OneMain Holdings (OMF): a misunderstood subprime lender, but waiting for an entry point

Idea introduction

Consider the following question: what would you pay for a company that produces ~30% returns on tangible equity, possesses a durable competitive moat, and is growing ~5-10% (leading to ample reinvestment opportunities)?

With only those facts, you’d probably say ~20-25x earnings. If I clarified that this company is cyclical and moderately levered but comfortably profitable at the bottom of the cycle, you would become more conservative and might suggest something closer to ~15x mid-cycle earnings. If I further clarified this was a subprime lender, most would probably laugh me out of the room. The brave might suggest ~8x earnings.

As you’ve likely guessed – this company is OneMain Holdings, and it currently trades at ~5.5x 2019 earnings. In short, I think this subprime lender is both misunderstood and exceptionally high quality. There is an extremely negative bias towards anything “subprime” following the financial crisis, but it’s exactly these negative biases that can lead to incorrect valuation discounts and opportunities.

With all of that said, I don’t own any shares today. Although OMF is a high quality business, it’s still inevitably cyclical. I rarely anchor investment decisions around a macro view, but acknowledge that most metrics argue that we are late cycle today. I also try to focus on extreme dislocations in security prices, and don’t think we have that with OMF today. But to the extent things turn for the worse, this is exactly the type of security that I want to be ready for.

For example, during the violent market sell-off in December, OMF traded from ~$30 to ~$23 before immediately rebounding ~50% off the lows to $34. This move led me to refresh my research and start working on this post (making the ultimate decision “to pass” a bit harder). But despite not being an active position, I think there’s real value in being prepared for a misunderstood stock that has demonstrated prior illogical price swings.

Longer-term, I believe the presence of Apollo and Varde Partners (40% owners) creates strong optionality on a future liquidity event. If the public market remains unwilling to properly value OMF, the private market should step in. After reading this post, I’m optimistic that readers will agree that an informed private market buyer would love to own this business outright. My goal is to pay a price that attributes little value to this M&A optionality.

Some background

OMF has an interesting background, and it’s worth spending some time on it. Similar to COOP, this colorful story involves Fortress Investment Group.

In 2010, AIG was still struggling and desperately trying to raise liquidity to pay back U.S. TARP bailout funds. AIG aggressively sold non-core business lines, and one subsidiary for sale was American General Finance (“AGF”). AGF was AIG’s subprime lending business, and this unit funded subprime mortgages, retail sales loans, and installment loans to U.S. borrowers. With mounting subprime mortgage losses, AIG had no patience to try and salvage any remaining value. In August 2010, AIG agreed to sell 80% of this subsidiary to Fortress for ~$150mm (well below AIG’s existing mark of $2.4bn).

Fortress had a plan. The firm knew that AGF’s mortgage and retail loans produced sub-par returns, but installment loans generated strong profitability. AGF also had a vast existing branch network and centralized servicing operations – this existing infrastructure was valuable and would scale well. Fortress re-positioned the business to exclusively focus on installment lending, wound down legacy mortgage and retail loans (through natural run-off and divestitures), institutionalized an ABS securitization program, cleaned up the balance sheet, and drove positive operating leverage through growth. The business was truly re-positioned, and Fortress re-branded the company to be called Springleaf. The business IPO’d in 2013 to solid demand, bringing in new capital to pay down additional debt. Fortress was already up >10x on its bet.

Another large financial institution was similarly re-evaluating non-core business lines in 2015 – Citi. Citi’s OneMain subsidiary was actually the largest subprime installment lender in the industry (Springleaf was #2). OneMain was a solid business, however new post-crisis capital regulations made subprime lending less attractive for regulated banks and depositories. Citi was also re-positioning the bank to focus on more affluent customers, so the time finally seemed right to divest this business. The logical tie-up had been rumored for years.

The deal was finally announced in March 2015. Springleaf would acquire the larger OneMain business for $4.25bn in cash, and the pro forma entity would be re-branded to OneMain (going forward, I’ll refer to the pre-merger Citi business as “Legacy OneMain”). By combining the two largest players in the industry, substantial synergies were promised. Citi was a somewhat forced seller for this unique asset, so Springleaf also got it on the cheap (<10x pre-synergy earnings). Springleaf investors were elated, and the stock jumped >30% to over $50/share.

The party was unfortunately short-lived. Although the deal was partially funded with an equity raise, Springleaf was levering up to acquire Legacy OneMain. The company was exiting 2015 with only ~4% tangible equity-to-assets, although management planned to quickly de-lever through organic earnings and capital generation. Initial purchase accounting suppressed GAAP earnings, extending this de-levering a bit. In hindsight, it was a tough time to lever up. In 2016, broader economic worries led investors to indiscriminately sell the equities and bonds of all levered companies and OMF was caught up in this.

This transitory market panic would eventually subside, but separate cracks began to appear in OMF’s initial pro forma EPS guidance. Management alluded to integration issues when converting the back-end servicing systems to one platform. Loan net-charge-offs (“NCOs”) began to climb higher. Management argued that rising NCOs were due to declining employee productivity during this period of botched integration, but the market didn’t care. An “over-levered subprime lender with merger integration issues” is one of the worst investment pitches imaginable. The stock traded below $20 (~65% off the highs), and this dislocation is what first attracted me to the business.

The back-end system issues were eventually resolved, employee productivity gradually improved, NCOs plateaued and started to decline, and the stock recovered somewhat to the mid-$20s. The other wrinkle is that Fortress still owned a majority stake in the business, and its 2010 investment was becoming long-dated. Fortress itself was being acquired by Softbank in 2017, and there were signs that Fortress was aggressively winding down legacy PE funds and investments on this development.

In October 2017, the WSJ leaked that OMF was running a sale process and the stock jumped to the low $30s. I understood that Fortress had run out of time on this highly successful investment, but it was still was odd timing given that value wouldn’t be maximized in a sale until (1) OMF had fully de-levered and (2) it was 100% clear that the integration/credit issues were behind it. A process was nonetheless underway.

The was another key question: who would buy this business? As mentioned, deposit-funded banks weren’t natural owners of this asset – the Basel III risk-weighting framework makes high-loss unsecured loans overly expensive to hold. This left private equity as the likely buyer, however PE funds were constrained in layering in additional leverage to generate attractive financial returns. I wasn’t optimistic on an attractive bid.

Things actually turned worse though. Through a single line buried in its Q3 2017 10-Q, OMF noted that they indeed ran a sale process, but this ended without a transaction. Concurrent with this news, Fortress sold a large block of stock, and it was clear that they were aggressively liquidating their majority stake. Given all of the recent operating issues with the business, none of this was encouraging.

Holders were forced to suffer a couple block sales from Fortress, but were rewarded when an investor group led by Apollo and Varde Partners acquired FIG’s remaining 40.5% stake in January 2018. The vote of confidence by two well-regarded investors again validated the business model, and sent shares back above $30.

In July 2018, it was also announced that Jay Levine was retiring as CEO and would be replaced by Doug Shulman. I thought Jay did a phenomenal job leading AIG’s distressed subsidiary through its restructuring, repositioning, IPO, and acquisition of Legacy OneMain. With most key initiatives complete (and Jay very wealthy), the timing made sense. Jay has stayed on as Chairman and still owns a ton of stock. Doug comes with experience from BNY Mellon, McKinsey, and most notably the head of the IRS. Doug was credited for modernizing many of the IRS’s technology systems, and my checks on him have been largely positive. Given the increasingly critical role of technology solutions in lending, Doug seems like a great fit. The next chapter of OMF will be far less transformative and more focused on optimization and late cycle risk management.

NCO rates continued the downward trajectory and OMF exited 2018 with 11% TCE and 6.9x leverage (debt-to-tangible equity), the upper end of its 5-7x target. The business has been performing well, but OMF nevertheless got caught up in the violent market sell-off at the end of 2018. The stock has since mostly recovered.

Q1 2019 earnings were coupled with a milestone for OMF – the initiation of a $0.25 quarterly dividend – its first ever return of capital. Uncertainty around pending accounting regulations is somewhat constraining capital return in the near-term, but this is an impressive turnaround relative to earlier years.

Industry overview

The unfortunate reality is that many U.S. households are financially unprepared for an emergency. According to Bankrate, ~24% of households have fewer than 3 months of income saved, and another ~25% have no material savings. Unexpected life events happen to everyone, and many events bring unanticipated bills. For those households lacking liquid savings, there may be few options.

Another issue is around credit. The U.S. underwriting system has evolved to rely on a consumer’s FICO credit score (provided by one of the three major credit bureaus). This backward-looking score analyzes the borrower’s historical credit history to determine one’s future likelihood of default. Although it’s far from perfect, this system has dramatically reduced aggregate underwriting costs, and in theory these savings are passed through to borrowers.

The problem is for those that lack a good credit score, as the ingrained system works against these households. Of the ~250mm Americans with a credit score, ~50mm are deep subprime (<550 FICO), ~100mm are “near-prime” (550-700 FICO), and only 100mm are prime/super-prime (700-850 FICO).

OMF targets that middle bucket of higher quality subprime and near-prime customers – a large estimated TAM. These are certainly credit-challenged customers, although they aren’t necessarily shut out from traditional lending markets like deep subprime individuals.

Let’s assume that an unexpected bill indeed comes up. The best option would be to cover this bill from savings, but many subprime customers lack material savings. The next best option would be tapping equity in one’s home. A home equity loan or HELOC can be a great low-cost method to generate liquidity, but not all subprime customers own a house or have sufficient equity to utilize this option. For most borrowers lacking home-ownership, the only material asset they own is their car, but this depreciating asset is often already burdened with an auto loan. After that, the options are less attractive. For most, the chosen option is a credit card. Less frequently, the chosen option is an installment loan. For deep subprime customers, options may be limited to extremely onerous payday loans and title loans.

One recent trend has been the rise of “fintech lenders” that lack physical branches (Lending Club, SoFi, Prosper, Goldman’s Marcus, Avant, etc.). These fintech lenders often don’t directly extend credit, but instead match potential borrowers with interested investors (generally investing through ABS). These lenders most commonly target tech-savvy millennial borrowers for student loans and installment loans. Volume growth has been rapid in recent years, however defaults have come in much higher than expected at several peers, leading many to pull-back from subprime and non-prime customers.

One should always be suspicious of any lender that doesn’t retain its underlying credit risk, and it’s not surprising that loan performance has been worse than expected at many of these start-up companies. The approval process is truly easy and streamlined for borrowers, however recent studies have suggested that this no-sweat online process is actually a material contributor to adverse selection. Separately, it will be interesting to see if 3rd party investors (often hedge funds) are just as excited to purchase ABS and provide funding in a deteriorating economy.

These fintech lenders certainly brush up against OMF today, but the interaction is less frequent as fintech lenders move up the credit curve. OMF also receives competition from banks and credit unions, although their focus will always be on higher quality credits, established customers, and larger loans. Similar to Fortress, other private equity firms have built stakes in smaller non-bank installment lenders – Mariner is owned by Warburg Pincus and Lendmark Financial Services by Blackstone. Some payday lenders have re-focused on installment lending following threats of onerous regulations in recent years, however these installment loans are generally lower credit quality, shorter-term, higher APR, and closer to payday loans than OMF’s typical installment loan.

Experian estimates that ~18mm borrowers in the U.S. currently have an unsecured installment loan and there is ~$130bn outstanding principal (~$7,000 average loan balance). These figures exclude secured installment loans – a key pillar of OMF’s business – although secured products are broadly less frequent. Despite strong recent growth, this product remains very small vs. other consumer borrowing products (mortgages, HELOC, auto, card, student loans), and it’s unlikely this will change.

As mentioned, OMF’s installment lending model operates in a bit of a niche – above the deep subprime payday/title lenders, but below the prime lending of national banks. This characterization understates the uniqueness of the business model and approach at OMF, however, and it takes a bit more work to fully grasp this. We’ll get there…

The customers & products

First, let’s talk about OMF’s typical customer. The average FICO score is 630, so these customers have certainly had some issues with credit in the past. On traditional underwriting metrics, these borrowers are higher risk. If you were forced to lend to these customers, what mitigating characteristics would you require? To start, I’d probably look for some evidence of stability – decent income levels, some continuity in work and residence, the responsible utilization of other financial products, etc.

And when we look at the data, we do see this. The average borrower has lived in their residence for 12 years, ~50% are homeowners, ~90% have a checking account, ~75% have a credit card, and ~50% have an auto loan. The borrower’s average income is ~$45k (vs. ~$60k median income in the U.S.) and 60% have had the same job for 5+ years. My first reaction to this data: it’s not that bad!

Why do customers need a loan? In Q4 2018, the breakdown was:

  • Debt consolidation: 36%
  • Household bills: 23%
  • Auto-related: 10%
  • Home repair: 8%
  • Family-related: 7%
  • Other: 15%

For those borrowers pursuing debt consolidation, it’s generally to consolidate multiple higher cost credit card and other debts into one monthly payment. And remember – an installment loan is a fixed monthly payment that includes principal and fully amortizes to $0 over the life of the loan. By converting a revolving debt balance into an installment loan, these borrowers are responsibly forcing themselves to budget for principal pay-down each month.

It’s similar to buying a house. For many families, their net worth is concentrated in their home. But if you actually do the math, you’d conclude that the long-term returns of investing in a home isn’t compelling vs. renting and investing your down payment in the stock market. It’s still a good financial decision for most Americans, however, because your monthly mortgage payment forces you to pay down principal and “invest” additional equity in your home each month. The unfortunate reality is that most Americans wouldn’t adequately save or pay down debt if not forced to.

Next, let’s talk about OMF’s three major loan products. All three products are installment loans, so they are fixed-rate, fully-amortizing loans. The average tenor is also consistent across loan products at ~4.5 years. The key product differences are around collateral:

  • Unsecured loan: personal loan without collateral
  • Hard Secured: loan collateralized by a 1st lien on a 10+ year old vehicle
  • Direct Auto loan: loan collateralized by a 1st lien on a <10 year old vehicle

Collateral matters for a few reasons. A lender’s overall loss rate will naturally be driven by: default frequency x loss severity. If a loan is unsecured, the severity is generally 100%. If a loan is secured, there exists the potential to repossess and liquidate the underlying collateral and reduce the severity rate. The realized severity rate will then be driven by the loan’s loan-to-value (LTV) ratio and the relative cost of repossessing and liquidating the collateral.

Lower severity rates makes secured loans less risky, but that’s not the only benefit. First, think about the mindset of an unsecured borrower. Most borrowers genuinely try to repay their loan, but some inevitably suffer financial hardship. When faced with the prospect of defaulting on their unsecured loan, they are weighing the reality of their difficult situation vs. the consequence of ruining their credit score. Most would drop <600 FICO, putting traditional financial products out of reach. This isn’t great, but it may not be significantly worse than their current financial reality.

Next, consider a secured borrower. When faced with the prospect of defaulting on a secured loan, they are weighing the reality of their situation vs. the consequences of ruining their credit score and losing their car. For many of these borrowers, their car is their most valuable asset and the key in getting to/from work. The stakes are much higher, and this impacts borrower behavior. As a result, OMF’s secured loans exhibit lower default frequency rates. Lower frequency rates, in conjunction with lower severity rates, makes OMF’s secured product a much lower loss product. And based on OMF’s historical experience, this loss mitigation benefit is greater when the collateral is a newer car – borrowers are more likely to walk away from a clunker. This makes the newer Direct Auto product (launched in 2014) OMF’s lowest loss product.

Through-the-cycle loan performance

Now that we understand the basic products, we can analyze some data and see how loans have performed over a full economic cycle. For most lenders, we could simply pull 10-Ks to see long-term performance. The OMF we know today, however, was two separate non-public subsidiaries during the financial crisis. This complicates things, but thankfully OMF does provide pro forma loan performance data as part of its ABS disclosures. If we look at page 48 of the 2019 Vegas ABS investor presentation, we can see the combined portfolio loan performance. This data effectively assumes that Springleaf and Legacy OneMain merged in 2004:

I’ve provided a summary analysis below. One complication is the fact that the Direct Auto product hasn’t existed through a full economic cycle. To deal with this issue, I’m simply assuming that Direct Auto cycle loss rates settle at ~50% of the Hard Secured product’s loss rates (consistent with current relative performance).

There are a few important observations to make. As suggested, Direct Auto loans currently produce the lowest loss rates, followed by the Hard Secured product. There isn’t a large difference in pricing (APR) between the Hard Secured and Unsecured products, however much of the Direct Auto’s credit “benefit” (lower loss rates) are passed through to borrowers via lower pricing. The lower APR and higher collateral values of the Direct Auto product in-turn allow for higher average loan balances.

This analysis suggests that – if Springleaf and legacy OneMain merged in 2004 – the combined pro forma portfolio would’ve produced 7.5% average cycle loss rates and 11.8% peak loss rates in 2009. This is pretty solid given the typical customer mix and realized APRs, but it would also ignore a few important nuances. To start, the current product mix is much higher quality than the average historical product mix. This should mean that the current portfolio would produce lower losses in a comparable recession today. To appreciate other less obvious nuances, we need to go back to the 2016 Vegas ABS investor presentation. At this time, management actually bifurcated the through-the-cycle performance data into (1) Legacy Springleaf branch performance and (2) Legacy OneMain branch performance:

The following table summarizes the relative cycle performance of Legacy Springleaf vs. Legacy OneMain branches. I’m ignoring the Direct Auto product, because this wasn’t really introduced in Legacy OneMain branches until 2016.

We can now see that Legacy Springleaf branches have produced much lower loss rates than Legacy OneMain branches. More impressively, Springleaf achieved better loss rates despite focusing on lower FICO customers! I was very confused when I first discovered this.

To better understand this, let’s focus on the unsecured product. Legacy Springleaf branches have averaged 6.9% cycle losses, well below Legacy OneMain’s 8.4% cycle loss rate. Even more odd, however, is the fact that Legacy Springleaf branches achieved a lower loss rate despite having consistently higher delinquency rates – 4.6% at Legacy Springleaf vs. 3.8% at Legacy OneMain! This means that customers were more frequently falling behind on their payments at Springleaf branches, but Springleaf was significantly more effective at returning those customers to “current” (non-delinquent) status.

Not all delinquent customers default on their loans. When a customer falls behind, the lender should reach out, gather information, walk the borrower through their options, explain the implications of default, and hopefully convince the borrower to make a late payment and become current again. These activities are collectively referred to as “delinquency servicing,” and some lenders are simply better at it than others. If a lender is relatively good at delinquency servicing, then it should have a lower level of NCOs relative to delinquencies (and vice versa). A lower ratio means that the lender is relatively more successful at “curing” delinquencies when they inevitably occur. Put differently, although the volume of delinquent customers is driven by initial underwriting and economic factors, the conversion rate of delinquent customers to ultimate defaults is driven by employee blocking & tackling.

The ratio of NCOs to delinquencies was 1.5x for Springleaf’s unsecured product, but much higher for Legacy OneMain branches at 2.2x. This means that Springleaf branches were consistently better at delinquency servicing. Why is that the case? I’ll explain this more later, but I ultimately believe that Springleaf operated with a more energized culture, superior late-stage delinquency management capabilities, and better designed compensation incentives. Remember that Springleaf was on the verge of insolvency and run by a heavily motivated private equity owner, while OneMain was neglected and ignored within a massive bank – the difference in performance shouldn’t be surprising.

We can now revisit our through-the-cycle loss analysis. As you can imagine, we have a strong focus in understanding just how bad losses could be in a future recession, and this may not necessarily align with performance in prior recessions. If we take our combined loan portfolios and re-weight historical product performance for OMF’s current loan mix, then the combined portfolio losses would peak at 10.0% (better than our prior 11.8% peak loss estimate). Going one step further – if you assume that management can extend Springleaf’s delinquency servicing performance to all Legacy OneMain branches – such that all branches perform in-line with Springleaf’s superior branch performance – losses would instead peak at 7.8%! Remember – Legacy OneMain’s higher loss rates were in no way attributable to customer quality, so this gap should mostly be fixable.

Let’s put some perspective around this. Private label credit card loss rates peaked at 11.3% during the financial crisis. Subprime auto loss rates (a fully secured product) peaked at 9.0% during the financial crisis. I believe that OMF, despite having ~50% of its portfolio in unsecured loans, could produce 7.8-10.0% peak loss rates in a comparable recession today. It gets even better when you compare to fintech lenders. The Kroll Bond Rating Agency tracks ABS performance for online installment lenders and creates indices for credit performance. OMF’s customers probably fit somewhere between the “Tier 2” KBRA index (Upstart and Lending Club Prime) and “Tier 3” KBRA index (Avant and Lending Club non-prime). Amazingly, the Tier 2 Index is currently running at 12% losses, and the Tier 3 Index at 22% losses! However you measure it, OMF’s relative underwriting is impressive.

A fair question is: why is OMF’s underwriting performance so strong? Why are loss rates so low for a subprime portfolio that is ~50% unsecured? To help answer this question, I visited a local OneMain branch to learn more about its underwriting process and incentive structures. I was able to spend multiple hours with entry level employees, the branch manager, and the regional manager. I’ve also read dozens of online customer reviews. The takeaways were very interesting.

Unique underwriting model

Before getting to underwriting, let’s talk about customer acquisition (which can have an effect on customer quality if done poorly). OMF primarily markets its loan products to customers through direct-mail, credit aggregators, SEO, and other online channels. Some of these potential customers could benefit from debt consolidation or have a liquidity need and become interested. At that point, 80% of customers apply online and the rest over the phone or at a branch. The initial application process is relatively straightforward and you can check it out here:

After inputting desired loan characteristics, personal information, employment, and income data, OMF’s centralized underwriting system checks attributes from traditional and alternative data sources to provide a conditional approval decision. The final approval requires the borrower to provide documentation that verifies employment, income, and key expenses. At this point, the vast majority of customers finish the application process at one of OMF’s branches, although tech-savvy customers have the option to digitally upload documents and finish the approval process remotely.

Following the merger and post-merger branch consolidations and divestitures, OMF has an impressive network of 1,600 branches. ~85% of Americans live within 25 miles of an OMF branch, and this is actually the largest network breadth in the industry (Wells Fargo is #2 at 76%). In today’s digital world – where new fintech players operate entirely digital platforms – it’s natural to question whether this branch network is efficient or even needed. OMF really operates a hybrid model where applicants have a choice. But it’s worth reiterating that most applications still prefer to visit a physical branch, so that should speak for itself. In addition, there are business benefits to closing in-person, which we’ll get to.

How does OMF really make loan approval and sizing decisions? Before we answer that question, let’s review how other lenders traditionally manage underwriting. For prime credit cards, the lender will predominantly rely on your FICO score. They will also ask for your employment information and salary, although many lenders won’t verify employment or income information for every borrower. Your FICO score will generally drive the approval/denial decision, with your (potentially unverified) income level driving your credit limit. This simplified process keeps underwriting costs low, and for low-risk prime customers, it’s generally fine.

Let’s think about the flaws in this model. To start, your FICO score is backward looking at your historical credit payment tendencies. This is generally a good predictor of future tendencies, but it ignores situations where major life events have impacted one’s go-forward default likelihood. And although tying credit limits to income makes sense, it ignores how much residual income is actually available to service debts. For one household, a $50k annual salary and modest lifestyle provides ample residual income. For another household with multiple kids and several cars, $50k isn’t enough to cover the bills. These details truly matter for riskier customers, and the FICO-based underwriting model starts to break down.

The other problem with credit cards is that lenders don’t fully control the timing of drawdowns. There are many customers that have perfectly fine credit, apply for a credit card, get approved, and never revolve an outstanding monthly balance. They are great low-risk customer until one day they lose their job and fully draw down their credit card to pay bills. There’s an inherent conflict where credit card lenders are frequently forced to “lend” to customers at the exact moment that customer default risk is highest. In contrast, installment lenders control the initial underwriting decision, and amortizing loans force borrowers to gradually repay principal during the good times.

On the other end of the spectrum, let’s look at payday lenders. These loans are generally for $500 or less, reserved for the riskiest customers with FICO scores <600, and extremely short-term in nature. These loans are for borrowers that have jobs but need liquidity to cover an expense that is due before their next paycheck. These lenders are then repaid on your payday (hence the name). Loan amounts are extremely small, which means that paperwork and processing costs are high relative to loan balances. Most states have some form of cap on loan interest rates through usury laws, but there are multiple loopholes for lenders to charge effective annual interest rates (inclusive of fees to cover processing costs) that are 100%-500%. Payday loans deservedly get a bad rap because many borrowers are unable to pay back the loan with their next paycheck. Instead, borrowers are forced to pay back their old loan with a new payday loan and incur a new wave of processing fees, perpetuating this high cost indebtedness or spiraling out of control.

How do payday lenders underwrite loans? The reality is “not at all.” A large number of these borrowers will inevitably default. One way to cover these credit losses is through the extremely high effective APRs that lenders charge. The other mechanism is through ACH transfers. When you receive a payday loan, you are generally forced to give the lender ACH transfer access to your checking account. Instead of tracking you down for payment, the lender automatically debits your account for the owed balance (whether you have it or not). This can lead to overdraft penalties and other bank fees. Put differently, payday lenders effectively manage credit losses through aggressive recollection policies (as opposed to underwriting). The borrower is rarely better off.

Our final example will be a traditional auto loan. Although auto loans are provided by banks and other financial institutions, they are generally processed and closed at the car dealership. Banks begin by approaching large and small dealer groups and express a desire to offer financing solutions for their customers. These banks will offer underwriting guidelines that they are comfortable with (FICO score, vehicle LTV, etc.). If the underwriting guidelines are too strict, the bank might be turned away or fail to capture meaningful volume. Right away, there is a natural competition and tension for lenders to modestly push the credit box and drive enough volume to cover the fixed costs of running an auto program.

In addition, most dealers negotiate some degree of flexibility to adjust certain approval terms and limits. Dealers may accept certain recourse for defaults on nonconforming loans, but generally only for a small initial window. Let’s also think about customer intent. Many customers go to a dealership with the full intention of utilizing dealer financing, but not all. Some customers are cautious of borrowing and initially plan to partially or entirely pay in cash.

Let’s think about a dealership’s incentives. In addition to earning a gross profit on the vehicle sale, they generate commissions and revenue whenever they close a loan, attach an extended warranty, or sell a lifetime oil-change contract. As a result, the compensation structure is highly variable, mostly driven by vehicle volume, and inclusive of additional compensation when financial and ancillary products are sold. Financing is an enabler, because it facilitates sales that could otherwise not occur, allows customers to purchase higher price (and higher margin) vehicles, and allows dealerships to roll ancillary product costs into the up-front loan amount (frequently at >100% LTVs). It should always be assumed that dealerships are trying to push the credit limits as far as possible. These loans aren’t “bought” by customers, they are “sold” by dealership employees. Lenders will prudently design underwriting controls to manage credit risk, but this natural tension is impossible to fully overcome.

Let’s now return to OMF. After verifying your income and employment, the focus turns to verifying and estimating your household’s monthly expenses. OMF will verify your address, any rent or mortgage payment, auto loan or lease payments, required payments on outstanding credit card balances, utility costs, cell phone bills, child care costs, and other estimated monthly expenditures. OMF will deduct all of these major expenses from your after-tax income to estimate your truly disposable income. If your requested loan is for debt consolidation, OMF will adjust your household “income statement” for paying down other debt balances. Based on your perceived risk and level of residual disposable income, OMF will underwrite to a conservative debt service coverage ratio (“DSCR”). Based on a responsible and serviceable monthly payment amount and approved APR, OMF will then back-into a maximum approved loan limit.

For example, let’s assume that a 650 FICO individual incurs an unexpected and uncovered $5,000 medical bill. Their credit score and other metrics suggest that a 29% APR is appropriate for an unsecured loan. Let’s assume that the loan’s tenor is 48 months and OMF’s residual income analysis determines that they can responsibly cover an additional $200 monthly payment. At a 29% APR and 48 month duration, a $200 maximum monthly payment implies that the borrower could responsibly borrow up to $5,645. If they have the proper paperwork to validate key underwriting assumptions, they would qualify for the $5,000 loan request.

The vast majority of borrowers enter the branch with an expectation to obtain an unsecured loan. With that said, the loan officer will also inquire whether the borrower has a vehicle that they’d be willing to utilize in a secured loan. Let’s assume that this borrower indeed has a mostly unencumbered and newer vehicle. This borrower also has separate outstanding credit card balance at a high APR of 28%. Based on this, the loan officer might suggest taking out a $13,000 Direct Auto loan (secured by their vehicle), paying the $5,000 medical bill, and using the remaining $8,000 to pay off the credit card balance. At a 28% existing credit card APR, this $8,000 repayment would save ~$190/month in credit card interest.

The presence of collateral would improve the expected credit performance of OMF’s loan, so OMF would be willing to provide a Direct Auto loan at an APR closer to 19%. A 48 month $13,000 loan at a 19% APR would have an all-in monthly payment of ~$390. After netting the $190 of credit card interest savings, the borrower is similarly paying an incremental debt service payment of $200/month (an amount that OMF has already determined they can handle). This loan is just as responsible on an ability-to-pay basis, however the borrower was also able to prudently swap high-cost, variable rate, revolving credit card debt with a fixed rate, fully-amortizing loan. This will force the borrower to budget principal pay-down each month, a positive for both sides. And that medical bill is also taken care of.

This is definitely a win-win. The borrower dealt with their liquidity need and swapped riskier revolving credit for more responsible amortizing credit. OMF won by issuing a larger loan ($13k vs. $5k). This is important because OMF incurs roughly the same underwriting costs when issuing a $5k, $10k, or $15k loan. A larger loan spreads operating costs over a larger loan balance, making it more profitable. It’s worth noting that OMF wouldn’t give you a larger loan simply for the sake of it – borrowers need to demonstrate that larger loan proceeds will indeed be utilized for debt consolidation or other purposes.

At this point, the borrower has the option of choosing either the smaller and higher cost unsecured loan, or the larger and lower cost secured debt consolidation loan. OMF generally prefers the larger loan, but simply underwrites each product according to the residual income model, presents the options, and lets the customer decide. Some borrowers prefer to not re-encumber their car or simply lack a vehicle that could serve as any material collateral. As cars age past 10 years (out of the Direct Auto product and into the Hard Secured product), the declining vehicle value also constrains the incremental borrowing capacity and benefit of a secured loan. With recent origination volumes at ~50/50 secured/unsecured, customers are finding value in both options.

Remember – the key to the secured math working is the fact that the presence of collateral and threat of losing one’s vehicle alters borrower behavior – they are simply less likely to default. Direct Auto NCO rates are 2%, well below the 9% rates of unsecured loans. And because credit and operating costs are lower for these secured loans, OMF is then able to offer lower APRs to customers. It’s these lower APRs that facilitate higher approved loan limits and improves the customer economics of consolidating other higher cost credit card debt. It all works nicely.

To the extent the borrower elects the secured option, the OMF employee will verify the title and walk outside to inspect the vehicle. This brief inspection is an important step to secured underwriting, and a key reason why the fintech model doesn’t work well with secured loan products. The final approval process is entirely paperless on a touchscreen monitor, keeping the process efficient. All verification documents are scanned and digitally stored. Borrowers are encouraged to digitally upload documents before visiting a branch, further streamlining the process. In either instance, a borrower could fail to qualify for any loan if they don’t provide sufficient documentation to verify key underwriting criteria or if their disposable income is too low. Most applications are turned away for one of these reasons.

Hopefully it’s becoming clear why OMF’s underwriting model has produced surprisingly low credit NCOs over the cycle. Relative to prime lending and subprime auto lending, OMF is actually doing real underwriting work – analyzing a household’s ability to service its debts – and only lending what a household can afford. This simple distinction is key. And through directly offering secured auto loans – as opposed to the “indirect” lending model at car dealerships – OMF is avoiding all of the agency conflicts and tensions that naturally push the credit box in dealership purchase transactions.

OMF isn’t the only one performing ability-to-pay analysis though – most installment lenders (in this small and niche product) perform one form or another. And many credit card companies are doing quasi ability-to-pay analysis by using data to estimate the conversion of household salaries to residual income. But few lenders are fully verifying key model inputs and fully considering the same breadth of competing expenses. It’s also worth noting that OMF’s Direct Auto product is relatively unique within installment lending. This business was built by former Wells Fargo Auto executives and not every firm has collateral underwriting capabilities. By having the ability to offer customers multiple loan options, there’s a much better chance of OMF finding the best product for the customer. The branch network is another key edge, because OMF’s customers have repeatedly expressed that in-person closing is preferred.

I learned a lot about OMF’s underwriting model through my branch visit and employee conversations, but I also learned a lot about OMF’s incentive structure, which I believe is equally important. At most traditional lenders, there is a clear bifurcation of responsibilities. Loan officers generally focus on pursuing leads, closing applications, and driving origination volume growth (within existing underwriting constraints). Once a loan is made, however, a loan officer’s responsibilities generally end and they move on to the next lead. To the extent a borrower becomes delinquent, delinquency servicing is generally handled by a separate dedicated team of employees. By clearly bifurcating these responsibilities, employees can specialize in specific skillsets and become more efficient.

This is perfectly fine, but OMF takes a different approach. Branch employees are responsible for both loan origination growth and initial delinquency servicing. This makes OMF’s entry-level branch job a bit more complicated to train and learn, but there are certainly benefits. The biggest problem with loan officers at traditional lenders is the fact that they are entirely incentivized by volume growth. Lenders have underwriting constraints to manage credit, but this doesn’t remove the ongoing volume vs. quality tension. Loan officers can start to act like salesmen, and there are always negative consequences when a loan is sold instead of bought.

At OMF, conversely, loan officers are responsible for servicing any delinquencies arising from their own personally originated loan pool. Similar to most lenders, OMF loan officers receive monthly variable bonuses for exceeding origination volume targets, however these bonus payments are separately constrained by delinquency targets. If an employee’s previously originated loan pool is producing a delinquency rate that is above key thresholds, then their volume-based bonus could be reduced or even eliminated. This compensation structure eliminates traditional conflicts and tensions while creating strong incentives to manage delinquency levels when they do flare up. Humans are inevitably incentivized by money, and OMF has structured its compensation structure to prioritize delinquency management. Branch Managers similarly receive quarterly bonuses that are driven by volume and delinquency targets, so Branch Managers will push employees if branch-level delinquencies start to pick up.

Remember that OMF exclusively focuses on installment lending and only offers 3 loan products. Compared to most banks, this is an extremely narrow focus. Bank loan officers are required to be knowledgeable about a wider breadth of product offerings, so it would be impractical for them to also handle delinquency servicing. OMF’s employee structure is therefore only practical in its narrow focus and lending model.

All things considered, roughly ~75% of a branch employee’s time is spent on new originations and ~25% is spent on delinquency management. The first thing employees do in the morning is follow-up with anyone that filled out an online application the night before. Prospective borrowers generally apply to multiple lenders, so OMF wants to quickly capture the best prospects. If employees are less busy, they will check in with prior customers and call prospects from lead aggregators. Existing customers frequently become repeat-clients and generally exhibit superior credit performance when they borrow again, so this opportunity set is attractive. When customers come in to finish a loan application, employees will pivot and focus on them.

Branch targets are measured against 30-89 day delinquency levels, so employees focus a lot of effort on borrowers that have recently fallen behind on their first payment. If they can cure these initial delinquencies before it hits the 30 day mark, then the loan won’t slip into the measured delinquency pool. The longer a loan is delinquent, the less likely it will again become current, so the natural sense of urgency here is beneficial.

What do employees do when a customer falls behind? Employees are taught to have a personal touch. Instead of opening up by demanding repayment, employees will ask what happened, figure out of they lost their job, explain borrower assistance options, explain the potential implications to their credit score, and offer to discuss further in person. Remember that most employees finish applications in-person and employees are responsible for servicing their own delinquencies. This means that delinquent customers are receiving calls from an employee that they’ve previously met face-to-face. This adds unique familiarity and comfort to what could be a depressing phone call. Employees told me that they believe this approach is a significant contributor to OMF’s strong delinquency servicing performance.

Once a delinquency drifts to 60 days, the servicing responsibilities shift to OMF’s central servicing operations. OMF discovered that although branch-level outreach improved outcomes for 0-59 day delinquencies, it didn’t materially improve late-stage delinquency outcomes. After a certain period of time, it makes sense for branch employees to refocus their time on new originations and other early delinquencies. Late stage delinquencies also involve certain unique skills like dealing with cease & desist orders, personal bankruptcies, litigation, and collateral re-possessions. Instead of training all of OMF’s branch employees for these specialized nuances, OMF’s centralized servicing operations can specialize in these complexities. Once a delinquency reaches 60 days, it’s handed over to the centralized team.

A typical branch might have ~3-4 loan officers and 1 branch manager. OMF’s 1,600 branch managers are the engine behind this machine, and new openings are frequently filled by experienced loan officers. The Branch Manager I met was highly impressive and she took great pride in her branch performance. I believe that Branch Managers have the opportunity to do relatively well financially if their branch performs well, and this is evident by the average Branch Manager tenure of 13 years. Branch managers are generally focused on driving branch volume & delinquency targets, ensuring that the Direct Auto product is properly marketed and sold, dealing with initial customer complaints, and managing hiring needs.

A Regional Manager network exists above the Branch Managers. I met with a Regional Manager that oversaw 7 different districts. His primary focus was on sharing best operating practices, helping to manage hiring decisions, and deal with regulatory compliance. To the extent a Branch Manager left one of his branches, he would generally drop most of what he was doing and focus on filling the gap. There was a real sense that retaining Branch Manager talent was critical. I asked whether the tight labor market was hurting turnover, but it hadn’t at that time. Regional Manager compensation is also impacted by branch-level returns on capital. I found this impressive, because it created an owner-oriented mindset at a relatively low level of the organization.

I’ve read a couple hundred Glassdoor employee reviews and will give a couple observations. OMF’s 3 star rating isn’t terribly impressive, but it’s only ~0.5 stars lower than most other financial services firms. Most complaints were around the competitive and demanding environment. Employees repeatedly mentioned the pressure-filled end of the month when branches would focus on meeting delinquency targets. In many instances, delinquency servicing would result in late nights and weekends at the office. Although this may not be ideal from an employee satisfaction point of view, it is exactly the type of thing you’d want to hear as an owner in this business. Management has demonstrated an interest in improving the culture and replies to most complaints.

If you read reviews immediately following the merger, it was also clear that there was an initial culture clash between the Legacy Springleaf and Legacy OneMain branches. It seemed like Springleaf’s competitive culture was more intense than Legacy OneMain, and this resulted in an uptick in employee turnover (exacerbating integration issues). None of this should be too surprising given Springleaf’s prior private equity ownership and tenuous financial position. Legacy OneMain was neglected within Citi, and its culture probably became a bit sleepy. The constant outreach and calling isn’t for everyone, but many positive reviews importantly acknowledged that top performers can make solid all-in compensation levels. After reading these reviews, the gap in observed delinquency servicing performance between Legacy Springleaf and Legacy OneMain branches makes sense. I believe that management has mostly extended this higher-energy culture and incentive structure to Legacy OneMain branches post-merger, and we should see better combined cycle credit performance as a result.

I’ve separately read a couple hundred Better Business Bureau (“BBB”) reviews/complaints for OneMain. Following the Wells Fargo account scandal, I believe that this type of diligence if mandatory for financial services companies. In short, OMF received relatively strong feedback. OMF has an A+ BBB rating and has received 4/5 stars for customer reviews. The vast majority of customer complaints are centered around OMF’s conditional pre-approval process. Prospective borrowers would receive conditional approval for a $5k loan, visit a branch, and fail to ultimately qualify due to a lack of documentation or residual income shortfalls. Of all the things that could be complained about, this isn’t bad.

The primary regulatory risk is from the CFPB, although OMF has historically done well from a compliance perspective. Even before Mulvaney began aggressively constraining CFPB initiatives, prior CFPB priorities were not particularly relevant to OMF’s business model. The most important and relevant focus is properly marketing and selling credit insurance to customers. When I visited the branch, they stressed how this controversial product is conservatively marketed to borrowers. The product requires explicit opt-in and employees provide multiple opportunities to subsequently opt-out. The prior Cordray-led CFPB was even focused on implementing ability-to-pay requirements in payday lending. This initiative has since been undone, but it actually speaks highly to the regulatory view on OMF’s underwriting and business model.

Still… if other installment lenders are similarly doing some form of ability-to-pay underwriting, is OMF’s business model really that differentiated? Shouldn’t competition gradually erode excess returns? And isn’t all of this underwriting expensive? Although other installment lenders may indeed do thorough underwriting, this ignores the other key part of the equation – scale – and OMF is the largest subprime installment lender in the industry. And as we will see next, scale is a critical factor in managing expenses and driving profitability.

The scale advantage

As noted, OMF is the clear leader in subprime installment lending. This has allowed for profitable investment in technology solutions and efficiencies. Its centralized underwriting and late-stage delinquency servicing capabilities scale well. In addition, OMF has driven growth and operating leverage at the branch level.

The best way to appreciate the role of operating leverage in this business model is to compare OMF today vs. the smaller Springleaf in 2014. Please note that all displayed metrics are for OneMain’s Consumer & Insurance (“C&I”) segment, which excludes certain non-core segments. These non-core segments have historically included run-off subprime mortgages, run-off real estate finance loans, and a JV in a consumer loan portfolio (OMF sold its JV stake in 2016, but continues to service these loans). The contribution from non-core segments is currently immaterial following years of natural run-off and divestitures.

There are several interesting takeaways when comparing 2018 (post-merger) vs. 2014 (Springleaf pre-merger). Post-merger efficiency gains, investments in technology, and increased centralized capabilities have actually reduced the total number of OMF employees per branch (including corporate & centralized servicing employees) from 7.2 to 6.4 (-11%). It’s likely that some of this is driven by a similarly-sized centralized servicing team being spread over a larger branch footprint. Despite having lower headcount per branch, however, these branches service 13% more loans on average. This means that each employee is servicing 27% more loans – a large gain in productivity. Other tailwinds to productivity include the fully touch-screen enabled digitization of the loan closing process and continued reduction of in-person monthly payments at branches.

In addition to these strong productivity gains, we have the underlying Direct Auto trend. Direct Auto loans are larger and we can see a 65% increase in the average loan size over this period. This has meant that the average receivables per branch (in dollars) has actually increased from $5.4mm to $10.1mm (+86%). The higher secured mix has pressured the average loan APR, however revenue is still up from $1.6mm to $2.6mm per branch (+63%). With opex only up +19% per branch, OMF has driven strong operating leverage. Taking everything together, after-tax earnings per branch has increased from $213k to $430k (+102%).

Remember that the Direct Auto product is a relatively new offering for Springleaf. Unlike other secured installment products that exist in the marketplace (older autos, household goods, boats, jewelry, savings accounts), newer vehicles are earlier in their depreciation curves, higher value, and therefore a bit trickier to underwrite. The remaining depreciation curve is generally much easy to predict for a 12 year old car vs. a 3 year old car. Before merging with Springleaf, Legacy OneMain branches didn’t even offer the product. The new OneMain has gradually built a Direct Auto ABS program, educating investors along the way. Very simply, not every competing installment lender has this capability.

We also know that competitors are smaller than OneMain, and therefore lack the scale benefits and operating leverage that this analysis demonstrates. Put differently, OneMain’s key private competitors probably look more like Springleaf in 2014 than OMF today. And although most customers apply online, most online applications still visit a branch to close the loan. This means that OneMain’s expansive branch network and leading branch density perpetuate its customer capture rates, leading market share, scale advantages, and profitability edge.

Financials & recent trends

Immediately below are key financial and operating trends for OMF’s C&I segment. As a reminder, the OneMain acquisition was completed in Q4 2015 (distorting certain year-over-year comparisons).

I’ll provide a few observations on the above. Although OMF’s consolidated yield on receivables has declined over this period (27.0% to 23.9%), this is entirely driven by the increase in lower-yielding secured loans. On a product-by-product basis, Unsecured yields are up +50bps, Hard Secured yields are up +40bps, and Direct Auto yields are up +220bps. Put differently, OMF has tightened pricing over this period on a mix-adjusted basis. The last thing you want to see is a lender getting aggressive in the late stages of an economic cycle, but I don’t think we have that here.

A related pushback I’ve heard is that OMF has grown its loan balance at a healthy clip over this period. When adjusting for the merger, OMF’s 2014-2018 loan growth CAGR is +7%. It’s a similar idea – you’d prefer to see slowing loan growth at the late stage of an economic cycle. I think this interpretation would be a bit misleading though. First, remember that the unemployment rate had dropped from 5.6% to 3.9% over this period. This has boosted the broader subprime population’s disposable income levels and debt capacity. Next, remember that all of OMF’s recent loan growth is from incremental originations of larger-sized secured loans. When adjusting for the merger, OMF’s unsecured loans have actually declined by -3% per year. Over this same time period, TransUnion estimates that the broader unsecured installment loan market has grown from $63bn to $132bn – a +20% annual CAGR! Most of that growth has been from new fintech firms, which we know are still licking their wounds on recent credit issues. When properly analyzing the data, we actually see that OMF is prudently giving up share in the riskier unsecured market. As we potentially enter the end of the current cycle, OMF is doing exactly what you want to see – tightening up the credit box, giving up share in riskier loans, and focusing on lower risk secured growth.

We’ve already talked about expenses, but it’s worth bringing up again. OMF has reduced its operating costs – measured as a percentage of receivables – from 13.5% to 8.1%. Management has guided for 2019 opex growth of +3% – well below receivable growth of +5-10% – implying another year of solid operating leverage.

NCOs were stuck at ~7% immediately post-merger, but this was impacted by the declining employee servicing productivity during the system integration issues. In addition, the oil-patch weakness caused a spike in certain exposed states like Texas. Following some of these speedbumps, NCOs declined 50bps to 6.5% in 2018. Management is guiding for NCOs <6.5% for 2019. It’s worth pointing out that the broader consumer lending universe (ex-mortgage) has seen NCOs creep higher in recent years for auto, card, and other consumer products. OMF’s improving credit performance is an outlier, and a testament to recent underwriting and product trends.

OMF exited its 2013 IPO over-capitalized, somewhat mitigating funding needs for the Legacy OneMain acquisition. Fortress was nonetheless aggressive in its acquisition funding strategy – exiting 2015 at 4% TCE – however OMF has rapidly de-levered from organic earnings generation. GAAP earnings were initially well below adjusted C&I earnings immediately following the deal, however the two figures are close to converging today. The primary historical drivers of the delta between adjusted earnings and GAAP earnings were integration costs in 2017-2018 and ongoing purchase accounting adjustments. Legacy OneMain’s receivables were initially recorded at a premium, resulting in excess non-cash and non-economic amortization expense. OMF has also actively repurchased certain high cost bonds at a premium in the open market, resulting in additional GAAP losses. Very few financial companies provide “adjusted” earnings metrics, so some traditional Financials Analysts were uncomfortable in looking past depressed GAAP earnings. In 2020+, this should be an immaterial issue.

The company guides to a 5-7x adjusted leverage ratio target, which approximately equates to a ~13% TCE ratio. Given OMF’s through-the-cycle profitability profile, I find this target perfectly adequate (large-cap banks are comfortable closer to ~8%). I model OMF getting to 6x adjusted leverage at Q4 2019, which is in-line with guidance on their Q1 2019 call. Although excess leverage has flattered ROEs in recent years, OMF should still comfortably produce ~30% tangible returns with a 13% TCE ratio in the current loss environment. This level of profitability is better than virtually all U.S. lenders that fund and retain loans on their balance sheet. I’ve spent a lot of time explaining why I think this is a great business, and the returns should speak for themselves.

With all of that said, there is one unfortunate nuance to all of this… a generational shift in loss reserve accounting! Starting in 2020, most of the industry will be required to shift from the current “probable and incurred loss” reserve methodology to the new “current expected credit loss” (“CECL”) framework. Although this accounting shift has no impact on OMF’s underlying free cash flow production or economics, we sadly need to spend some time on this topic. I’ll go through more detail in a later section, but in short, this accounting change will result in an immediate step-up in reserves and reduction in capital.

Profitability scenarios

Given the underlying macro sensitivity and uncertainty, I don’t think there’s tremendous value in trying to accurately predict near-term EPS for OMF – I have an in-depth model, but will refrain from posting it. With that said, given our detailed understanding of the unit economics, we can do some analysis to help understand the range and potential earnings power in different economic environments. Most importantly, we want to better understand how much earnings OMF could generate in a recession. I’ll model two 2020 earnings scenarios: “Base Case” and “Recession.” All of my estimates are “pre-CECL” and instead conform with current GAAP accounting and provision methodologies (I’ll deal with accounting impacts in the next section).

Base Case assumptions:

  • Receivable growth: +7.5% for 2019 (guidance of +5-10% for 2019), +7.5% for 2020
  • Average loan yield: 24.0%, flat with Q4 2018 (guidance of “stable yields” for 2019)
  • NCOs: 6.5% (guidance of <6.5% for 2019)
  • Average funding costs: 5.5% of receivables (consistent with 2018)
  • Ancillary revenue & credit insurance claims: consistent with 2018 levels
  • Opex: +3% for 2019 (guidance of +3% for 2019), +3% for 2020

A few things to note. In recent years, the portfolio mix was dramatically increasing its secured mix, distorting certain consolidated metrics. It appears that trend is dramatically slowing and mostly over, however, so I’m not explicitly modeling any mix impacts going forward.

Management has guided for the average portfolio yield to remain “stable” for 2019, and I extrapolate this further into 2020. A frequent question is: how much higher could this go (if needed)? It’s a tricky question, because different states have different APR caps for installment lending under local usury laws. For example, the 10-K shows that 9% of OMF’s loans are in Texas, and we can look up and see that Texas has a 30% APR cap for this product. 7% of loans are in North Caroline, and North Carolina has a 27% cap. Some caps are as low as 24% (Florida), some as high as 60% (Georgia), and certain states lack explicit caps. Based on OMF’s geographic disclosures, I calculate that OMF’s weighted-average local cap is ~29-30%.

Relative to OMF’s average yield of 24%, this suggests that OMF has some latent pricing room. It’s a bit less than it seems though. If we look at the ABS disclosures, we can see that OMF’s average Unsecured product APR is already 29.0%, its Hard Secured APR is 28.0%, and its Direct Auto APR is 20.8%. OMF is probably mostly tapped-out on raising unsecured yields, has some modest room on its Hard Secured product, and therefore the majority of any price increases would need to be on the Direct Auto product. I’m assuming that OMF won’t aggressively raise APRs in states that lack APR caps and materially “subsidize” capped states. I think OMF has some further room to raise Direct Auto APRs, but management should want to maintain an APR discount relative to its other product offerings. The value proposition of the Direct Auto product is its ability to responsibly upsize loans for customers, enhancing OMF’s economics. But this only works if it’s cheaper and customers choose it. In summary, I think OMF easily has ~100bps+ of realistic residual pricing power before it starts to run into tension with state caps. From a customer’s perspective, a 100bps increase in APRs would have negligible impacts on total monthly payments.

Offsetting some of this pricing constraint is the fact that OMF should continue to realize positive operating leverage. Entirely consistent with recent years, OMF has guided to 2019 opex growth of +3%, well below the guided receivable growth of +5-10%. This should result in further declines in the opex ratio. I expect this trend to continue for 2020. Wage pressures should continue to mostly be offset by employee productivity gains. Management has generally demonstrated strong focus on controlling costs. Despite characterizing 2018 as an “investment year,” total C&I opex only grew +4%.

A few comments on funding costs. From 2016 to 2018, OMF has reduced its lower-cost secured funding mix (ABS) from 58% of OMF’s total debt to 48%. In addition, LIBOR increased from ~0.25% in 2016 to ~2.30% in 2018. Both of these trends are headwinds to funding costs, yet OMF’s interest expense (as a percentage of receivables) has actually remained flat at 5.5%. OMF was able to achieve this because it has de-levered from 17.4x to 6.9x (reducing unsecured spreads) and ABS collateral performance has been incredibly strong (reducing secured spreads). The rating agencies have rewarded OMF over this period, upgrading its corporate rating from B3/B to Ba3/BB-. On the ABS side of things, rating agencies have upgraded the most senior tranches from A+ to AAA, reducing top tranche spreads from ~180bps to ~60bps. OMF was the first consumer lending company to achieve a AAA rating on a senior ABS tranche. SoFi would become the second company to achieve a AAA rating, however SOFI’s securitization needed an average FICO of 750 to achieve this milestone.

To summarize, OMF’s strong corporate de-levering and loan performance has allowed it to maintain flat funding costs despite base rate pressures and growing unsecured funding. As OMF continues to further de-lever and issue new ABS at tighter spreads relative to existing issues – holding market conditions constant – OMF’s average funding costs should further benefit.

By shifting the capital structure towards more unsecured funding, OMF has also extended its average funding duration and termed out its future maturities. Given today’s tight credit spreads and low interest rates, this makes perfect sense. Only $1.3bn of unsecured maturities are due through 2020 (~15% of outstanding unsecured bonds). This means that, to the extent a recession occurs in 2020 and credit spreads rise, OMF should have a manageable amount of funding that needs to be rolled in a stressed environment. Although ABS credit spreads blew out significantly in 2008-2009, they were really only elevated for ~12-18 months and a “normal” recession should have less pain. As a result, I think modest funding pressures could mostly or entirely be offset with higher APRs. This was the strategy during the last recession, and OMF increased APRs by ~200bps in 2008-2010. OMF has less room to use this strategy again, but it’s certainly a tool.

With all of this in mind, I’ll lay out my “Recession Case” assumptions below:

  • Receivable growth: +7.5% for 2019, +0% for 2020
  • Average loan yield: 25.0% (+100bps to offset funding pressures)
  • NCOs: 9.0% (vs. our prior peak loss estimate of 7.8%-10.0%)
  • Average funding costs: 6.5% of receivables (+100bps vs. 2018, implying a couple hundred bps of spread pressures)
  • Ancillary revenue & credit insurance claims: consistent revenue with 2018 levels, but credit insurance claims double
  • Opex: +3% for 2019, +3% for 2020

Taken together, our EPS estimates are below.

A few quick observations. If the economy merely chugs along, OMF could do ~$6.60 in 2020 EPS. OMF is effectively properly capitalized at that time, and would still be doing 32% ROTCEs. For perspective, prime-focused consumer finance banks like ALLY, SYF, and COF are currently doing ~16% ROTCEs, but with more leverage (10% TCE ratio). If you look at subprime and near-prime focused auto lenders, CPSS, CACC, and SC are currently doing ~18% ROTCEs (although there are some earnings quality concerns in these numbers). Banks aren’t doing any better. OMF is earning better returns than virtually every major U.S. lending model.

I hope we’re at a point where this isn’t too surprising! We know that depositories are dis-incentivized from making these loans due to overly-punitive Basel III regulations, leading to some modest pricing power. The underwriting model is extremely thorough and somewhat novel, resulting in attractive through-the-cycle loss rates for this type of customer. This is expensive to do, but OMF has aggressively reduced costs through scale, centralized operations, and process optimization. Employees are properly incentivized and the culture is competitive and energetic. There’s no single secret, but many factors work together to create OMF’s superior economics.

Still, we know that OMF is inevitably cyclical. My Recession Case EPS of ~$2.60 is well below current earnings, although I believe this case is highly conservative. With that said, this still implies a solid ROTCE of ~13%. Until I performed this analysis, I was surprised at just how profitable OMF could be in a recession. For most lenders, you pull up the 2009 10-K and see just how bad things got during the financial crisis. We can’t simply do that for OMF, so it takes a lot more work and effort. I think OMF’s potential recession profitability is much higher than what investors expect, because most investors aren’t doing this work.

CECL & capital return

We now sadly need to talk about accounting. I want to reiterate that CECL will change none of the underlying business economics of OMF, yet it will potentially change how regulators and rating agencies view capital adequacy. Capital levels affect the ability to return capital through dividends and buybacks, so we need to think through this at least a little.

Let’s first discuss how lenders currently provision for loan losses. The existing provisioning framework is the “incurred loss” method. Very simply, lenders have historically reserved for losses only when such losses are probable. This method relies on recent loan performance, minimizing the need to make forecasts and assumptions. Most lenders will look at their current level of delinquencies, look at recent roll-rates to estimate the percentage of delinquencies that will ultimately end up in losses, and provision for this amount. For some loan products, there is a relatively long period of time between the identified event that caused the borrower to default and the loan’s ultimate default. For other loan products, the timing and visibility is less. This has led to different levels of “loss emergence periods” for lenders.

For example, OMF’s 2018 NCO rate was 6.5% and the company maintained a 4.8% loss allowance exiting the year. We can therefore determine that OMF’s loss emergence period is ~8-9 months (4.8% รท 6.5% x 12 = 8.8 months). OMF is looking at recent delinquency trends to project how many currently delinquent borrowers will likely default. A 4.8% reserve does not mean that OMF expects 2019 full-year NCOs to be 4.8%, but instead reflects the reality that OMF generally has ~8-9 months of visibility before a loan actually defaults. OMF only reserves for probable and incurred losses, so its reserve policy is relatively constrained by actual data experience.

This method has generally worked fine, however some regulators believe that it broke down during the financial crisis. Lenders knew that the world was melting down around them, however they couldn’t reserve for higher losses until the “probable” loss threshold was crossed. This delayed the recognition of losses for many lenders, potentially making historical financial statements less useful to investors and regulators.

Separately, some argue that the incurred loss methodology is overly pro-cyclical. Lenders are relying on recent delinquency roll rates to forecast current delinquency balances. When times are good: strong recent trends are extrapolated further, loss provisions are low, profits on newly originated loan are high, additional capital is created to lend against, and this flywheel continues to turn. When things are bad: loss provisions finally catch up, weak recent trends are extrapolated further, capital is eroded, and lenders respond by reducing lending. The incentives naturally encourage lending growth in the good times and reduce credit availability in the bad times. Lenders provide the fuel that drives the economic machine, so this pro-cyclical behavior is believed to exacerbate business cycle peaks and troughs.

Whether correct or incorrect (I have my doubts), FASB has responded by proposing a new reserving methodology. The result is the “current expected credit loss” framework, or “CECL”. Despite bitter challenges from the industry, CECL is currently scheduled to go into effect on January 1st, 2020 for most large lenders.

There are several key distinctions. Most importantly, lenders are now required to reserve for a loan’s lifetime projected loss rate at the time of origination. Instead of waiting for signs of a probable incurred loss, lenders get ahead and reserve for everything today. In general, this should reduce reserve volatility over the cycle. There is a certain amount of logic to this framework, but there are also several challenges. To start, projecting a loan’s lifetime loss rate involves a multitude of assumptions around the future business cycle and product-specific performance. Assumptions must be backed by data to stand up in an audit. Some lenders will try to conservatively ensure compliance by relying on objective industry-wide datasets, but this may be inappropriate for lenders with specific product and geographic focuses. CECL may have a certain degree of theoretical elegance to it, but there’s no doubt that its actual implementation is a pain for all involved.

This isn’t an opinion piece of CECL, however, so let’s discuss how it could impact OMF. I’ve spent a lot of time reading CECL white papers, rating agency notes, and auditor opinions to try and refine my understanding. That was mostly a waste of time, however, because CECL truly gives a lot of freedom (and uncertainty) in how lenders project lifetime losses. Some lenders have been more vocal than others when discussing potential CECL impacts (JPM estimates that CECL will increase reserves by 35%), but OMF has largely been quiet thus far.

What we do know is the fact that OMF will need to reserve for lifetime loss estimates. We also know that OMF is only reserved for ~8-9 months of probable losses under the current methodology. And because OMF’s loan products carry a much longer duration, it should be clear that OMF’s loss reserves need to go much higher. Partially mitigating this issue is the fact that OMF’s effective loan duration – its average loan duration when incorporating prepayment activity (OMF’s loans have no prepayment penalty) – is much less than its average contract term of 4-5 years.

The below analysis is an extremely simplified attempt to project OMF’s loss reserves under CECL. I originally utilized ABS pool data to refine lifetime loss estimates, but ultimately didn’t think the additional work was really enhancing precision (given the multitude of assumptions needed). We’ll know what the CECL reserve will be relatively soon, and really just need a rough estimate in the meantime.

To reiterate – this analysis is highly simplified and conceptually flawed. It effectively says: if OMF’s average loan duration is ~1.5 years, then it should roughly reserve for 6.5% annual NCOs x 1.5 = 9.5%. Given OMF’s current reserves are 4.8%, this means that OMF needs to increase its reserves by ~$820mm pre-tax on 1/1/20. This will reduce tangible book value per share by ~$4.60, increase my estimated Q4 2019 leverage from 6.3x to 8.3x, and reduce the TCE ratio from 12.0% to 9.0%. This is a relatively large hit but remember – OMF’s extremely high return on equity ensures that “lost” capital will be replenished relatively quickly (roughly in one year’s time).

All of the above math is related to the January 1st one-time capital impact of CECL. After that, the ongoing earnings impact is a bit more nuanced. The provision expense is forward looking and reflects a loss estimate on your ending portfolio balance. This means that a growing portfolio generally produces a provision expense that is in excess of the period’s actual NCOs. To simplify things, your provision expense not only reflects the actual NCOs realized during the period, but also the loss allowance on your incremental loan growth.

Going forward, OMF will be forced to reserve for losses on incremental loan growth at a higher allowance rate. I expect OMF to exit 2019 with $17.4bn of loans. Let’s use my base case and assume that 2020 growth is +7.5%. This means that OMF should generate an incremental +$1.3bn of loan growth in 2020 ($17.4bn x 7.5%). Under the incurred loss methodology, this incremental growth would increase the provision expense by $62mm ($1.3bn x 4.8% allowance rate). Under CECL, however, the incremental growth will increase the provision by $124mm ($1.3bn x 9.5% allowance rate). This incremental expense will produce an EPS drag of $0.34. Relative to my Base Case EPS estimate, this only results in 5% EPS dilution and is entirely manageable.

You should realize that the CECL EPS “headwind” is higher when loan growth is higher and vice versa. Our Recession earnings case assumes 0% loan growth, so there wouldn’t be an incremental CECL EPS headwind in that illustrative scenario. To summarize my CECL views, I expect a large but manageable one-time capital hit on January 1st, a subsequent quick replenishment of that capital, and a modest ongoing EPS impact.

It would be great if the company had already provided its preliminary CECL impact analysis, but it is what it is. We do have some clues though. To start, Moody’s upgraded OMF’s corporate ratings in February 2019. OMF also paid its first dividend in February 2019. Although management hasn’t been open to investors around discussing potential CECL impacts, there’s no doubt that they’ve been in constant contact with the ratings agencies about likely impacts and implications. Although Moody’s didn’t discuss CECL in its upgrade note, there’s also no doubt that estimated CECL impacts were taken into account. Management has also been very patient around finally initiating a dividend. Given OMF’s perennially constrained balance sheet, the dividend is a powerful symbol and sign of victory. The combination of the Moody’s upgrade and dividend initiation strongly suggests that management is comfortable with the pending CECL impacts.

OMF has generally traded at ~6x earnings following the merger. A fair question is: what gets this stock to work? Why should anyone pay 10x for a “subprime lender”? In short, I think everything changes with the initiation of the buyback. OMF’s high returns on capital ensure that it produces significant distributable cash flow that could be used to accretively buy back stock. OMF isn’t regulated by the Fed’s CCAR process, so it can also be more aggressive and opportunistic with its capital plans. The only reason OMF hasn’t yet bought back stock is because it was previously entirely focused on de-levering. CECL pushes this out slightly, but we’re very close. Management has been careful to avoid making hard promises around a buyback initiation, but this should be a serious discussion ~12 months from now.

To illustrate how a regular buyback could “force the market’s hand,” let’s do some simple math. Let’s say that long-term C&I ROTCEs (at a 13% TCE ratio) hold steady at 32%, implying a 29% GAAP ROTCE. Medium-term loan growth is +5% (2x household income growth to account for the low penetration of installment loans) and OMF continues to pay out 20% of C&I earnings in the form of a dividend. If you do the math, you’d conclude that ~55% of C&I earnings would be available for buying back stock after funding growth (on a leverage-neutral basis) and the dividend. Let’s assume that the stock is still stuck at 6x earnings, implying a 17% earnings yield. We can take our 17% earnings yield x 55% buyback capital return ratio = 9% annual float reduction. If we’re growing net income at +5% (in-line with loan growth) and reducing the float by 9% per year, then OMF should produce +16% medium-term EPS growth. If you add the implied 3% dividend yield, we get a total prospective annual shareholder return of +19%.

This is all hypothetical math, but my point is that the ability to redeploy distributable income into buybacks changes things. In the above example, OMF could either (1) continue to trade at 6x earnings, buyback stock, and grow EPS at incredibly attractive rates, or (2) finally see its stock re-rate to a more realistic PE multiple (somewhat mitigating the silly buyback accretion). We may go through a dip before getting there, but I still believe that the long-term valuation target should indeed be 10x+ PE.

Thoughts on the stock here

To summarize – I believe that OMF is an extremely misunderstood and attractive business. It possesses durable and structural competitive advantages that should produce high returns on capital for years to come. Barriers to entry protect the moat. Investors lack an appreciation for all of this because few are willing to roll up their sleeves and work on a subprime lender. And although anyone can look at OMF’s currently depressed PE ratio and high ROE (it certainly comes up on value screens), these extreme metrics are frequently brushed aside as being distorted by late-cycle dynamics. Investors can’t pull up a 2009 10-K and see how OMF would perform under stress. I truly think that it takes a lot of work to fully appreciate OMF’s business quality, although hopefully I’ve saved some people some time in that area.

What do I think OMF is worth? If I ignore where we are in the cycle and “fully” capitalize my $6.60 Base Case EPS at 12x, you would conclude $79/share of intrinsic value (+150% vs. the current price of ~$31). But I’m also realistic, and the stage of the cycle does matter here. If things indeed turn for the worse, the market won’t care and simply apply a low multiple on a declining EPS estimate. It could be very painful! I think the exciting part is on the other side of things. I believe that OMF’s trough recession profitability will be significantly better than most expect, and this validation of the model will provide a historical blueprint for investors to apply a more constructive valuation on the upswing.

And if the stock really does poorly, I could see Apollo trying to take you out at a premium to a depressed share price. Apollo/Varde are clearly smart, appreciate that they’ve bought 40% of the company late-cycle, and have a plan. This is actually another argument to potentially wait things out a bit. The last thing you want is to purchase the stock at $31, watch things turn for the worse, watch the stock irrationally dip, and then be taken out by Apollo/Varde below your original cost basis. Depending on where you buy the stock, Apollo/Varde could either be a positive or negative catalyst. I have no particular insight into Apollo’s plans for this investment, but I do suspect that they think about the business similar to me.

Even if a future recession is mild and trough earnings settle closer to ~$3.50 (consistent with 8% NCOs), I could still see the market putting 6x on this number and the stock bottoming out at ~$20. Part of the problem with OMF’s high returns on equity is the fact that – even at a low PE multiple – the stock continues to trade at a healthy premium to tangible book value. OMF should indeed trade at a premium to tangible book value – it has franchise value that produces strong excess returns – but many financial investors look to tangible book value as a simplified downside valuation estimate across their coverage universe. This is a highly simplified and flawed approach, but it is what it is. I sadly think we’re more likely to see $20 before we see $40.

That’s not to say that I need a recession to invest in OMF. As we saw in December, the stock substantially de-rated to $23 on virtually no company-specific news. At that price, I think you were getting properly compensated to step into cycle risk. This is the unfortunate reality of the stock, but also a source of opportunity. By spending a lot of time on the business up-front, I’m hopeful and ready to the extent another dislocation occurs. I’ve thought about trying to hedge a long position with different shorts or options, but think cash-hedging and patience is the best bet right now.

The next 24 months will likely be volatile, but I do think ~$79/share is the intrinsic value / takeout target on the other side of a cycle. In the low $20s, I think you get fully compensated for the uncertain path it may take to get there.