Who’s Afraid of Lending?
If, like me, you come from a commercial background, where you are used to looking at businesses that buy goods at a cost, sell them for a certain price to make a margin that (theoretically, at least) covers all the costs, you likely have one of two responses to lending: either you think it’s significantly easier to run than a commerce business (who doesn’t want a loan!) or you think it is too hard to even contemplate.
If you are someone who falls into either of these two buckets, I would urge you to read on for an understanding of:
- The key structure of a lending business
- How Fintechs capture value
- What VCs look for in lending businesses
An understanding of lending and its cost structure is important for everyone (founders, business heads etc.) especially when embedded finance presents large margin improvement opportunities.
Embedded finance is the opportunity to embed financial services such as loans into an existing business in the B2C or B2B commerce space. Take the example of Grab: Grab partners with lenders to give out loans to drivers who are part of the Grab ecosystem. Grab reports significantly higher supply-side stickiness from giving out these loans.
Key Structure of a Lending Business
In commercial terms, the costs of a lending business are:
(a) the cost of the capital you have borrowed;
(b) the money that does not get paid back to you (i.e., the “non-performing assets” or NPAs); and
(c) the operational costs of the lending business (i.e., cost of acquisition of customers, cost of underwriting, cost of collections etc).
The money you earn from lending is from the interest rate you charge to your borrowers. The profit margin is the difference between this interest rate and the costs involved.
Let’s look at a few real examples. Here are the financial statements of a few listed Non-Banking Financial Companies (NBFCs) in India, reduced to their basic forms:
The table above shows how all of these metrics come together to determine profitability of a lending business. Take Satin Creditcare, which is a micro-finance institution offering lending products to the underserved segment. Given their higher cost of capital and high operations costs, they need to charge a higher rate of interest compared to the other companies. Operations costs tend to be higher in micro-finance institutions on account of the segment they target and the low average ticket sizes of the loans (~USD 500 for Satin Creditcare). Looking at their P&L, arguably, the interest rates they need to charge should be even higher to make up for the high cost-base.
On the other hand, while the outcomes for HDFC Credilla (which gives education loans) and JM Financial (which gives loans against real estate) are quite similar, JM Financial charges a higher interest rate to make up for the poorer NPA profile of its loan book. The higher NPA for JM Financial was a result of the covid pandemic i.e., household borrower defaults arising from the loss of jobs and reduced incomes.
Value Capture by Fintechs
Fintechs that operate in the lending space try and capture value by using technology to:
(a) Scale while keeping operations costs low: This is achieved by digitising the entire process from loan origination to underwriting to collections. An example of this is Sachin Bansal’s Navi Technologies Limited. Having started the personal loans product in 2020, by Dec-2021 Navi’s AUM was 1/5th the size of Satin Creditcare’s AUM (a 30-year old company). Quoting from their IPO filing: “Our customers have experienced downloading the Navi App, completing the entire loan application and receiving the approval and the proceeds of the loan in their bank account in under 5 minutes.” Of course, this fast scale-up has to go hand-in-hand with an NPA % that works commercially.
(b) Open up newer segments: Lower cost to serve helps Fintechs give out loans to newer segments of the population (i.e, the NTC or New-To-Credit customer). Traditional banks and NBFCs may consider some segments unprofitable but technology-led plays could make those same segments attractive. For example, take the case of Grab, which gives out loans to its drivers. While a traditional NBFC might find it too expensive to acquire such small-sized loans, Grab would be able to operate at a much lower cost-base for acquisition, underwriting and servicing since the driver-borrowers are already part of the Grab deliveries ecosystem.
(c) Reduce NPAs: Reduction in NPAs can be achieved either by using significantly more data-sets or using proprietary risk assessment models to underwrite customers better. Quoting Navi’s IPO filing once again: “Our artificial intelligence based underwriting models learn from a rapidly growing training dataset that contained 2.64 million repayment events … with an average of 0.28 million repayment events added every month...” Without technology using such large datasets is virtually impossible.
Typically, these are the value capture levers available to a Fintech. Lower cost of capital is not usually a lever for a Fintech just starting out.
What VCs Look for in Lending Businesses
Based on the discussion above, here are a few things we usually look for in companies looking to start lending and disrupt established NBFCs:
- Differentiated distribution: Similar to any B2C or B2B commerce business, a lending company’s ability to keep operations costs low depends on how cheaply it can acquire customers, how easily it can acquire the data to underwrite them and how sticky the borrowers are. And like in B2C or B2B commerce businesses, in lending as well a differentiated distribution model is a moat. Examples of this are the various Embedded Fintech models that we have invested in across SEA: Ula in Indonesia, Vigo in Vietnam and Growsari in the Philippines. All of these companies have a distribution set up that gives them access to thousands of retailers to provide financial services.
- Ability to collect and cost of collection: It’s critical to look at how a company mitigates collection risk at the lowest cost. This impacts operations cost as well as NPA %. In the example of the B2B FMCG players mentioned above, a sales person visits the store frequently and gives the Fintechs the ability to collect. Other examples are Earned Wage Access (EWA) players, which can collect from the employer at the end of the month (if customer acquisition has happened in a B2B2C route).
- Quality of underwriting: Quality of underwriting is critical to keeping NPA % low. What makes an underwriting model interesting is if it has one of 2 things: (a) proprietary data; or (b) an algorithm/model that is superior because the team has an insight that is unique. To go back once again to the examples quoted above, the buying behaviour of retailers is proprietary to the B2B FMCG companies and becomes data that can be used to underwrite more effectively. Similarly, Grab is able to underwrite its drivers based on the history of earnings these drivers have on the Grab platform (ticket sizes, frequency, ratings etc.). This is an example of proprietary data.
What are some lending models that are interesting according to you. Are you building one that differentiates itself on any of these parameters. Do chime in or reach out at: visa@saisoncapital.com.
