularity compared to prior periods when it grew around the rate of total income. If we continue the adjustments and mark-up advertising to the rate of sales growth, the efficiency is 46.9%. This represents my best estimate at Axos' normalized efficiency ratio over the most recent period. The spike in efficiency is not merely a transient byproduct of UDB investment.
However, this process is imprecise on two fronts. First, Q1 2020 was the first reported period in which the Securities Business was fully consolidated on the financials, and as such, the only period in which such estimates could be made. It is foolish to consider 10-year forecasts based off of one quarter as reliable. Second, Axos is likely to possess some degree of operating leverage whereby the efficiency ratio decreases as sales grow. Therefore, I am going to use a range of efficiency ratios in my forecasts.
The period of normalized 46.9% efficiency coincided with the Company surrendering 2% of its market share. Therefore, should market share continue its declines at a similar pace, I will assume that the cost structure remains stationary. I will also assume that investment in Universal Digital Bank persists through 2021. In this scenario, I will model Axos' efficiency ratio as 53.6% in 2020 as the Securities Business costs remain high, 51.9% in 2021 as the Securities Business stabilizes but investment persists, and 45-47% in 2022-2029 as the estimated long-run range. If, however, Axos sacrifices efficiency to maintain market share, I believe that long-run efficiency will remain around 50-52% as the business continually invests in Universal Digital Bank and other services to retain and grow customers. In this circumstance, 2020-2021 efficiency will remain identical to the former scenario but 2022-2029 efficiency will stay elevated at 50-52%.
Valuation: $28-35 per Share
To value Axos, I built an "excess capital available for distribution" model, where net income generated beyond the amount necessary to maintain the capital structure, at a predetermined level of leverage, creates value for shareholders. I discounted available capital over the life of the model and received a fair value - nothing more than a modified Discounted Cash Flow format.
Assumptions embedded in the model, other than those heretofore discussed, include an equity to assets leverage ratio of 8.99% (median between 2011 and 2019), a 27% tax rate (fiscal year 2019 rate), a 10% cost of equity and a 2% terminal growth rate.
The model produces an intrinsic value estimate of roughly $28-35 per share across all ranges of the value drivers, suggesting that Axos is fully priced at $29-30 per share. I would not recommend purchasing Axos Financial because the business is facing intense competitive pressures and is fully valued.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.