What should Capital One do with Discover's networks?

Partnering is better than going it alone

Key insights in the post

• The Capital One/Discover merger seems more likely than not at this point

  • Both boards approved the transaction

  • The regulatory environment has become more favorable

  • Discover estimates the deal will close in May

• The card issuing side of the merger is straightforward, but the network side faces challenges

  • How to close the global acceptance & scale gaps in Discover Network?

  • How to take advantage of the Pulse PIN-debit network?

• The conventional wisdom is that Capital One will remediate any gaps in the networks and convert portfolios when the networks are at parity with Visa & MasterCard

• I respectfully suggest that Capital One can accelerate its strategic goals with partnering strategies

  • Partner with a global incumbent for non-US transactions

    • Instant, ubiquitous global reach

    • Significantly less complexity

    • Motivated partners

  • Federate the networks with major US issuers to build domestic scale

    • Partner with JPM and BAC who have both aspired to own networks in past

    • Share only the technical infrastructure, with each issuer setting their own interchange — this gets around historical antitrust concerns

    • Improves scale economies quickly as these issuers represent 30%+ of card issuing although cobrands could be a challenge

    • Capital One has to set its own interchange with or without partners

• Together these two approaches solve the key challenges faster than going it alone without giving up any meaningful advantage

Introduction

As many of you know, I am a skeptic about most industry developments. But I am not at all skeptical of Capital One’s pending Discover acquisition. Scale benefits aside, Capital One is best-in-class in credit card lending and can no doubt increase value in the Discover portfolio. The deal is a masterstroke from this perspective. Another benefit may be using the Discover brand to target the middle segment that is left out of Capital One’s famous barbell strategy.

But the networks are a wild card. Discover is the fourth biggest US credit card network, with a patchy global footprint. Neither Discover network nor Pulse network has material signature debit volume and Pulse is disadvantaged in its core PIN market. It is unknown whether banks that issue on Pulse will continue to do so under Capital One ownership.

The networks have scarcity value. Both JPM and BofA have publicly coveted their own networks in the past – and JPM tries with ChaseNet. All issuers have uneasiness about Visa’s expansion strategies, market power, and economic rents – even as they benefit from Visa interchange strategies and global footprint. Love-hate describes it pretty well for Tier I issuers. Tier II and below are more on the “love” side even though they pay for it all.

All public comments I have seen suggest Capital One intends to keep the networks proprietary and remediate any performance gaps itself. But is that the right approach? I have put off covering this topic given the regulatory uncertainty about the merger, but now that both boards have approved the transaction, I feel more confident the transaction will close.

This post is pure speculation, but I think there may be attractive alternatives to the full proprietary strategy. I will describe these in my usual excessive detail.

How are Discover Network and Pulse Network positioned today

Both networks are among the smallest within their orbit; but, networking is a scale business; Capital One’s volumes only close some of this scale gap.

Discover Network

Discover Network has about 5% share of US credit card network switching. The only material issuer they support is Discover itself. They also provide access for non-US card brands that need a US acceptance footprint – JCB (Japan) is the primary example. These initiatives don’t add that much volume.

The network has near universal US acceptance by outsourcing virtually all acquiring. Discover directly acquires only for the top 100 or so merchants. Like Amex & MasterCard, it is not accepted at Costco, which is its only material merchant gap.

For global acceptance, Discover partners with domestic networks around the globe —for example, with Interac in Canada and Prosa in Mexico. This patchwork has gaps and visibility issues. Discover Cards don’t really appeal to the globe-trotting set, so this has not been a massive issue, but it is a downside. For example, Discover does not have Corporate T&E cards as Amex does.

As a three-party network, Discover doesn’t make much of a distinction between network fees and interchange. They have a Network Authorization Fee (NAF) of 0.25¢ compared to Visa’s 14bps + 2¢ assessment fee (for Credit). Discover interchange is roughly comparable to Visa.

In summary, domestic acceptance is at parity, interchange is comparable, network fees are lower, and global acceptance lags. That is the hand Capital One is dealt.

Pulse Network

Pulse is a PIN debit network. It’s roots are serving Texas community banks, but it now operates nationally. PIN debit networks mostly started as associations and only later became units of public companies and Pulse is no exception.

The Durbin amendment helped Tier II PIN debit networks. Durbin mandates every debit card have two unaffiliated networks so that merchants can benefit from network competition. That gives independent PIN networks some presence on debit cards

PIN debit network fees have always been lower than Signature Debit fees. Post-Durbin, PIN fees have been pressured by big merchants and big acquirers. Where V/MC Signature networks charge 13bps + 1.5¢ per transaction the merchant, PIN debit switch fees vary from 2-5¢ per transaction to the merchant and a similar amount to the issuing bank. Those are published rates, but they are negotiated down at larger merchants, acquirers and banks:

• The biggest merchants and acquirers may receive volume-based incentives, making the net fee lower

• The biggest banks don’t pay their side of the switch fee. The PIN networks need their cards to carry the brand, so they heavily discount to get the unaffiliated slot

Pulse is among the 5 biggest PIN networks, but with the worst distribution:

• Interlink is the share leader as a subsidiary of Visa. To understand why, read the DOJ debit routing lawsuit and my post on the subject. The quick explanation is that Visa uses its leading position in Signature Debit to advantage Interlink in PIN Debit. Visa has ~80% share of Signature Debit

• Maestro is a distant second because MasterCard has a much smaller share of the Signature market (20%). Otherwise Maestro benefits from the same tactics as Interlink

• STAR & Accel are Fiserv units. STAR was the First Data network and Accel was the pre-merger Fiserv network. Fiserv cross-sells STAR & Accel to its debit processing and core processing clients. Most small banks take the PIN network as part of the core processing bundle and regional banks take it as part of a debit processing deal

• NYCE is an FIS unit. NYCE has a similar role at FIS that STAR/Accel have at Fiserv – it is cross-sold as part of a core processing bundle. FIS also sells it independently to some regional banks

In contrast, PULSE never had a core processing business nor a meaningful Signature Debit product. It’s does debit processing for some small banks, but it is more dependent than the others on independent sales to bigger banks. Pulse and the other independents have fewer levers to win merchant routing as switch fee discounting is their only tool.

PIN debit is rarely used for cross-border transactions, so global footprint is less important, but where it matters, Visa/Interlink and MasterCard/Maestro are advantaged.

Advantages of network ownership

I heard several early claims about owning networks but I only find one convincing.

Convincing: Exempt interchange

The Durbin amendment not only introduced debit routing competition, but also an Interchange cap on larger banks (>$10B in assets). As a result, smaller, “exempt” banks can earn almost 2x as much interchange as capped banks for similar transactions. One loophole in Durbin is that 3-party networks are also exempt from the cap. Both Discover & American Express launched bank accounts with exempt debit cards as a result. But neither offering generated much volume.

One attraction to C1 of getting both Discover & Pulse is the prospect of earning exempt interchange on the banks debit cards. That would improve their competitiveness for low-balance, debit-centric accounts, similar to those offered by Neobanks (see my post on Neobanks for more detail). Capital One could target the high end of debit-centric customers and then convert the best of them to credit card users when they qualify. Effectively, good debit behavior creates a pipeline of good credit card prospects.

Exempt status is likely a key benefit driving Capital One’s decision-making.

Unconvincing: Network Data

I am a broken record on this topic, but Capital One gets limited extra insight from owning a network than it has as an issuer – particularly if it is the sole issuer on that network. Networks see exactly the same ISO 8583 data record that issuers see, so the incremental value would only come from seeing multiple issuers’ data together. But if Capital One were doing anything special with that data, no other issuer would use their network. Catch 22.

The key is that Networks see no more or less data on any transaction than an issuer sees or an acquirer sees. And none of those parties gets the SKU-level data that would be truly valuable. Network data helps with fraud prevention, but is not otherwise that useful. Note that none of Visa, MasterCard nor American Express have a material data business based on their network data. Nor does Discover.

Unconvincing: Direct merchant relationships

There was some early talk about Capital One benefiting from Discover’s direct acquiring relationships with the top ~100 merchants. This is unlikely to amount to much.

Discover accounts for under 5% of a typical merchants volume, so it stands a distant fourth in priority for any retailer. Even with Capital One volume it will not exceed 15%. Second, acquiring is rarely an anchor to a meaningful relationship. Third, large retailers don’t really have relationships with their vendors that can lead to things like share of wallet increases or cross-sales – they buy mostly on a best-of-breed and lowest-price-wins basis.

While Discover Network will increase volume with Capital One cards, it will still be the fourth biggest network. The acquiring revenue is also not that attractive at the top 100 as these merchants typically pay a penny or less per transaction. The very biggest merchants are well under a penny. The aggregate revenue is just not that high.

Acquiring contracts are not senior leadership level “relationships” that could lead to unique propositions. Merchants will not willingly share SKU data with an acquirer. They view SKU as their crown jewels to be carefully guarded. With a lot more volume, management engagement would elevate somewhat, but Discover would still be smaller than the three other networks.

What to do with the networks

The conventional wisdom is that Capital One will keep both networks and remediate any gaps. They won’t convert off Visa & MasterCard for their high end cards until Discover & Pulse are at parity.

The highest-end portfolios, like VentureX and Spark will stay on Visa or MC the longest because those products appeal to high-spending cardholders who need seamless global acceptance. Subprime/Near Prime/Low Prime portfolios mostly spend domestically, so they can convert earlier.

The challenge here is that the lower end cards have much less spend per card, therefore, Capital One won’t benefit from the full scale lift initially. Until those high-end cards convert over, Discover Network scale won’t increase enough to shift its merchant bargaining power.

One of the key goals of the merger is unaffected by network gaps: Exempt status. Capital One will likely rebrand its debit cards Discover/Pulse as quickly as possible. But I don’t think it will be quite that simple. Capital One only has about 1% national share of consumer checking accounts, so it is vulnerable to merchant pressure to drop Discover/Pulse for debit – particularly in regions where Capital One has a low checking base. The idea that merchants will simply roll over as Capital One doubles prices is naïve. It might happen, but it is not guaranteed.

Discover network will need time to broaden global acceptance, particularly if most of the cards with cross-border needs don’t convert. American Express has been at it for 50+ years and still has acceptance gaps, particularly in countries where it doesn’t issue cards – and outside the US, Capital One only issues in Canada and the UK.

Another complication is Capital One’s unique approach to network affiliation. They issue on both Visa and MasterCard with a much closer split than most other issuers. That gave them negotiating leverage as they can re-brand origination volume from one network to the other if a network is uncooperative. Most issuers are 80%+ on one network for their issuer-branded portfolios; major cobrands choose network brand themselves.

As Capital One shifts volume to Discover, it will be splitting volume across three networks, and negotiating leverage with the two majors will drop. Even if it keeps all volume on V/MC until Discover is 100% ready, the majors will give it less attention. Capital One is the master at network management, but that is partially because its volumes generally grow faster than other issuers. If its V/MC volumes have an end-date, those tactics won’t work.

So what alternatives are there?

Option 1: Partner with one network for Global Acceptance and exception portfolios

Cross-border transactions account for one-third of net revenue at Visa and MasterCard (revenue less incentive costs). The networks charge an inter-regional fee to switch transactions across borders and capture FX revenues of ~1% as well.

Instead of remediating the Discover network’s global footprint, Capital One could partner with one of the majors to become the non-US acceptance brand on Capital One/Discover Cards. Domestically these cards would always clear on Discover network, but globally they would clear on either Visa or MasterCard.

The prospect of keeping some of the revenue for cross-border transactions should be motivating to the majors. This would allow Capital One to transition high-end domestic volume to Discover Network faster. While there are likely technical issues to conquer, all sides should be motivated toward speed.

As an additional inducement, Capital could convert any residual portfolios to its global partner’s brand; for example, Discover has no Corporate Card program, so Capital One’s portfolio must stay on a major anyway; there may be other examples like this. The winning brand would get the loser brand’s share of non-convertible volume to partially make up for its loss of commodity domestic volume – which should be motivating.

Capital could keep some of the cross-border revenue, either as a revenue share from the partner network or by keeping certain countries proprietary; for example, Canada and Mexico could stay as local partnerships since those countries likely account for a material share of the combined portfolio’s cross-border volume and Discover has strong partners in each:

• In Canada, Discover partners with Interac, the national Debit network, and Capital One does issue credit cards domestically

• In Mexico, Discover partners with Prosa, also the national debit network. Capital One does not issue credit cards in Mexico, so all volume would be cross-border

Whether these corridors are in or out of a network branding deal could be a leverage point for negotiating terms and revenue share.

This becomes a game of chicken between the two majors. Neither will like it one bit, but if they blink, their rival gets the most lucrative part of switching revenue from a big domestic issuer – and loses out on those exception products.

Option 2: Federate the Networks with other big card issuers

The horror you say! Capital One now has proprietary networks and I suggest they share them? Read on. I don’t think I’m crazy, but you decide.

The first objection might be that when the big issuers last owned a network they lost billions in interchange litigation settlements. That eventually led to Visa & MC going public so that interchange was set by an independent entity. But reversing history is not quite what I am suggesting.

I suggest that Capital One spin-out the networks into a consortium with only JPM & BAC. These three account for almost half of the card volumes and about a third of debit. What I propose they share is solely the technical infrastructure: Acceptance, Authorization, Clearing, Settlement, fraud and some rule setting. But crucially, each issuer would set interchange independently.

Credit

The prior litigation succeeded because the network set credit interchange on behalf of its owners. But if each owner sets interchange independently, that cause of action goes away. I am not a lawyer, but that seems logical to me.

Capital One itself is better off on cost and neutral on interchange under this arrangement:

• In the conventional wisdom model Capital One has to negotiate interchange independently anyway; they still do under the federated model

• In the federated model, they get help funding network costs, whether that is for pure domestic volume or a global footprint as well

BofA and JPM are likely to view this model favorably as they have publicly aspired to own a network.

• JPM already has ChaseNet, a subnetwork on Visa. That allowed them to set some interchange themselves. They use ChaseNet when they are both the issuer for a card and the acquirer for a merchant

• Bank of America has publicly toyed with the idea of a network in the past. Like JPM they have a large acquirer that might benefit and they have tried self-settlement in the past

This arrangement should make merchants happy as it is what they have been asking for. It does not increase their integration or operating costs, but it does given them negotiating rights at the individual issuer level. That makes lawsuits less likely.

If the network infrastructure were run as a utility, its fees to merchants could drop from the 14bps + 2¢ that Incumbents charge to just what is needed to fund Opex and Capex. The network doesn’t have to return a profit to its owners, and being transparently run as a utility provides more insulation from anti-trust suits. Amex doesn’t charge an explicit network fee for example.

This is essentially how Zelle and Paze are run as well as TCH, Akoya, NACHA and other successful bank consortia. These three banks are used to managing together as they are among the EWS owner/founder group.

While I would keep the door open to other issuers joining, I don’t think any will do so.

• Citi is the closest call. It has close relationships with large merchants due to its Retail Cards business – bundling PLCC, Cobrand, and Citi-brand pricing in a single deal may plays to their strength. They might even route PLCC over the shared network to reduce costs, but they are also the most global issuer and this consortium only solves for the US

• None of the other issuers have enough share to leave the Visa or MasterCard interchange umbrella. They are more likely to be negotiated down by merchants than the big three. But if they want to join, or just join on the debit side, I would let them in. However, ownership and governance might be reserved for the “group of three” to keep things simple.

The biggest confounder in all this is cobrands. Big cobrands pick a network themselves, so their volume is not in play for re-hosting on the new network. BAC only has a few big cobrands, particularly Alaska Airlines, but as much as half of JPM’s volume is in big cobrands like Amazon, United, Southwest, and Disney. They are publicly in the running for the Apple Card. Even Capital One has a few meaningful cobrands that will need to stay on the incumbents. Those incumbents pay the cobrands handsomely for the branding, so conversion is unlikely. This problem can be dealt with over time, but in the short run it will require some residual relationship with the incumbents at much poorer economics.

Another cobrand point worth making is that while JPM & BAC dwarf Capital One in total assets, their card portfolios are of similar size after excluding cobrands. BAC is the smallest in aggregate but has the fewest cobrands. JPM is the biggest in aggregate, but up to half their volume is cobrand. Capital One with Discover may actually have the most issuer-branded volume — they have cobrands, but none are enormous, particularly after disengaging from Walmart. So from an issuer-branded cards perspective, the three partners would be relatively close in size, balancing the power within the consortium.

The consortium has no impact on acquirers as they already serve Discover Network and would continue to do so. However, their volumes would shift from 5% Discover to 30%+ Discover.

Debit

The key regulatory question is whether this structure counts as a three-party network for Durbin purposes. All the major pricing contracts would be directly with each bank. The network itself just provides infrastructure.

If it does count as 3 three-party networks instead of 1 four-party network, the owners could negotiate debit interchange with the merchants directly. They could certainly anchor on the Durbin Cap, but might be able to charge more in regions where they have density. The likelihood that a merchant drops JPM or BAC debit volume is low given they each have up to 15% debit share and, in some regions, as much as 30% each. If the owners drop Interlink/Maestro and add Pulse instead, Visa’s market power in Debit drops substantially – which should make the DOJ and the merchants very happy — giving merchants an incentive to be flexible.

If the consortium only counts as a four-party network, Capital One would not get a unique premium above the regulated cap, but my sense is that they may not get that anyway given their low share – merchants can simply turn off Pulse to pressure rates without much consequence on consumers.

If the consortium can count as 3 three-party networks, sharing also wouldn’t reduce Capital One’s advantage much in banking the debit-centric population. JPM & BAC do not actively seek out high-risk credit segments and don’t actively pursue low-balance checking accounts. Capital One may have this segment to itself anyway.

And finally, JPM & BAC would have to buy in to network ownership. Given the scarcity value, those investments would offset Capital One’s acquisition cost to some degree.

Both options together is better than each one alone

Option 2, which addresses scale, works best if combined with Option 1, which addresses global acceptance.

Conclusion

I am fully prepared for a knowledgeable anti-trust lawyer or compliance expert to school me on how wrong-headed this all is. I would not push back. I likely haven’t identified all the holes in my logic.

But networks are a scale business and the conventional wisdom approach does not build scale fast enough to achieve what Capital One actually wants. Only by partnering can they accelerate the end-game. Option 1 addresses the global footprint problem and Option 2 addresses the domestic scale problem. Together they create a meaningful competing network for the first time since Discover first emerged in the early 1980s.