Sure. But the default and reasonable presumption is that at that scale, companies do things because they're profitable; in this case, they implement rewards programs because they make more money by doing so. In particular, the default and reasonable presumption is that companies don't spend more money acquiring a class of customers than they expect to make from those customers. The claim by the Atlantic article is that credit card companies lose money on the rewards programs for rich people. There's no evidence of that, and some evidence to the contrary.

> Without more detailed data it’s not possible to rule out the possibility that points programs are a subsidy for the rich.

That's not where the burden of proof lies. Without more detailed data it's not possible to demonstrate that rewards programs are a subsidy. The conclusion is not presumptively valid.

I'm not suggesting, either, that the argument from the linked Twitter thread is ironclad, just that it's compelling evidence against.

> The second assumption he makes is that volume is the same across the board. It’s not, there’s way more customers on the lower end, and the company charges them even more fees. So sure, rich customers individually pay more interchange fees, but the company could still be making most of its money from poor customers.

"more customers on the lower end" is in fact refuted by the data; see figure 2 (F) in the linked paper.

"making most of its money from poor customers" might be true, but note that figure 2 (E) is scaled by "percentage of average daily balance", and if you were to view it in absolute terms, it seems less likely to be true. Even if it were true, though, it's not evidence of a subsidy of rich customers by poor customers. The credit card companies very obviously think that rich customers are profitable to acquire and keep, or they wouldn't pursue them so heavily. The evidence from the thread supports that claim.

The argument being made in the thread is, precisely, if the rewards program pays for itself an individual customer, that customer is not being subsidized in any way.

If you want to claim that the rewards programs offered to richer customers aren't in fact profitable for the credit card companies, that seems like an even more extraordinary claim, and not one that seems to have supporting evidence.

Alternatively, if you want to define "subsidy" so broadly that any business making more money from one group of customers than another is having one group of customers subsidize the other, even if they are making money from both, then I'd question your definitions and use of evocative terminology. They're spending money to get a group of customers, and if they're spending more to get those customers than they make from those customers, they wouldn't spend it in the first place; if they're spending less to get those customers than they make from those customers, then those customers are not being subsidized.