October 21, 2023 - 11:15am EST by
2023 2024
Price: 82.19 EPS 11.93 13.59
Shares Out. (in M): 250 P/E 6.9 6.0
Market Cap (in $M): 20,543 P/FCF 6.9 6.0
Net Debt (in $M): 0 EBIT 0 0
TEV (in $M): 0 TEV/EBIT 0 0

Sign up for free guest access to view investment idea with a 45 days delay.


Discover Financial (DFS)

DFS has been written up 7 times on VIC before, as it is a high ROE business with stable NIMs and a solid compounder.  EPS has grown by 16% / year since 2006.  But the company has a lot of problems right now outside of just the cycle starting to turn (potentially).  

Discover recently lost their long time CEO, Roger Hochschild who built DFS from its spin off in 2005 with EPS of $2.26 to $15.50 last year.  Compliance issues (the 2nd in a year) meant that compliance costs are going up (~200mm per year or 60c in EPS), buybacks have been halted and to top it off, delinquencies are rising and hitting the income statement pretty hard right now.  It is messy and we haven’t even gotten to the complications of somewhat new CECL accounting rules.

It all started last July when Discover announced that they had overcharged certain merchants dating back to 2007. The amounts were small, less than 2 basis points of sales per year, and only about $255 million in cumulative earnings, but it opened a small Pandora's box of 2 compliance issues in one year. Roger Hochschild was tossed out a week later and a new interim CEO was appointed (John Owen).

DFS stock has fallen by over 30% just since mid July.

On the plus side, the company reached a consent order with the FDIC on the consumer side at the end of September, with no fines but broadened oversight of the company. The final decision on the merchant misclassifications are reportedly also finished at DFS; the company is now awaiting regulatory approval for this.

So, the regulatory overhang seems mostly behind the company, but share buybacks remain halted until the final nod from the FDIC is given on the merchant stuff.

As for recent trends and third quarter results, DFS reported a somewhat sizable EPS miss last week. Revenue was solid, up 16% YoY, beating expectations by $80 million. But EPS came in at $2.59, missing estimates by $0.59. The company earned $3.56 last year in Q3.

The miss however was entirely driven by higher provisions for loan losses. Pretax, pre-provision net interest income was $4.0 billion, up an impressive 16% YoY. NIMs were 10.95% in the quarter, about typical for DFS and entirely unaffected by changes in rates or the slope of the yield curve; loans revolve in tandem with Fed Funds so Discover is not seeing pressure on net interest margins.

Below is a good bridge.

A screenshot of a computer screen Description automatically generated

Note that Provisions above (which we highlighted in yellow) were a $2.59 headwind to earnings in Q3.

In the quarter, net charge-offs (which are essentially realized losses) were $1.1 billion. Provisions, which flow through the income statement, were $1.8 billion.

See below.

A screenshot of a credit loss report Description automatically generated

Over the life of a loan, provisions should equal charge offs. But today the company is frontloading a lot of future loan losses it seems. Provisions made on a quarterly basis are partly driven by new CECL rules and partly tied to expectations for future credit losses.

Under CECL accounting, new loans (or card balances) now must be provisioned for on day one. As loan growth was quite strong, day one provisions grew more than expected.

That was roughly 50% of the increase in provisions in the quarter.

The other half of the increased provisioning was driven by the company’s current view of the macro environment. With delinquencies up to 3.1% in Q3 (vs 2.6% in Q2), management is seeing stress in the “lower to mid FICO bands” as they put it. 

To give some background, many investors have been worried that excessive loan growth in 2021 to 2022 was due to lax underwriting standards. As the company explains, the company has maintained the same underwriting standards since 2021.

But, in 2022 for a couple of quarters, Discover “ran some tests” with marginal prime and sub-prime card candidates. These tests ended in Q3 2022. Those accounts did not meet return or volatility thresholds and so were halted. The bucket remains strong, but this “testing” in 2022 led to loan growth that was larger than normal.

Given that it takes 12-24 months from origination for delinquencies to become seasoned, these lower quality card delinquencies and charge offs are hitting the books starting this quarter, but should peak in mid-2024.

Management then expects a plateaued level of delinquencies for 2 quarters (through 2024) before they start to fall in 2025.

If delinquencies don’t slow in the first half of 2024, the company further suggested that economic conditions must be worse than expected (and they are already expected to slow substantially, unemployment to hit 5% and DQs to normalize).  

In 2020, the company took provisions of $5.1 billion, but only charged off $2.7 billion of loans that year. While not quite as large this year, DFS is on track to provision for $6 billion of losses (yes, higher than 2020) while forecasting charge offs of $4 billion.

Given that they over-reserved for losses in 2020, DFS cut provisions basically to zero (well, $200 million) in 2021 (against charge offs of $1.6 billion).  So EPS was too low in 2020 but overstated in 2021.

For 2023, the company expects 3.5% of loans to be charged off in 2023. The 2006-2022 average charge off rate is 2.94%.

A graph showing the amount of loss of a loan Description automatically generated with medium confidence

So net net, DFS is this year expensing more in provisions than they did during 2020. That should be baking in a lot of potential recession pain already.

In a soft landing scenario, DFS probably ends up cutting provisions again to a small number ala 2021.  That could make 2024 and 2025 earnings quite high in a soft landing scenario.  

2024 and 2025 Estimates

We took management at their word that delinquencies on 2022 vintage balances will peak around mid-2024. John Owen on the call mentioned that they expect mid-single digit loan growth next year.

Therefore, we modeled 4% loan growth (which is just inline with inflation and will mean far less upfront CECL provisioning). Charge offs probably move higher, so we assume they move from 3.5% of loan balances (per guidance) to 3.8% in 2024.  In 2025, they then normalize to 3.5% (which is still higher than their long term average at 2.9%).

Admittedly, we are not expecting a pop in delinquencies to 7.7% as happened in 2009, but a recession of that magnitude would obviously be problematic for more than just DFS. A mild recession DFS can easily handle from an earnings/reserves point of view.

In short we come up with this.

A screenshot of a computer Description automatically generated

Our “normalized” runrate EPS figure today is $13.84.

Back in 2019, EPS was $9.07 (when DQs were fairly typical). Today, 4 years later, loan balances are 32% higher. It makes sense that normalized EPS should be in the $12+ range.

The valuation with DFS trading at $84 per share certainly bakes in a lot of bad news. Even with provisions where they are, pure run rate EPS is $10.36. We model about $12 in EPS this year, lower than Street estimates by about 60c (and inline with the quarterly miss).

But we model EPS potential of $13.50-16.60 over the next 2-3 years. DFS tends to trade around 9-10x earnings, so we see a lot of value here. But the value trap potential is pretty high now given delinquency trends/economic weakness potential. The name is also complicated by the FDIC compliance issues, lack of buybacks, no permanent CEO and the messiness of CECL accounting.

The news that DFS dabbled in subprime customers in 2022, even if only on a small basis, also perhaps spooked some investors.

Long term, it is hard to see how normalized credit, even at higher than long term average charge off rates, do not result in DFS able to earn mid teens EPS in 2-3 years.  In our bull case, once past the economic cycle, the stock could easily be a double.  So we like the risk reward here.

A close-up of a table Description automatically generated

The cycle is the big risk here, but likely heavily baked into the stock. DFS has fallen over 30% from its peak just a few months ago.  The company is hated, but quality in the sense that 1) ROEs are quite high (typically mid 20s), 2) the card business is getting less competitive given higher rates (and fewer disrupters in the card space), 3) NIMs are borderline immune to rate moves and 4) free cash flow is solid.  

At 7x earnings, DFS now offers a healthy 3.4% dividend yield (that looks 3x covered). Soon we think the company restarts buybacks which have averaged 5% of shares outstanding each year (from 2019 to 2022).

Finally, the balance sheet is solid with common equity tier 1 capital ratios of 11.6%.  Deposits are up 25% YoY and grew 5% sequentially. 



I do not hold a position with the issuer such as employment, directorship, or consultancy.
I and/or others I advise hold a material investment in the issuer's securities.


Resumed buybacks, provisioning headwinds to moderate

    show   sort by    
      Back to top