January 23, 2018 - 9:32pm EST by
2018 2019
Price: 78.00 EPS 7.75 8.35
Shares Out. (in M): 366 P/E 10 9
Market Cap (in $M): 28,550 P/FCF 0 0
Net Debt (in $M): 0 EBIT 0 0
TEV ($): 0 TEV/EBIT 0 0

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


DFS is a large cap stock that has been pitched to this community before.  My first real work investigating the stock was sparked by Shoe's pitch in July 2014.  Today, the stock looks mispriced again. 

(1) Low P/E.  Wall Street forecasts that Discover will earn over $7.75 per share in 2018, for a 10% earnings yield or a 10x P/E ratio.  The median of constituents in the R2000 and S&P 500 is a 5-6% earnings yield or a 18x P/E ratio.  This excludes the 350 companies in the R2000 that are not even forecast to make a profit, so the concept of earnings yield isn't useful.  A 10% earnings yield is higher than 95% of the R2000 and 95% of the S&P 500.  And these are real cash earnings, in fact cash earnings are a little higher than GAAP net income.  Discover's equity capital is viewed as more than adequate, so the Federal Reserve has approved the company's plans for the past several years to return around 100% of earnings to owners by way of cash dividends and share repurchases.  The dividend yield is 1.8%.  

Because 10% is nearly twice 5.5%, many years of high growth are required for 5.5 to catch up to 10.  The scenarios are endless, but imagine one in which the earnings of 10 do not grow, and the earnings of 5.5 grow at an annual rate of 10%.  More than six years are required for 5.5 to reach 10.  Another six years are required for 5.5, once it surpasses 10, to make up for the defecit created during the first six years when the growing 5.5 was still below the 10.  This is a pretty obvious and basic concept.  But the market seems to ignore it today. Please help me with a correction if you disagree.    

My guess is the market fears Discover's earnings could decline because credit expenses are rising.  This is an important concern, but I believe the fears are misplaced and overblown.  Credit quality is deteriorating slightly from unusually benign levels.  Management reports a rise in loss given default because of higher consumer debt-to-income levels, but not a rise in frequency of defaults.  The bigger cause of rising credit costs is actually a fundamental positive:  Discover is adding significant numbers of new customers. For a cohort of credit card loans to new borrowers, the losses are front end loaded in year one and two.  After two years, the lender has sorted out which customers are profitable.  Around 10% of the new loans are charged off, and the customers are disgarded.  But the remaining 90% supply a highly profitable annuity.  Discover reports that the average tenure of its borrowers as Discover customers is 12-13 years.     

The level of Discover's current profitability and returns must be considered in relation to the 10% earnings yield.  One of my concerns is that Discover's high 22% ROTCE is not sustainable (22% is both the current level and the 10-year average).  Credit card lending is one of the most profitable products that banks offer.  And since 15% is a good ROTCE for a bank, credit card lending is attracting capital and has become increasingly competitive.  Discover has a lower cost structure so I believe it can earn superior returns.  But management's long run target is a 15% ROTCE.  I have not found an explanation for why this number makes fundamental success, and as I pointed out, the average of the past ten years is 21%, and I note that the past 10 years include the highest level of credit losses in the history of the product.  Nonetheless, there is likely to be some pressure on incremental returns and the ROTCE could decline further.  If the ROTCE declined to 15%, the earnings would provide a 5.5% yield on today's $78 stock price.  And a 5.5% yield is still the median for the R2000 and S&P 500. 

There are some positives to Discover's 22% ROTCE.  Consider an alternative investment in JPM.  The stock trades at an earnings yield of under 8% based on an ROA that is above the 20 year median and a 13% ROTCE.  Imagine that JPM and DFS can both grow assets by 13% and are both adequately capitalized.  That means that JPM must retain 100% of its earnings to support the growth. Banking is sort of the definition of capital intensive.  In this case, JPM has no earnings that it can distribute to owners.  The situation is different for Discover.  With a 22% ROTCE, the company needs only to retain 13 of the 22 in earnings (% of TCE) to support asset growth of 13%.  The remaining 9% of tangible common equity is available for management to distribute to owners.  JPM has $2.5 trillion of assets, including $0.9 trillion of loans.  DFS has $0.1 trillion of assets. Because JPM is more than 25x the size of DFS, it is possible that Discover has higher long term growth prospects.  

(2) Durable earnings.  As we think about the market's fears regarding Discover's level of earnings and returns, which I argue is the best explanation for the very high earnings yield, I point out that Discover has not incurred an annual loss since 1987, the year after the product was launched by Sears.  The company did post two quarterly losses during the Great Recession, however (both a manageable <2% of tangible common equity).  Discover's resilient earnings are due to a strong underwriting culture that has contributed to credit costs for Discover that are lower than peers, and one of the highest risk adjusted yields (loan yield minus charge-offs) among banks over the long-term.  The company's high loan yields (10%) and low operating costs (<40% efficiency ratio) cushion cyclical increases in credit costs when they occur.  Today, charge-offs of 7.5% would be required to wipe out pre-provision revenues and bring the company to break-even.  In 3Q17, the net charge-off rate was 2.6%. The highest charge-off rate for Discover was 7.8% in 2009, well below the industry peak of 9.4%.  And for context, the industry peak of 9.4% in 2009 was 1.5x the peak in 2002 and 1.8x the peak in 1997.  So 2009 was bad compared to the two previous peaks.     

Although there is no clear relationship between tax rates and returns on capital for banks during the past 75 years, it is worth pointing out that Discover's cash taxes run at 36-38% of income before tax.  So a reduction to 21% could boost net income by 25%, all else equal.  Moreover, customers that pay lower taxes may be more inclined to borrow, to spend, and to repay.          

(3) Excellent stewardship.  A 10% earnings yield is high, but the cash dividend yield is only 1.8%.  A significant portion of our return will depend on how well management invests the earnings it does not pay out in cash dividends.  The historical record posted by Discover's management is excellent in this regard.  Over the long run, management has retained 80-85% of earnings (paying out 15-20% in cash dividends) and strong capital allocation has delivered 12% per annum growth in assets per share, which I consider a proxy for earning power assuming stable returns on assets. This is exceptional. More typical is JPM, which produced 5% per annum growth while paying out 30% in cash dividends.  

Management operates with a fiduciary mindset, driven by stock ownership among top executives equal to 25x the target annual compensation, and 40x for the CEO, who owns $120 million in stock.  The average tenure for 11 executive officers is 15 years, and the CEO has held the role for 14 years. 

I like their disciplined pursuit of a well-defined strategy.  Discover seeks middle income customers with excellent credit who value the company's excellent service and will revolve a credit balance.  

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


If the recent increase in the loan loss provision stops and the expense levels out, either because the growth rate moderates or because credit quality stabilizes or improves. 

    sort by    
      Back to top