Quick Thoughts – AXP: Releasing the Distribution Bottleneck will Lift ROE
Our thesis is that in 2016 Amex will raise its ROE target (currently 25% or more) as the Global Network and Merchant Services or GNMS business grows relative to the proprietary issuing business. Presently GNMS, which includes earnings from network processing and merchant relationships but not cardholder relationships, accounts for 25% of total segment earnings and generates a reported ROE of over 70% (vs. 36% for the more balance-sheet intensive business where AXP acts as issuer); in practice, these ROEs are likely overstated because of the substantial unallocated equity but, adjusting for this, generates estimates of 50% and 25% respectively.
The long-standing challenge for AXP is that distribution constraints imposed by V and MA (and specifically the exclusionary rules which operated until 2004 and prevented banks issuing V or MA branded cards from also issuing Amex-branded cards) have subjected growth of the high-return GNMS business to the distribution and balance-sheet constraints faced by the issuer business. Even today, a decade after the legal system struck down the exclusionary rules, the legacy of the distribution bottleneck faced by AXP is reflected in its consumer penetration: AXP has 53mm cards-in-force in the US vs. 254mm for Visa and 178mm for MA (685mm and 336mm if we include debit/prepaid cards as reported by Nilson, issue 1034).
However, the bottleneck is being released. Amex has grown “GNS” (as opposed to GNMS) volumes where it provides network and merchant services but where the card is issued by a third-party at a CAGR of 20% since 2004; 2013 GNS volume of $140bn represents 15% of the total. The overseas portion of this volume at $117bn is nearly as large as the volume from AXP’s international card services business (where AXP issues cards to international consumers). In the US, however, GNS volumes have lagged and at $24bn (as reported by Nilson, Issue 1033) represent less than 5% of the US card services business (where AXP issues cards to domestic consumers).
If AXP is to continue to grow GNS volumes at 20%, it needs to build the US business and persuade banks to issue in scale on the Amex-branded network for the intrinsic business case rather than in token amount (there are just 10mm Amex-branded cards issued by third-parties) to mitigate anti-trust risk. WFC’s decision to make Amex-branded cards (launched last week) an integral part of its strategy to build the relevance of its card business towards that of its deposit franchise is therefore a watershed event.
Balance-sheet efficient growth in US network volumes will also be driven by Amex’s Serve product which is positioned as a low-cost alternative to a traditional bank account and takes advantage of exemption from the Durbin cap on debit interchange. Chase Liquid, for example, had to remove online bill-pay capabilities to avoid the Durbin cap but Amex can offer this feature (and, indeed, check access) without tripping Durbin. While Serve volumes are relatively low, annualizing at around $5bn, they are growing rapidly as a result of distribution through non-traditional (i.e. non-bank) channels including retailers such as WMT (where Serve is branded as Bluebird) and carriers (who, as part of activating new ‘phones, will enable the Isis wallet which means – for customers who do not have an account with either of Isis’ two bank partners, Chase and WFC – opening a Serve account).
By end-2015, we expect Serve to be generating at least the same US network volume US as Amex’s third-party credit card business does today, and note that Dan Schulman (who heads the Enterprise Growth unit of Amex which includes Serve and is reported as part of Corporate/Other) reported yesterday that consumers have loaded more than $1.9bn on Serve this year through April which is almost three-times the equivalent year-ago figure and is a leading indicator for purchase volume. Indeed, there are already 9.4mm Serve cards-in-force (with 2.4mm issued this year through April) compared with the 10mm Amex-branded credit cards issued in the US by third-parties.
Segment Results Reported by AXP