21st Century Lending

The earliest known records of lending date back about 4,000 years to Mesopotamia where the Sumerians were borrowing and lending livestock and seeds that would later be repaid from the offspring and yields harvested from the original “capital”.

Shortly afterwards the Sumerians discovered that it was more convenient to use silver as a medium of exchange, and not too long after that the Code of Hammurabi defined the price of silver and how the interest charged on silver loans was to be regulated. It seems that Hammurabi was quite concerned about usury and preventing abusive payday loan practices. If he were alive today, he may be disappointed (but perhaps not too surprised) to see that the need to regulate overly greedy lenders has not gone away. Four millenniums may have passed, but human nature does not seem to have changed much!

On the other hand, what has changed is the technology available to borrowers and lenders, and the good news is that there is now a new breed of lender emerging that uses this new technology to drive better outcomes for all parties involved.

This new type of lending is characterized by five main features: It is embedded, it operates in real-time, it unlocks and underwrites with sources of data that were previously locked up or inaccessible, the collection mechanism is “default on”, and the use of proceeds are laser targeted.

The net result of these five features is that the losses for this type of lending is orders of magnitude lower, in some cases approaching zero. Coupled with operational cost efficiencies enabled by technology, this in turn enables the lender to charge the borrower rates that are dramatically lower than other lenders, creating a virtuous cycle. Let’s now quickly examine all of these five factors in some more detail.

The first factor, embedded lending, has been written about and discussed a lot in the fintech community, including in my previous blog on the topic. In summary, it refers to the lending itself being invisible to the borrower and moving in harmony with the needs of the customer. An example we used in the past is how mortgages can be embedded into the property purchase to make the whole process last hours not weeks (or months as is the case in the UK). Imagine being able to buy a house with a few clicks!

Secondly, these new lenders operate in real time. In practice this means that they in real time ingest not just the data from the borrower, but also the third-party sources of data about the borrower, and hence can also make and communicate their decisions in real time. To take small business loans as a case study, consider how those were made in the past. The business owner would have to print pages and pages of financial statements that were already out of date at the time of submission, the loan officer would review these, and make the loan some weeks later. The business might have ended up in a totally different state by then, but again the loan officer would only find out once the new set of financial statements were prepared, reviewed by an accountant, and submitted some months later. Now contrast that with having real-time access to not just the customer’s bank account but also their sales data. Quite a difference!

One can only hope for lower interest rates in the future!

Thirdly, yes, the data is in real time, but crucially it also contains new and relevant information that is usually locked up inside the business. This can include anything from recent bank account movements to real time sales data or data on how many shifts an employee has worked that week. This incremental data not only provides better performance from a credit underwriting perspective, but it also ensures way better customer outcomes by making sure calculations like affordability and the potential vulnerability of the borrower can be made with superior precision.

The fourth factor is a collection mechanism that is “default on”. In simple terms this means that in the ordinary course of business, the collection of the loan is automatic and does not require any extra effort. Following the insights of behavioral finance that have emerged over the last decade thanks to Professor Richard Thaler from the University of Chicago (where I had the good fortune to take his class) and his mentors Kahneman and Tversky, this actually makes the loans much easier and painless to collect, resulting in substantial costs savings that can in turn be passed onto the borrower. A good example of such a mechanism is collecting a portion of sales receipts at the point of sale to repay the loan, but there are many other examples.

Finally, the use of proceeds is targeted with laser precision. The loan is not made as a general disbursement for the borrower to spend as they wish, but rather for a specific purpose. Another way to look at this is that the lender understands very well why the borrower needs the funds, and how that fits into a sustainable pattern where the borrower can pay back the loan without falling into distress. This leads to a much more borrower friendly situation, where funds are only drawn down subject to affordability and a sensible use of proceeds.  

In closing, it is also very important that one does not lose sight of the most crucial element of the narrative: Done rightly, 21st century lending enables far superior borrower outcomes where consumers and small businesses can borrow money when it is most needed, at a fraction of the cost, with minimal disruption to their lives, knowing that affordability and sustainability has been embedded into the process.

To Lend or Not To Lend

Over the last decade many fintechs have contemplated this very question, with some choosing to lend, some choosing not to, and yet many others ending up in the undiscovered country from whose bourn no traveler returns as a result of doing lending badly.

When dealing with such existential questions, a simple framework can be of great use, so let’s examine one here, with some real-life examples of each instance where it makes sense or not to embark upon a lending journey.

In the simplest of terms, the economic equation from lending can be described as follows:

Π = NPV(Y) – X  where

Π = Profit from lending

NPV = Net Present Value of future cash flows at the firm’s cost of capital

Y = ( [Interest + Fee Income] – [Interest + Fee Expense] – OpExCost – Losses)

Χ = Customer Acquisition Cost (CAC)

So in simple terms, a firm can expect positive economic profits from lending if it is able to charge its customers interest and fees that exceed all its expenses which include cost of funding, operational expenses, losses from credit and fraud exposure, and customer acquisition costs including all marketing expenses.

Now, it is very important to keep in mind that this equation exists in the context of a very competitive market, and in striving to generate a positive profit from lending, a firm will be competing with others that are trying to do the very same thing. Much like the story in which two campers come across a bear outside their tent, the key to survival is being able to run faster than their fellow camper even if one cannot outrun the bear. 

In other words, to embark upon lending successfully one needs to have a competitive advantage in ideally all (being able to outrun the bear), or at the very least one (being able to outrun the other camper) of the key areas outlined in the equation above.

We can now look at some real-life examples from each of the categories.

Traditional players like banks have a clear competitive advantage when it comes to cost of capital given their vast branch networks where they hoover up deposits at close to zero marginal funding cost. A tech enabled fintech may have an advantage when it comes to operational cost given state of the art infrastructure that automates many manual processes. Players like Capital One used vast databases and superior analytics to keep their losses low relative to their interest income, in other words data mining pockets where the risk reward equation was superior. Yet other players may have found superior customer acquisition tools by virtue of a unique partnerships or value propositions to their customers.

Interestingly, while having a competitive advantage in some of these areas is a prerequisite for not being eaten by the proverbial bear, it does not mean that going down the path of lending is the best decision for that particular firm. Playing in just one aspect of the value chain may in fact provide a higher return on equity than lending which is capital intensive.

It is hard to run faster than a bear

Let’s take the prior example of a fintech company that has the infrastructure that enables it to originate and service loans at a lower operational cost compared to anybody else. It may in fact be more profitable for that company to outsource its origination and servicing technology to other players, thus being an infrastructure provider vs an actual lender itself. QED’s portfolio company Amount is a great example of this very case.

Another example would be a company that has a clear advantage in customer acquisition costs. Rather than using this to become a lender, the company could become a loan originator or broker, and originate loans for lots of other lenders. Again, from CreditKarma to Capitalise, there are many examples in the QED portfolio of companies that have adapted this approach.

So, to summarize, founders that are thinking about taking their company down the path of lending first need to ask themselves if they have a competitive advantage in any of the key areas needed for success. If the answer is affirmative, then the next question is to determine if the company is better off selling its unique skills to the broader market, or becoming a lender itself. The answer to this question is to a great extent determined by how tradable those skills are: licensing tech infrastructure, or a loan originator selling leads to a lender is relatively easy. On the other hand for Capital One to transfer the complex analytical framework it has developed over the years is somewhat harder and likewise it is hard for a big banks to “sell” their cheap cost of funding (though they can monetize their advantage in other ways, by for example becoming a wholesale funder of other smaller lenders).

Finally, let’s look at a few examples where it does not make sense to become a lender. One example that we have often seen is when founders mistake having good data scientists as a sufficient condition for being able to create a good risk reward equation (interest income relative to credit losses). While having good data scientists (or statisticians as we used to call them back at Capital One) is undoubtedly a crucial ingredient, it is by no means enough in and of itself. The models and scores that are produced by the data scientists need to be placed in a wholistic business context, where the business analyst takes the model output and determines where to make the right cut offs and tradeoffs. For example, a data scientist with a model can produce an expected fraud probability score for a credit card transaction, but a business analyst needs to determine the optimal threshold where a transaction is declined or approved taking into account the business impact of false positives as well as false negatives. In other words, credit is a culture, not a model. Building the right organizational structure to get it right is not easy.

Another example that comes to mind are companies trying to engage in what could be described as “multiple arbitrage”. The classic example of this is a fintech start up (for example a mortgage broker) that gets allured by the high revenues that lending can provide, and starts lending, hoping that it can attract a tech multiple on its lending revenues. While this may work with some investors initially, if the company does not have a competitive advantage for lending in at least one of the key areas, the bear will catch up with them eventually.