An analysis of digital banks’ business models
All of the content below is my own view and does not reflect the view of my employer. I do not represent my employer in this article.
I’ve been studying digital banks worldwide to understand their business models for the past 1.5 years, and would like to share some of the observations I made. I won’t say that there is anything groundbreaking to those who have some degree of familiarity to the banking sector, but it did provide me some insight into what made the profitable digital banks what they are. Just to note, this analysis is conducted based on public sources i.e. their annual reports, media reports etc., so it is based on facts either audited by third parties or that comes from the horses’ mouth itself. Out of personal interest and professional necessity, I combed through a number of these annual reports to search for any useful trends and details that may explain their current predicaments. These observations may assist applicants in thinking through the business model they want to propose, and I also appreciate any extra views to explain these trends if readers have access to and can publicly disclose information I don’t possess. I’m not advocating for one business model over another here, to be clear.
I started the analysis with identifying the profitable digital banks from the loss-making ones, based on their 2019 numbers. The criteria I applied is that it must at least have 3 years of operations behind it, is prominent enough i.e. has raised significant amount of funding or has received strong backing from corporates, and operates in a regulatory jurisdiction that is roughly akin to Malaysia. I only identified three digital banks that are profitable: Oaknorth (UK), Kakaobank (Korea) and WeBank (China). Oaknorth is the only digital bank without a BigTech backer, so naturally it drew my attention. What’s striking about it is that it was profitable within 6 months of operations. KakaoBank and WeBank turn profitable a little bit later, aided by the deep pockets of its benefactor. As to the loss-making ones (based on their 2019 audited numbers), I chose Monzo and Starling Bank as a benchmark, although I am aware that there are numerous other digital banks that can be compared against e.g. Revolut, Tandem, Atom Bank. Its just more convenient to compare it against Monzo and Starling due their more prominent positions as established digital banks. As referenced in my previous article about track records, Monzo has expressed “significant doubt” over its ability to continue after doubling its losses during the pandemic. Interestingly since I did the analysis last year, Starling claims it has broke even by Q4 2020 and is currently profitable, after starting its business in 2015.
It goes to say that these digital banks employ very different business models, with very different results. I will explain these business models briefly, but first take a look at how their asset sizes have grown in 4 years of operation.
Admittedly, all these digital banks with the exception of KakaoBank did not have parents with deep pockets to support their growth, so we can caveat KakaoBank’s figures for a moment. The fact remains that even after 4–5 years in operation, they’re still a small part of their own banking system (total assets in the UK banking system as of 2018 is around RM 57.4 trillion). What’s even more interesting to me is the composition of those assets, as shown below:
It is unique that a large chunk of their assets are not loans, but cash deposited with the central bank, particularly for the unprofitable ones. This may be indirectly tied to their business model, but I still consider that there are better ways to deploy their capital than keeping it at the central bank. I can only speculate, but maybe this cash is the dry powder they’re keeping to fund more growth. So while it is still unused, it may be the case that these parties thought they may as well just put it with the central bank to earn some interest in a safe manner. If there are any other explanations please let me know in the comments
Looking more deeply into the business models of these digital banks, I started to observe general trends on how their business model was affecting their performance. I should caveat that painting broad strokes on this relationship is not the most scientific way to approach the question, but I will try to support any assertion I make with publicly available data. To add, I welcome inputs from other parties whether agreeing or respectfully disagreeing with what I am trying to say. Its better for all of us who are interested in this topic to learn as much as we can from each other.
The first dimension I’d like to propose is where digital banks derive their income from i.e. whether they are fee-income focused or interest-income focused. The trend I observed is that digital banks focusing on fee income seems to be in a weaker financial position. Monzo and Starling started out with a stronger emphasis on fee income, derived from interchange fees, premium subscription models, fees obtained by being a platform for other digital banks/fintechs etc. Loans weren’t offered until after Y2/Y3, so to increase their runway, my guess is that they relied on the revenue stream from fees while focusing on competing for deposits and building the customer base. Contrast this with the business model of OakNorth and Kakaobank, which were focused on interest income from loans. Both parties started out with giving out loans, leveraging their advantages by catering to specific segments as will be demonstrated later. Its important to note that Monzo and Starling have been shifting to increase their interest income as well by offering loans, while OakNorth and Kakaobank continues to put a lower emphasis on fee income.
The second dimension I’m proposing, which I think is particularly important for parties without a platform with existing customers, is whether a digital bank is business customer-focused or retail customer-focused. Applicants in Malaysia have to bear in mind that the requirement to mostly serve unserved and underserved segments implicitly implies that in all probability, the margins per customer will be slim. This increases the importance of getting the target segment right from the outset, and from my observation, digital banks focusing on business segments seems to have a sounder footing. My guess is that serving business segments allows for loans with larger ticket sizes and better margins than serving retail segments. Oaknorth, for example, only serves SMEs and business segments, which some in the market argue as too niche for a bank, but they’re the fastest digital bank to turn a profit that I’m aware of. Even Starling Bank’s earlier shift to serve business customers compared to Monzo is credited for their stronger financial position as reflected in this article. Subsequently, this has been alluded to by Starling themselves when explaining their turn to profitability. I need to get into the numbers to substantiate this further, but 2020 numbers aren’t out yet. The trend however, is there.
The diagram below shows where the digital banks I’m benchmarking stands relative to the dimensions I proposed.
The final dimension I’d like to propose that applicants should consider when applying is their plan to scale. The advantage of a digital bank having no branches and hosting its systems in the cloud is well known to all, but that’s just one part of the jigsaw puzzle. If an applicant wishes to go for the volume play, there are other associated operational costs that one have to bear in mind. My take on Monzo and its current predicament is that it scaled too fast. The variable cost it incurred to gain market share, coupled with its reliance on fee incomes at the start, may have increased the burn rate too much. Based on my conversations with some consultant friends and from my own observation of the annual reports, operational costs for digital banks are higher across three key factors: staff costs, marketing costs and technology costs.
Running a digital bank would require an organisation to have a higher proportion of staff with technology expertise, and not only is this talent pool scarce globally and in Malaysia particularly, a digital bank is effectively competing with IT companies. I think that inevitably this will increase the staff cost as compared to a traditional bank, although I lack data to support this. Marketing costs is expected to be exorbitant if an applicant doesn’t have an existing customer base, which is the situation I expect some applicants to face. As for technology costs, I suppose at some point if a digital bank’s revenue flow can’t keep up with the rate of customer base expansion, logically the cost to serve per customer will be very prohibitive. Other parties can comment on this better than me. Just to be clear, all the above can be the inaccurate reasons too, which is why I value input from other parties on this issue.
I’ll end this article with one last diagram to highlight the observations I made. This is based on the latest annual reports I can find and is simplified for the reader’s convenience.
The chart above highlights my points earlier about income and managing the scaling up of operations. The advantages of going digital will still be negated if the business model deployed is wrong. Hopefully applicants will take this into consideration when proposing their business models.
Until next time,
Syahril
