What's actually stopping US interchange from collapsing like it did in Europe?

You're right that the empirical case for 2% as a functional minimum is thin, but I'd push back on the framing slightly. The EU comparison works as a thought experiment only if you're comfortable ignoring what actually changed in the mechanics of how payments moved money around. The networks didn't crater in Europe because they adjusted what they *do*, not because the cap was harmless. Issuer rewards collapsed—that's not a side effect, that's the main mechanism by which interchange funds the system. In the US, that slack got picked up by other revenue streams: annual fees, balance transfer fees, premium card programs with higher annual fees. Banks found ways to extract the same economics from richer customers and let everyone else downgrade to commodity cards. The EU cap worked partly because EU banks tolerated lower overall margins on payment products. American issuers... I'd be skeptical they'd accept that. We're not a market with patience for lower margins. The real question you're asking—can *US* networks function at EU rates given US banking structure and consumer behavior—is actually different from "did European networks function." And I don't know if anyone's seriously modeled that. You'd need to account for how much of the 2% actually funds things that have become expectations here (fraud prevention, speed, the security theater), versus what's just economic rent. The pass-through question is interesting but backwards—knowing whether merchants can absorb lower interchange doesn't tell you what the issuers will actually do with the revenue pressure. That's a choice, not a constraint.

What I'm actually curious about is whether anyone's traced what happened to the *training and technical infrastructure* on the issuer side after the EU cap. Because this is where I think people miss the labor dimension of a supposedly technical problem. When you compress margins that hard, someone's budget gets cut. In the US, that's often the operations and training budget for bank employees—the people who actually understand fraud patterns, who know how to configure risk systems, who onboard new staff into the actual mechanics of payment processing. I watched something similar happen in electrical apprenticeships when contractors got squeezed on labor margins; the knowledge work didn't disappear, but it got distributed to fewer people, often less formally. The work still got done, but slower and with more institutional fragility. Did European payment processors just absorb those margin cuts cleanly, or did they shed the equivalent of institutional knowledge? If they shed it, how did they avoid the fraud and stability problems that *should* have followed? That seems like the actual empirical question worth asking—not just whether transaction volumes held up, but how. Did they lean on automation that the US hadn't yet built? Did they accept higher fraud costs as a trade-off? Did someone's training budget just evaporate and we never noticed because the systems still technically worked?

You're pointing at something real but maybe naming it wrong. The distinction isn't whether networks *need* 2% to function—they clearly don't, the EU proved that. The distinction is between what's economically necessary and what's politically extractable, and those are different animals. Here's the thing I'd push on: "antitrust enforcement is weak" treats this like a bug in the system when it might be closer to a feature. US merchants have fragmented leverage. A bodega in Milwaukee can't organize with a thousand other small shops the way European grocers coordinated political pressure. So you get individual merchants trying to absorb or pass through costs, and the networks stay insulated from collective action. The EU cap worked because the merchant lobby had actual parliament access and could make this a trade issue, not just a pricing problem. That's not really about whether networks *can* function at lower rates—it's about who has standing to force the question. The empirical gap you're naming is real though. Nobody's seriously mapped cardholder behavior shifts in the EU post-cap, or traced how issuer-side costs propagated versus got absorbed. That would matter for an honest US conversation. But I suspect the reason you're not finding good work on that is because the question assumes a genuinely neutral institutional space where we're just measuring physics, when actually the whole framing—what gets studied, what gets funded, what counts as a "crisis"—is already shaped by who has power to define the problem.

The problem with the Europe comparison is it doesn't tell you what you think it tells you. European payment infrastructure didn't collapse, sure. But you're comparing systems that were already structurally different when the cap hit. European issuers had been operating on thinner margins for years before 2015. They'd already consolidated more aggressively. Rewards programs were never as central to their competitive model. When the cap came down, yeah, some rewards got stripped back and cardholder acquisition costs went up—but they were adjusting around an existing equilibrium that wasn't built on 2% interchange subsidizing consumer behavior. US issuers, by contrast, have spent fifteen years building entire business units around churning cards to people with rewards value propositions. A sudden cap here wouldn't just reduce revenue; it would demolish pricing structures a lot of institutions bet their franchise on. That's not the same as saying the 2% is *economically necessary* for the network to function. You're right that the better question is whether it's necessary to maintain the *current competitive strategy*, and those aren't the same thing. But the EU evidence mostly just shows that you can run payment rails at 0.3% if you build your business around 0.3% from the start. It doesn't actually test whether you can get there from 2% without a decade of visible disruption and bank consolidation that American antitrust law would probably catch anyway. The reason nobody's cited hard work on pass-through is probably because the answer is: it depends entirely on the transition path, and that's not a satisfying policy argument.