Lending is easy......

...getting your money back is hard!

It’s been a while since I last wrote something here - so it’s good to be back! So first and foremost - what week! I will NOT pontificate on what happened and why with SVB et al. - plenty of amazing content available out there - check out all of Matt Levine's takes and this one from Marc Rubinstein. As we recover from the events (thanks to the collective effort from many), one thing has become very clear to me. As a lending and credit risk exponent who 'grew up' in the GFC era, I had always considered credit risk to be THE existential threat for Banks (and it probably still is for the most part). However, it is becoming clear to me (and I am sure to many others) that in a rising rate environment, interest rate risk is equally important and foundational. Prudent risk managers need to make sure they have adequate protection/hedge against asset-liability timing mismatch, especially in a volatile rate environment - ok, it's a bit more nuanced than that, but you get the point!

With that, let me segway into a topic that I know a tiny bit more about than interest rate risk :-) As the GM of a newly launched credit card business, the last two years have been incredibly exciting, rewarding, enriching and also humbling! I have been in the eye of the storm, starting and scaling a credit card business as the credit environment started to normalize post-pandemic - and through it all have learnt many lessons that I would like to share with the broader community through a two-part post. 

As you talk to financial services and fintech experts, you will hear a common theme - to build a profitable, sustainable and scalable consumer financial services business, you need lending - period. It’s one of the most attractive businesses boasting high margins, high ROA (often in the 3-8% range) and high ROE (sometimes upwards of 35% depending on the segment). But when dig one level deeper, it becomes obvious that very few fintechs/new entrants have really been able to breakthrough into lending. So why is that? Well, it goes back to the popular adage - ‘Lending is easy but getting your money back is hard’ :-)  In this two-part post, I will dive a bit into the details and draw on my years of experience as a practitioner of both secured and unsecured lending for big banks and fintechs.

So let’s start off with the very fundamentals of lending. At its heart, lending is a simple concept - "You borrow money short and lend out long". You have a source of funds that you pay for - borrow either from the capital markets or other Banks as warehouse loans or as a debt facility or as low cost deposits if you are lucky to be a Bank. You then turn around and lend that capital out at a higher rate. You deduct your cost of funds and you are left with what is called NIM or Net Interest Margin. You then deduct fraud and credit losses, allowances, fixed costs and operations cost and as long as you are net positive, you are making money. Simple right? Well, not quite! 

In part 1 of this analysis, I will dive a bit more into the fundamentals of starting a lending business - the decisions and choices you have to make at the very onset and in part 2, I will go into the details of launching, scaling and running the business. 

Decision 1 - Product -

I have often heard and seen institutions jumping into the lending bandwagon with the most generic products and concepts hoping to take on the incumbents. The first question you have to ask is in a crowded field dominated by big banks with access to seemingly unlimited capital and better data and tools, what incremental value does your product offer? In other words, what is the unique problem or unmet need that your product addresses? From credit cards to personal loans to home loans to auto loans - almost every segment of the lending business has myriads of players offering mostly the standard set of features. So in this crowded market, having another generic personal loan, or a credit card that does not offer anything differentiated/innovative is not going to cut it. That much I think is obvious to most. What is not often obvious though is the concept of positive selection. In lending, you will always have a certain set of people who will NOT pay you back - either because they cannot (ability to pay) or they do not want to (willingness to pay) and there is no 0/1 way to completely isolate them; at least without running afoul of regulatory requirements - more on these concepts in part 2. The key is always to find enough ‘good’’ people who will pay you to offset that. While you will likely use sophisticated targeting and underwriting policies to get those people (again, more on these in part 2), it usually starts with having the right value prop and product offer at the very top to attract the best quality people. Otherwise, even with the best targeting and underwriting models your mix of ‘good to bad’ people and balances will go out of whack leading to higher than expected losses. 

Decision 2 - Identifying your target segment -

Somewhat related to the first, the next major decision is your target segment. As you think about the jobs-to-be-done with your product and value prop, you have to be deliberate about the segment you want to play in. Each segment is unique with its own pros and cons and requires a different strategy and capabilities. For instance you may find that a segment like new to credit which consists of sub-segments like students or recent immigrants is very attractive because it aligns with the innovative product value prop you have developed and is somewhat underserved by incumbents. However, by definition, this segment is much harder to underwrite, tends to have a lot of synthetic fraud and has low unit economics to start with. So if you decide to play in this segment, you will need to invest in things like alternate data sources to underwrite, modern fraud detection capabilities and have a business model to cross-sell to drive up the lifetime value or longer term unit economics. On the other hand, segments like prime+ are attractive because of lower risk, but have much lower margins and longer payback periods because of extremely high competition. 

Decision 3 - Identifying your source of funds

The obvious element of lending is securing the funds that you will lend out. As we lived through an unusual era of zero interest rate over the last decade or so, this question was less critical - everyone had access to cheap capital. However, as days of 'ZIRP’ (Zero Interest Rate Policy) start to fade behind us, any lending business needs to grapple with identifying a source of reasonably priced funds. This is where having a bank charter becomes a game changer. If you have invested the time, effort and money to obtain a bank charter, you are sitting pretty now. In a rising rate environment there is no better, cheaper and more stable source of funds then those deposits (as long as they are sufficiently diversified!). That said, a bank charter comes with a significant level of regulatory scrutiny and requires massive investments in building up the capabilities to meet those demands. For non-banks, it often boils down to a couple of options - securing a warehouse line of credit or a debt facility. Think of both as a loan you take from one of the big institutions and pay interest on at a predetermined rate as long as you maintain certain conditions, especially with your credit quality - one is a revolving line of credit and the other is a fixed loan. It's worth pointing out that SVB was a pioneer in offering low priced debt facility to the fintech lending ecosystem and enabled many a start-ups to get into lending.

Decision 4 - Selecting the right accounting treatment

This is another big decision you have to make before you get started. There are typically two options - held for sale and held for investment and like with most things, there are pros and cons. Held for sale is the model where you originate a loan and then sell it to a buyer as an Asset Based Security (ABS). As the name suggests, the asset is basically the loan receivables you have originated and the price you get depends on the quality of the asset. The biggest advantage with this approach is this allows you to have a much lighter balance sheet as you are not carrying the loans in your books. This also means that you need minimal reserves freeing up capital (more on that later). You also get an immediate boost to your P&L- as you sell the loans the proceeds positively hit the P&L immediately. Even when you sell the loans, you could also choose to be a servicer of the loans and charge a fee for it, thereby adding an additional source of revenue. On the flip side, you have to be on the hamster wheel constantly generating and selling loans. The moment you stop, the P&L dries up. This can happen for many reasons - especially in a tight economic environment or a rising rate environment. The second model is held for investment. In this model, you generate the loans and keep them in your balance sheet. You can choose to service the loans (more common, particularly for Credit Cards) or choose to outsource the servicing to a 3rd party (sometimes used for fixed term, closed-ended loans like mortgages and auto loans if servicing is not in your wheelhouse). This model allows you to have a steady source of income and not be on the perpetual hamster wheel of having to originate loans everyday to stay afloat. However, there is a big issue with this model -  the dreaded J curve! In this model you take many of the costs upfront - the marketing cost, acquisition/set up cost, fraud losses and the dreaded allowance for credit losses. The allowance for credit losses is basically setting aside a certain amount of money depending on the quality of the asset you generate + macro economic factors to cover for future losses. Per the new accounting rules called CECL introduced in 2020, you have to take reserves for the life of loan losses as soon as you make the loan and even before you've generated a single cent in revenue - BRUTAL! All this means you have dug yourself a massive hole before you can see any revenue. However, as the business matures over time (the exact timeline depends on the segment and product and could be upwards of 7-8 years) and reaches a steady state, you’ve built a healthy and seasoned back book - or "the goose that continues to lay the golden eggs!" And that back book is what allows you to generate healthy returns year after year without having to rely on new acquisitions. Most big banks rely on this model but building this ‘healthy and seasoned’ back book takes time, patience and above all discipline (more on that in part 2) and I have seen many banks, fintechs and non-banks falter in their pursuit of the golden goose and run out patience and funding!

Decision 5 - Building the core capabilities -

So this one deserves its own post - so I am not going to delve too deep into this in this post. As you can imagine, the success of any go-to-market strategy is built on a handful of critical capabilities. In lending, one of the first things you need to build is your underwriting model and credit policy. As a new entrant, this is one of the most challenging aspects - how do you build a powerful underwriting model that is able to detect and identify risk, especially around the margins when you do not have any historical data? Now compare that against the incumbents who have tweaked and perfected their models using performance data from multiple vintages representing a variety of micro and macro conditions. One way to do that is to use anonymized historical performance data from the bureaus (you can buy them, or use a sandbox provided by them) for segments that closely mirror your target population. While a viable option, this approach has its gaps and no matter how closely you try to mimic the segments and expand the timeline to capture various macro/micro factors, the data will still be biased. Another newer approach that is starting to emerge is using ML to produce synthetic data that more accurately represents the target segment and a diverse set of macro/micro conditions and building your model using it. I have personally not used this technique so cannot speak to it's efficacy first hand but the idea sounds intriguing. Once you have your model, you will need to align on the policies -  what is your risk tolerance and what are your 'cuts' based on your risk tolerance. This is another challenging exercise for new entrants. Usually established players will use what is called a valuation model (an estimation of future cash flows) to create the policies - usually cutting out the marginal segments with low or negative valuations under various stress scenarios. Unfortunately, as a new entrant, you will likely not have a valuation model out of the gate - so you have to start by marrying your models scores with a heuristic based approach to create custom policies - e.g. "decline anyone who has applied for more than 5 credit cards in the last 30 days". The same approach will be taken for your Fraud models. The next step is to build a set of decision engines that will ingest all these data elements - bureau data, application data, and any other additional data sources depending upon the segment, run the underwriting and fraud model, apply the policy you have created along with the hard cuts, run your regulatory checks like KYC and spit out a decision. And then for certain products like Credit Card (which IMHO, is one of the most complex lending products), you will have to determine the processor and the network (more on those in a separate post). Admittedly, in today's API-driven plug and play environment, it is possible to automate some of these services using workflow based solutions from many of the 'platform as a service' providers. And then you can layer on top of it a typical program manager to run the end-end service for you. While this will definitely help you get to market faster, you will lose some control. Finally, depending on the product, you build the servicing capabilities - for closed ended loans like personal loans, or auto loans, it's a lot simpler. For revolving loans like a credit card, the servicing and portfolio management capability is a lot more complex and requires a lot of iterative investments. This is because for revolving loans like Credit Card, a big chunk of risk management happens post acquisition through steps like credit line management policies, balance build policies, authorization policies, account closure policies etc. This overall aspect is one of the most complicated and critical aspects of building a lending business. I will dedicate a post entirely on this subject in the future - so stay tuned. 

PHEW!!!! Ok, if you have made it thus far, kudos to you! You have done all the hard work to set the stage and the spotlight is shining. By now you have identified the target segment, built a viable product, you have established access to funding - either through a debt facility or a warehouse and you have finalized your accounting model and built the core capabilities to start approving and servicing loans. Now all that is left is to go to market with your killer product and rake in the moolah, right? Well, this is where the real fun starts - the ride is often bumpy and unpredictable. In part 2, I will delve into the challenges of launching, scaling and running a lending business. Sneak Peek - starting day 1, you will get hit with something that rhymes with BROAD! :-)

If you enjoyed reading this post, please consider liking and sharing it and please do not hesitate to leave your thoughts or questions in the comments section below or @ddas1992.