There are a number of different business models at work within the subprime auto lending industry. I find it helpful to categorize these models using a spectrum of loan quantity versus loan quality.

I will start with business models that prioritize quantity, since this model predominates the industry.

Most of the large lenders (e.g. Wells Fargo, Santander, Capital One) are subsidiaries or related parties of large financial institutions. These lenders need to lend enormous amounts of money, so their subprime lending arms prioritize quantity. Wells Fargo, for instance, holds a trillion dollars in deposits. If it wants to earn a return on some of these deposits by lending them to subprime car buyers, it must lend billions of dollars just to make a dent on its income statement.

In short, lenders with a lot of capital behind them need a lending model that can scale large and scale fast.

The way a lender does this is by automating and centralizing as much as possible. Instead of having individual loan officers and regional branches, the large lenders use algorithms to assess loan applications, and a centralized location to process the loans. Their goal is be able to make average-quality loans at the click of a button, with a minimum of information from the dealer and borrower.

Because this is a scattershot approach to lending, the returns are not great. A 1-2% return on assets is probably the best you can expect over the cycle. So to earn a respectable return, the large lenders typically lever their loan portfolios by 10x or more. Going into the crisis, CPSS, one of NICK’s public peers, was levered 20x.

In addition, the lenders that want to grow large but don’t have direct ties to banks with large deposit bases often securitize their auto loans. They do this because lending is seriously capital-intensive: you hand your money over to the borrower at t=0, and then you get it back in small increments over a number of years. This isn’t a problem if you’re Wells Fargo and your capital is practically unlimited. But if you’re not a money center bank, getting your capital back quickly so that you can re-lend it is vital to your growth. By securitizing, you get your capital back in a matter of months rather than a matter of years. The cost to you is the interest you pay to the securitization purchaser, as well as the increased leverage in your capital structure.

If this is your business, then your business is about quantity. You are a loan processor more so than you are loan underwriter.

I am not being critical of the big lenders. In most cases they are doing what they have to do, namely make mediocre loans and lever them to the hilt. If you are Wells Fargo, this is only way $10,000 loans can be worth your time.

But this approach has clear risks, most notably the risk of going bust. If you are making mediocre loans and you’re levered up, you may get in trouble if loan performance sours, or if the securitization market closes and you fall into a funding gap. (Anecdotally, during the GFC most of the subprime auto lender distress was due to funding issues rather than loan performance).

In short, blow-ups happen in this industry. Sometimes a lot of them happen. Executives who have been around for a while can recite the long list of the fallen without blinking. But, just as with banking, an auto lender with a strong underwriting culture and prudent capital structure can thrive despite the frequent industry turbulence.

I think that sufficiently covers the quantity end of the business model spectrum. At the opposite end, you will find NICK.

NICK prioritizes loan quality to the point of zealotry. Wells Fargo can’t carefully pick and choose each $10,000 loan, but Nicholas Financial is small enough that it can. NICK looks for good credits with bad credit scores, such as reliable borrowers who are in temporary financial distress due to medical bills. It also looks for borrowers with combinations of traits (performance of most recent loan, age of vehicle being financed, length of stay at current residence, etc.) that for one reason or another are unacceptable to algorithmic lenders.

NICK uses a branch model that employs real people in real offices close to NICK’s dealer customers, as opposed to a centralized, automated model, which typically has an HQ and no branches. The branch model helps NICK do detailed research into potential borrowers via personal interviews, vehicle inspections, reference checks, and so forth.

Further, NICK does not securitize its loans. It holds all of them on its own balance sheet until maturity. NICK uses leverage, in the form of a credit line with a bank it has known for 20+ years, but the leverage has historically been just 2-3x equity.

As you can see below, there is a pretty dramatic difference between the quantity models (SC, CPSS) and the quality models (CACC, NICK). On average, the volume-focused lenders receive lower interest rates and experience higher loss rates, but make up for the shortfall from an ROE perspective by using three to four times more leverage:

Because NICK’s loans are of such a high quality, NICK’s profit margins dwarf those of its volume competitors. But then, their loan books dwarf NICK’s.

And therein lays the downside to the NICK model. NICK makes better loans, but it can’t make hundreds of thousands of them. In fact, it makes only about 19,000 a year. The company can grow—recent loan book growth has been 5-10% a year—but it can’t grow at high rates. NICK’s model is too labor-intensive and there simply aren’t enough good loans out there.

There is another shortcoming of this model: it makes NICK an unlikely acquisition target. Historically the acquirers in this industry have been big banks looking to build or expand their auto lending platforms. Not surprisingly, they have wanted to buy lending businesses capable of scaling up quickly. NICK is not that business. NICK doesn’t make much sense for a private equity buyer either because the added leverage could make NICK’s capital structure unsafe.

There are, however, a few possible buyers. I will discuss them—as well as NICK’s aborted attempt to sell itself for $16 per share a year ago—at greater length later on. But first I am going to delve deeper into the NICK lending model in particular.