June 14, 2018 - 4:08pm EST by
2018 2019
Price: 100.00 EPS 7.25 8.05
Shares Out. (in M): 861 P/E 13.8 12.3
Market Cap (in $M): 86,100 P/FCF 14.0 12.4
Net Debt (in $M): 26,000 EBIT 7,900 8,450
TEV (in $M): 112,100 TEV/EBIT 14.2 13.3
Borrow Cost: General Collateral

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


This is an odd opportunity: a special-situation short based on litigation risk that appears unknown to most investors, sitting on the $112 billion enterprise value of one of America’s best-known, all-time-great businesses.  Although it has some disadvantages relative to Gyrodyne (100% gain in six months) – the percentage change in equity value is much smaller, extraneous factors could move the stock in the other direction, and there is a small chance that the litigation damage is already be priced in, despite outward appearances to the contrary – it has three advantages:  The large market cap and trading volume allows one to trade in size.  The initial catalyst of a court decision is no more than two weeks away and could be days away (next Monday).  And the litigation outcome appears even more certain than in Gyrodyne.  This is the only lawsuit I have seen for which one side’s odds of success, even prior to the appeal’s oral argument, were 90%.  The oral argument did not change those odds.

I recommend shorting American Express (AXP) to earn what should be ~10% alpha (return vs. market) within two months and perhaps 20% by year-end.  By the end of June, the U.S. Supreme Court will rule on Ohio vs. American Express, an antitrust case brought against Amex by the U.S. Department of Justice and 17 state governments.  Amex won in the appellate court, but with 90% certainty, the Supreme Court will reverse and Amex will lose the case.  That ruling should ultimately reduce Amex’s enterprise value by 20-40%, by causing a double-digit reduction in Amex’s merchant fees and by enabling private lawsuits from merchants with claims totaling tens of billions of dollars.  It appears that investors are largely ignoring this risk; the only substantive public discussion we have seen in the press or from the sell side came from a few analyst reports in the last week, and they were insufficiently bearish.  (To our knowledge, no one has mentioned the damages liability from follow-on merchant lawsuits.)  The Supreme Court’s opinion will bring the risk to investors’ attention and should cause the stock to fall as investors begin to realize and estimate the likely consequences.  Further stock falls should occur as the case reaches conclusion on remand, private lawsuits begin piling up, and profits fall.



The trial court’s decision:

The Second Circuit’s opinion:

Well-organized links to all of the Supreme Court briefs:

Specific Supreme Court briefs discussed below:

The Supreme Court oral argument recording:

The Supreme Court oral argument transcript:



The case involves Amex’s anti-steering rules or “Non-Discrimination Provisions” (NDPs), which for decades have been contained in its contracts with every merchant who accepts Amex cards.  The NDPs prevent a merchant from doing anything to steer a customer towards using a non-Amex card, including:

  • indicating or implying that the merchant prefers other cards;
  • trying to dissuade a customer from using Amex or to persuade them to use any other payment method;
  • criticizing any Amex card, program, or service; 
  • imposing any restrictions, conditions, disadvantages or fees on Amex users that are not imposed on other payment methods, other than cash, checks, or EFT;
  • promoting any other payment methods more than it promotes Amex, other than its own private-label credit card for use solely at the merchant.

Merchants pay credit card networks for their services through a merchant fee (a/k/a discount fee), expressed as percentage of the charged amount.  The Amex fee currently averages 2.4%; if a customer buys $100 of goods using an Amex card, the customer owes $100 to Amex, Amex pays $97.60 to the merchant, and Amex keeps the $2.40 discount.  The NDPs were intended to, and did, make it impossible for Visa, Mastercard, or Discover to compete with Amex based on the level of merchant fees.  Amex historically had much higher merchant fees than its competitors but also offered its cardmembers higher rewards and other services.  If Visa lowered its fee, its share of customer purchases at a merchant would not change at all, because the customers who choose how to pay could not benefit from those lower fees and, indeed, would not even know of the lower fees.  The merchant’s only recourse was to stop accepting Amex, and almost all merchants calculated that, if they did so, the lost profit from losing customers who wanted to use Amex outweighed the gained profit from lower merchant fees.

Over time, Visa and Mastercard decided that, if they couldn’t beat Amex, they should join Amex.  They added their own anti-steering rules to their merchant agreements and raised their discount rates by over 20% between 1997 and 2009, to levels near Amex.  Discover launched its card in 1985 with a low-fee strategy but eventually concluded the strategy didn’t work and raised its rates by 24% between 2000 and 2007.  Amex itself significantly tightened its NDPs in the 1990s and engaged in a systematic “Value Recapture” campaign that materially raised rates on 65% of its merchants from 2005-2010.

In 2010 the United States Department of Justice (DOJ) and 17 state attorneys general sued Amex, Visa, and Mastercard to challenge their anti-steering rules as antitrust violations.  Visa and Mastercard entered consent decrees and agreed to remove or revise their rules, but Amex did not.  The District Court held a seven-week trial in 2014 and ruled against Amex in early 2015.  It found:

  • Amex has market power, as demonstrated by (1) the NDPs’ proven anticompetitive effects, and (2) the market structure – Amex’s market share, high market concentration, high barriers to entry, and Amex cardmember’s insistence on using Amex.
  • The NDPs have caused anticompetitive effects on inter-card competition.
  • Amex’s proffered pro-competitive justifications – that reduced competition on the merchant side allows Amex to offer greater cardmember benefits, thus increasing consumer welfare and inter-card competition on the consumer side – were insufficient to justify the NDPs.

Amex appealed, and the Second Circuit Court of Appeals reversed and ordered judgment for Amex.  The case’s facts are complex, and the Second Circuit is full of smart judges with smart law clerks.  Yet it is easy to see that the Second Circuit botched things badly; its errors are surprisingly simple to explain and fundamental to how antitrust economics and antitrust law work.  The core issue is that credit cards are a “two-sided market,” in which each credit card purchase necessarily involves a consumer/card-issuer transaction and a merchant/card-issuer transaction.  Amex argued not that the NDPs had procompetitive effects in the consumer/card-issuer market that outweighed any anticompetitive effects in the merchant/card-issuer market, but rather that these two sides represented a single “market” for antitrust purposes, and that the governments bore the burden of proving market power and net anticompetitive effects in this single market.  That is nonsensical as a matter of antitrust analysis.  An antitrust “market” is a set of goods or services that are similar enough substitutes for each other to constrain each other’s prices.  If you don’t define “market” that way, the entire edifice of competition analysis falls apart.  Card issuers’ services to merchants are not substitutes for card issuers’ services to consumers.  The second problem is procedural yet just as fundamental: Longstanding antitrust doctrine requires the plaintiff to make an initial showing of anticompetitive effects, and, if the plaintiff is successful, shifts the burden to the defendant to prove countervailing procompetitive effects.  The court weighs both sides’ evidence and makes a judgment about the balance of effects.  The District Court’s ruling was a straightforward application of this doctrine, and the two-sided nature of credit card markets didn’t require a different framework.  Amex’s argument gave the governments a burden that no antitrust litigant or, indeed, any other type of litigant should have, which is to prove the absence of evidence favoring the opposing party.  The Second Circuit nevertheless accepted Amex’s argument and ruled that the governments had not met their burden of proof.

The states petitioned the Supreme Court to take up the case and reverse.  The DOJ did not join in the petition; it filed a brief stating that, although the Second Circuit was badly wrong on the merits, the case did not meet the usual standards for whether the Supreme Court should hear it.  The Supreme Court nonetheless granted the petition.  Merits briefs were filed last winter, oral argument occurred in February, and unless something highly unusual happens, the Court will issue its opinion before the June 30 end of term.



Five separate factors make the governments’ odds of success extraordinarily high:

(1)   The case’s merits.  Even when the correct outcome of an antitrust case seems “clear” to an antitrust practitioner who is not directly involved with the case, the actual case outcome may not be clear beforehand, because the fact situations are often complicated and antitrust law explicitly calls for courts to make judgments about competitive effects that are not directly observable or precisely measurable.  To a surprising degree, that messiness does not apply here, even though the original facts were complex.  The trial court already found against Amex on all necessary points, and its findings of fact are irreversible unless found “clearly erroneous.”  To reverse that ruling and give the win to Amex, the Second Circuit had to make the two easily-explained errors of law described above.  Once these two rulings are reversed, the case necessarily reverts back to a win for the government.  The Supreme Court will likely remand the case to the Second Circuit for further proceedings consistent with its decision, but no other issue of law exists that could reasonably support a win for Amex on remand.

(2)   The opinion of disinterested experts.  Unusually, this case’s most persuasive and influential Supreme Court briefs come not from one of the litigants, but rather from a group of amicus curiae (“friends of the court”): Professor Herbert Hovenkamp and 27 other antitrust law professors, who have no direct interest in the case other than to have courts “get it right” for the sake of legal precedent for future cases.  While the other professors are highly regarded antitrust experts in their own right, Hovenkamp’s presence is the key.  No other field of law has a professor who is more dominant than Hovenkamp is today in antitrust law.  The treatise he co-wrote with the late Phillip Areeda, whose dominance in the field was even greater, is the decades-old, continuously-updated bible of antitrust law.  To quote Wikipedia:

Professor Hovenkamp is generally regarded as "the most influential antitrust scholar of our generation" and the New York Times reported that many consider him "the dean of American antitrust law." He is the sole surviving author of Antitrust Law, the most cited legal reference on the subject.  In each of the last ten antitrust cases heard by the United States Supreme Court, either the petitioner or the solicitor general pointed to Hovenkamp as supporting the position the justices were being urged to take. Professor Hovenkamp’s writings have been cited in 36 Supreme Court decisions and more than 1300 decisions in the lower courts.

Hovenkamp is the chief guardian of antitrust law’s doctrinal integrity.  He and his peers told the Court, “the Second Circuit got it badly wrong, and you need to fix it now to avoid high-cost problems in later lawsuits.”  On top of his standing with the Court, Hovenkamp hired the right lawyers to write the briefs.  These two briefs are as good as it gets in appellate practice; they are masterful at simplifying a complicated situation and speak with an almost eerie sense of authority.  Given the DOJ’s demurral, the Court probably would not have taken the case but for this group’s amicus brief.  At least one justice has publicly praised these briefs, which is rare; near the beginning of oral argument, Justice Sotomayor led off by saying, “We have a wonderful amicus brief that explains . . . .”

(3)   The Supreme Court’s granting of certiorari.  The Supreme Court denies over 99% of petitions for certiorari.  Normally it grants certiorari only when there has been a “circuit split,” with one court of appeals deciding an issue one way and another deciding the same issue the opposite way.  The DOJ’s certiorari brief stated that Second Circuit had erred badly but opposed certiorari because there was no circuit split.  (The DOJ takes seriously its credibility with the Court in selecting which cases it does and does not recommend for review.)  For the Supreme Court to nonetheless grant certiorari suggests that it reached out to correct what it saw as an erroneous and damaging decision.  If the Court agreed with the decision below, it could have simply denied certiorari and let the decision stand.  The Supreme Court has reversed over 90% of the decisions for which it has granted certiorari without a circuit split.

(4)   Oral argument.  Before oral argument, we believed the most likely Supreme Court vote on the case was 8-1 in favor of the governments, with Justice Gorsuch dissenting.  The oral argument did not change that assessment.  Although Gorsuch spoke the most and clearly favored Amex, Sotomayor, Kagan, and Breyer clearly favored the governments, and brief questions from Roberts, Kennedy, and Ginsberg were neutral.

(5)   The court and the litigants.  Almost every lawsuit odds-making should allow for at least 20% odds of loss, even at the appellate stage and even when the case’s merits appear to be a slam dunk, due to three risks that can arise apart from the merits.  None of those risks apply here.

  • Sometimes judges just get things wrong – In lower courts, the judges can be poor or merely average legal analysts, or they are unfamiliar with a particular field of law, or the lawyers’ arguments can be sub-par.  For an example of an inexplicably poor yet good-faith decision, one need look no further than the Second Circuit opinion in this case.  However, the Supreme Court has the best lawyering possible, all of the justices are exceptionally bright, several are fluent in antitrust law, and all of them have ruled in antitrust cases before, including cases involving the “rule of reason” standards that are applicable here.
  • Philosophical/ideological bias or bias against one party – This case raises no social or political issues likely to color the court’s opinion, and Amex, a $100 billion business defending high rates for an ubiquitous, utility-like service, is not a sympathetic litigant.
  • Outright corruption – No one bribes the Supreme Court, and the justices do not need to worry about being re-elected or re-appointed.



Losing this case will likely cost Amex over $10 billion of entity value each from two different sources: reduced profits caused by lower merchant fees (and possibly by lost market share), and legal settlement payments for damages claims by hundreds or thousands of merchants, including most of the very large merchants.

First, a court loss will force Amex to eliminate its NDPs.  The entire point of the antitrust suit is that the NDPs allowed Amex and its peers to charge higher fees; those fees will fall once the NDPs are removed.  The potential hit to Amex’s profits is surprisingly large.  Amex’s 2017 discount revenue was $22.9 billion.  With 65% of Amex card-billed business in the United States, U.S. merchants should be paying roughly $15 billion of annual merchant fees to Amex.  (It could be a higher or lower if the average U.S. discount rate varies much from the reported global discount rate, but not enough to matter here.)  The trickier estimate is how much the discount rate will fall.  One must try to measure an alternative world without NDPs that has never existed in the U.S.  Most of the trial evidence and trial briefs on that point were filed under seal or redacted, and the District Court’s 150-page opinion judiciously does attempt to give a number.  However, under longstanding antitrust law, 5% is the bare minimum price raise that can support a finding of market power, which the District Court did find here.  And the trial briefs and opinion do give some numbers, particularly the 20-24% industry-wide rate increases from 1997-2009. 

A real-world test case in another country suggests that the number could be much higher.  Australia’s central bank forced card issuers to eliminate their anti-steering rules between 2003 and 2005.  Amex’s average merchant fee rate fell from over 2.5% prior to the rule changes to less than 1.5% afterwards – a drop of over 40%.  Retailers with presences in both Australia and the U.S. filed an amicus brief with the Supreme Court discussing this experience at length.  It is possible that the Australian credit card market was different than the U.S. in some material way, or that the 2003-05 credit card market was different than today’s in some material way.  The most obvious possible way is that, Amex’s rates in Australia in 2002 were probably higher than peers’ rates by an amount that is more than Amex’s rates in the U.S. in 2018 are higher than peers’ rates.  That said, another amicus brief from another Australia group supporting Amex had the chance to argue for such differences and did not do so, even though it highlighted differences that were relevant to other arguments.   (Links to the briefs are above.)

Taking all the (frustratingly scant) data together, and with an enormous range of reasonably possible outcomes (5% to 40%), our best guess is that the fee drop should be at least 10% and probably more like 20% (e.g., if the average U.S. discount rate is 2.4%, it will drop to 1.9%).

The decrease in Amex’s annual discount revenue would be $1.5 billion with a 10% rate drop, $3 billion with a 20% drop, and $6 billion with a 40% drop.  This is a pure price cut at 100% decremental margin and represents a theoretical drop in Amex’s pretax profit of ~20% with a 10% rate drop, ~40% with a 20% rate drop, and ~75% with a 40% rate drop.  The actual effect will be less, and potentially much less, because Amex and its peers will likely respond by raising other merchant fees, raising fees to cardmembers, and reducing cardmember benefits.  (If they do so, they would likely lose volume to other payment methods, thus undermining the mitigation.)  Whatever scenario one chooses as a base case, it is not hard to arrive at a value hit that is greater than 10% of profits.  One such scenario is a 10% rate drop with 50% of lost profits recouped through benefits cuts or other fee increases and not lost again to lower charge volumes.  Another scenario is a 20% rate drop with 75% of the lost profits recouped.

Amex’s private litigation liability is likely as large as the lost profits.  A final judgment of an antitrust violation in the government case will trigger “collateral estoppel” against Amex in follow-on antitrust lawsuits filed by merchants.  The final judgment in the government case will legally establish Amex’s antitrust liability to merchants; the only remaining issue in the follow-on merchant lawsuits will be the amount of damages for each merchant.  Under the relevant antitrust statutes, merchants will be entitled to treble damages, for 12 years of overcharges (four years prior to the governments filing suit, or 2006, through the NDPs’ removal in late 2018), plus prejudgment interest, plus attorneys’ fees and costs.  If every single merchant filed suit, and the merchant fee overcharge were proven to be 20%, Amex’s theoretical total liability would be:

·        $18 billion – Amex’s average annual discount revenue for the last 12 years

·        x 67% -- average U.S. share of total card-billed business (prior years were higher than the current 65%)

·        = $12 billion – average annual U.S. discount revenue

·        x 12 – years of damages recoverable

·        = $144 billion – total merchant fees at issue

·        x 20% – assumed amount of overcharge

·        = $28.8 billion – total base damages

·        x 3 – treble damages

·        = $86.4 billion – total theoretical damages, plus prejudgment interest & costs

The theoretically liability is half that much if the merchant fee overcharge is found to be 10%, and double that if the overcharge is 40%.

Amex will not actually pay anywhere near this amount, because not every merchant will sue, and Amex will be able to settle the lawsuits for less than the full damages amount.  A class action lawsuit cannot be used, each merchant will need to prove its damages amount in arbitration (as required by Amex’s merchant agreements), and doing so will be heavily disputed, fact-intensive, and require sophisticated damages models created by highly-paid economics experts.  For very small merchants, the litigation costs and management time will not be worth it, even though they can theoretically recover the monetary litigation costs from Amex after winning in arbitration.  For very large merchants, the litigation costs will be relatively trivial, and they will all sue or extract a settlement without suing.  For in-between merchants, they can reduce their individual costs by grouping together, hiring the same lawyers and expert witnesses, using the same damages models, etc. 

These groupings are already occurring, fostered by merchants’ trade groups and enterprising law firms.  Indeed, some merchants filed suit soon after the District Court’s decision in 2015; those cases were put on hold after the Second Circuit reversed and will be revived if the Supreme Court reverses again.  An easy way to observe the merchant interest already expressed is to examine the companies and trade groups that joined amicus briefs in the Supreme Court.  Here are the individual merchants, including 6 of America’s 7 largest retailers by sales volume and 12 of the top 20 (in bold).  Large travel & entertainment businesses like Southwest Airlines and United Airlines, also on the list, have lower revenues than the large retailers but have a much higher portion of their revenues paid with credit cards and a higher portion of Amex cards among their credit card payments.

  • Ahold
  • Albertsons
  • AutoNation
  • Bally Total Fitness
  • Bi-Lo
  • Big Sur Waterbeds
  • BJ’s Restaurants
  • Bridgestone Americas
  • Caleres
  • Crestline Hotels & Resorts
  • CVS
  • Denver Mattress
  • Festival Fun Parks (Palace Entertainment)
  • The Great Atlantic & Pacific Tea Company (A&P grocery)
  • H.E. Butt Grocery
  • Home Depot
  • Hy Vee
  • Ingram Micro
  • Innovative Dining Group
  • Jack in the Box
  • Jo-Ann Stores
  • Kroger
  • Luxottica
  • Meijer
  • Office Depot
  • OfficeMax
  • Public Storage
  • Publix
  • Rite Aid
  • Safeway
  • Sears Holdings
  • Sofa Mart
  • Southwest Airlines
  • Staples
  • Supervalu
  • Target
  • TKC Holdings
  • United Airlines
  • Walgreen
  • Wal-Mart

Additional amicus briefs were filed by the following trade associations:

·        The Retail Litigation Center has 23 large-retailer members and exists specifically to facilitate litigation opportunities such as this one on behalf of its members

·        The National Retail Federation is the world’s largest retail trade association

·        The National Association of Convenience Stores has 2,200 retail members and 1,800 supplier members with 154,000 U.S. stores; its members have $550 billion in sales

·        The Food Marketing Institute represents the food retail industry; its members have 33,000 retail food stores, 12,000 pharmacies, and $800 billion in sales

·        The National Grocers Association represents independent (non-large-chain) grocers with $131 billion in sales

·        The National Association of Shell Marketers represents 140 wholesale distributors of Shell-branded petroleum products with thousands of retail locations

·        The Retail Industry Leaders Association represents large retailers with $1.5 trillion in sales and over 100,000 stores

·        The International Air Transport Association represents 282 member airlines covering 84% of global air traffic

The top 100 U.S. retailers account for around 40% of U.S. retail revenues.  Most of travel and entertainment spending (which is very Amex-heavy) is more concentrated than that.  (Think about the number of airlines, hotel chains, and rental car companies.)  A rough-cut first estimate is that merchants accounting for around half of Amex’s U.S. spend will sue or extract a settlement without suing.

The settlement discount is just as hard to estimate as the percent of overcharge.  With liability already decided but with damages amounts that will be highly disputed and have a wide range of reasonably possible outcomes for each merchant, this follow-on litigation is unique and difficult to compare to settlement discounts actually agreed upon in earlier lawsuits.  For the purposes of this trade, it is sufficient to choose the easiest, roundest number and assume a settlement for half of the “expected” outcome.  (For example, perhaps Amex will argue that damages are 1x, a merchant will argue that damages are 5x, a neutral observer would conclude the most likely award by the arbitrators is 3x, and the assumed settlement amount will be 50% of 3x, or 1.5x.)

Using these figures, from a starting point of an $86.4 billion potential liability (ignoring interest & costs), Amex will face claims from merchants for perhaps half of that, or $43.2 billion, and will settle for half of that, or $21.6 billion.  Most payouts would probably not occur for 2-3 years because Amex can wait until the arbitrations are near completion, so a time-discounted value may be more like $16 billion (~16% of market cap).

The direct payout will reduce Amex’s business value in an additional, indirect way: The reduction in Amex’s capital base will reduce its capital ratios, reduce the available size of its loan book, and therefore reduce its net interest income.  That effect may be worth another few percentage points of market cap by itself.

Putting the two pieces together, Amex faces a loss of ~10-20% of its value from lower merchant fees and ~10-20% of its value from legal liability plus the resulting capital constraints, or 20-40% total.



Apart from the obvious risk of legal lightning striking and the Supreme Court ruling for Amex, three other risks could render the overall trade less profitable or turn it into a loss.  First, the market could already be pricing in the risk of loss.  It appears that the market is not doing so at all.  The case has received minimal coverage in the financial press or broader press, apart from a few general-interest stories on the two days immediately following the granting of certiorari and the oral argument.  None of that press discussed the likelihood that Amex would lose or the potentially large financial consequences of a loss.  The topic was not discussed on Amex’s last three earnings calls or its March 8 analyst day.  (They did include a generic disclaimer that their guidance “assumes no material changes in regulatory environment.”)  Few sell-side analyst reports have discussed the issue until this week.  Also, since just before certiorari was granted, AXP’s stock has risen slightly more than the XLF financials ETF and in-line with the S&P 500 and Capital One (a credit-card stock with no direct exposure to the Amex issues).  However, I do not read most sell-side reports, and it is possible I have missed some bearish discussions.  AXP’s stock price rose 8% on the day that it released its strong 1Q results; without that day, the stock would have underperformed.  It is possible that the buy side is more informed than the public information would suggest and that AXP’s current price would have been materially higher absent the litigation-related risk.

Second, the stock could rise materially for reasons other than this legal situation.  Amex could report surprisingly strong 2Q results, it could raise guidance, the entire stock market could rip higher, or financial services stocks could outperform based on macroeconomic changes and overall investor sentiment.

Third, it may take more time than expected for the stock to drop.  Investors may still not see the coming value destruction after the Supreme Court issues its decision.  Investors might see the risk but place hope in another Second-Circuit “save”; they might not price in the risk until the Second Circuit and District Court dispose of the case on remand, even though the true odds of another save are very low.  If the stock decline gets strung out, the ROI falls and the risk of a price rise from other factors increases.  I expect, however, that a Supreme Court reversal will receive significantly more attention from the press and investors than did the grant of certiorari or the oral argument.  At that point, sell-side analysts will be failing in their jobs if they do not estimate and publish what the decision should mean for Amex’s stock value -- including the private litigation payouts.


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.


  • Supreme Court decision within two weeks
  • Press coverage and analyst coverage of that decision and its implications (both for ongoing profits and liability to merchants)
  • Further litigation events in the lower courts during 3Q and 4Q
    show   sort by    
      Back to top