This week, I read an essay of mine about how credit card rewards programs are financed out of interchange revenue, and offer substantial live commentary. This comes hot on the heels of an article in the Atlantic claiming that rewards are subsidized by poor people. This view has a lot of currency among advocates and the media. It is false.

In addition to paying for rewards, interchange pays for many things businesses find valuable (recordkeeping, automation of business processes, etc), and some things society should find valuable (not needing to send poor people to jail over payments failures).

[Patrick notes: As always, I have enhanced the transcript with notes added after the fact, set out in this fashion.

To visually distinguish commentary in this episode from the published essay that I do a reading from, I will distinguish the essay like this:

This would be a hypothetical quote from the essay.

Quotes from other people are announced in text and italicized when they are paragraph-length or longer.

]

Sponsor: Vanta 

Vanta automates security compliance and builds trust, helping companies streamline ISO, SOC 2, and AI framework certifications. Learn more at https://vanta.com/complex

Timestamps

(00:00) Intro
(00:31) A recent Atlantic article's false thesis
(09:35) Rebating interchange to earn share of wallet
(14:48) Sponsor: Vanta
(16:56) Why isn’t every card a rewards card?
(27:56) The complexity spectrum of rewards products
(29:55) A fun rabbit hole about credit card acceptance
(33:53) Back to more complicated cases
(45:45) Further refinements in cat and mouse games
(51:24) Giving the customer more choices more frequently
(58:10) More directions to go in
(01:01:54) Patrick’s counterproposal

Transcript

Hideho everybody. My name is Patrick McKenzie, better known as Patio11 on the Internet. Last week on Complex Systems, we had an episode with my guest Gary Leff about the economics of airline mileage programs, which are tightly tied into their card issuing programs at partner banks.

By coincidence, almost the exact day that the episode dropped, there was an article in The Atlantic titled There Are Two Kinds of Credit Cards, subtitled "Yet Another Way the Poor Are Subsidizing the Rich" by Annie Lowrey. I'm going to quote verbatim a paragraph that I think is the thesis statement of this article:

High costs are weighing down working-class families, while driving big rewards to rich ones. Over the past few decades, the credit card market has quietly transformed into two credit card markets: one offering generous benefits to wealthy Americans, the other offering expensive debt to the poor, with the latter subsidizing the former. While balances are compounding at the highest average APR in decades, a brutal 21.5%, the haves are not just pulling away from the have-nots. The people swiping their cards to pay for food and gas are also paying for wealthy cardholders' upgrades to business class.

This is a very appealing thesis for many people. It is clearly articulated in the article, and Annie Lowrey has both arguments and some level of evidence to back up those arguments. It's very important to say this thesis is false. I'm saying that flat out - this is not the way the world works.

I should note that this essay ran in their "Ideas" sub-vertical, which I believe is what they call opinion pieces. So I'm not fully punching The Atlantic in the face, but I had somewhat higher expectations that they would fact-check even their opinion pieces when they made many false claims. Be that as it may, we've seen this thesis in a number of other places too. The New York Times ran a piece back in 2023, Vox ran one in 2021. [Patrick notes: I think by the standards of high-status journals of news and opinion, being able to find three comprehensively wrong articles with the same core thesis in peer or near-peer places is an argument you can use with your editor.] 

It's just not the way the world works.

Why isn't it the way the world works? Well, it's making a claim about the answer to a math problem where A is greater than B, and you can just math out this math problem. And A is in fact not greater than B. Indeed, the banks have actually mathed out that math problem and there is sufficient information elsewhere in the world, including publicly, to do the same math, come to the same conclusion, and conclusively disprove this thesis.

I think audio formats are the worst possible way to have a debate about mathematical formulas, so I won't force you to listen to that at the moment. I'll write about it somewhere where you can see links to embedded sources and graphs. But I would like to read an essay that I wrote previously about how credit card reward programs actually work, how they're connected to interchange, and how that - and not the interest charged to lower income customers - is the primary thing that drives the availability of rewards to customers in the United States.

The anatomy of a credit card rewards program 

(I originally published this in Bits about Money in March of 2024.)

There are only a few opportunities to make a living by getting good at creating tables to facilitate high frequency math that end users will find entertaining, but will have predictable statistical properties at scale. One of them is designing role playing games. That seems like an interesting topic, but someone else will have to write about it in this column.

The Dragon sleeps on a hoard of interchange revenue. You slay him to get credit card reward points and the card issuer running the game merely chuckles at players' misconception that they are Dragons. No, silly. They're much realer and much richer.

A disclaimer off the top. I used to work at and I'm still an advisor of Stripe. A major portion of the Stripe economic model is charging businesses money to take payments on credit cards. Stripe's two largest costs are paying smart people and paying interchange. And of the two, one would feel a lot better to cut.

[Patrick notes: Many people who read the Atlantic article, and then my Twitter thread, made comments like: “Of course he is in favor of interchange! He worked for a payment processor!” Which is not just the usual level of intellectually bankrupt it is to assume that someone’s financial incentives dictate all their preferences. It is a sign error.] 

Another disclaimer due to longstanding practice, I am homeopathically exposed to the common equity of financial services companies that my family uses so that I can call up investor relations if I ever need to escalate a routine banking issue. My family's main U.S. bank happens to be Chase, which is mentioned in this essay. 

Almost everyone writing about credit cards on the Internet receive some sort of spiff if you sign up after clicking through tag links in their material. This is not my business model. People pay me to write about financial infrastructure. This is probably something you want to be cognizant of every time you read about credit cards online.

As an aside, obviously, Complex Systems has advertisers. Those advertisers are clearly disclosed. None of them have a credit card issuing business as far as I'm aware. But I would mention that here if they did. [Patrick notes: Out of maximum scrupulousness: I recently learned that a future advertiser is a) a future advertiser and b) has a credit card issuing business.]

And The Atlantic has one citation in the piece, which they use as evidence for the claim that rich consumers are rebated more than the direct cost of interchange to those rich consumers. That citation is to a gentleman who works at Bankrate. I'll quote two sentences of the Atlantic article: 

The industry argues that swipe fees not interest in penalties on poor customers, finance its cashback and travel programs. Ted Rossman of Bankrate, the financial data and news site told me, but he does not quote, 'totally buy that' end quote. It's all quote, 'part of the calculation' end quote.

Back to the voice of me. I would not have described Bankrate as a financial data and news site. I would've described Bankrate as an affiliate site, which markets financial products. I'm going to quote verbatim from a disclaimer that is written on the bottom of bankrate.com:

Bankrate.com is an independent, advertising-supported publisher and comparison service. We are compensated in exchange for placement of sponsored products and services, or by you clicking on certain links posted on our site. Therefore, this compensation may impact how, where and in what order products appear within listing categories, except where prohibited by law for our mortgage, home equity and other home lending products. Other factors, such as our own proprietary website rules and whether a product is offered in your area or at your self-selected credit score range, can also impact how and where products appear on this site. While we strive to provide a wide range of offers, Bankrate does not include information about every financial or credit product or service.

When I say that Bankrate is an affiliate, the primary way that Bankrate is compensated, I say based on general industry information rather than having specific knowledge of the contracts involved in this case, is that they are paid on what's called a CPA model, cost per acquisition. This means that a credit card issuer, such as one of the largest banks in the country, has a review of their credit card published on Bankrate. That review might be heavily informed by material that that credit card issuer's marketing department gave to Bankrate, but will come through in the voice of Bankrate as editorial content. And there will be a link in that review to sign up—well, to apply for that credit card.

When you click that link, you'll be taken by bankrate.com to that credit card issuer's website. (Much more rarely a third party in the middle that provides technical services.) And if you successfully pass through underwriting review for that credit card and you get it in the mail, and sometimes depending on the specifics of the contracts, if you actually activate that card by, for example, using it for a transaction, the issuer will cause Bankrate to earn a often fixed amount of money for the acquisition. The typical rate for, well, qualified credit cards in the United States, meaning the user of the credit card is well qualified, is plus or minus $200 per acquisition. 

[Patrick notes: Source here is what I would describe as “general industry knowledge” or what the New York Times would describe as “multiple people familiar with the matter.” Payouts vary depending on a variety of criteria, including the issuer, specific commercial agreements, the stringentness with which they can exclude particular actions from being paid out, and the economics of the product you sent people to. You might assume “And banks pay more for better customers, right?” but largely this is only through segmenting them into higher-tier products.]

So if you understand all of that, you now understand the Internet economy of credit cards probably better than Atlantic Magazine does.

Oh, I'm being a little harsh, but I feel like that's utterly deserved. [Patrick notes: On reflection the next day… yep, still deserved.]

Returning to my essay:

Rebating Interchange to earn share of wallet. As we have discussed previously in my article about how credit cards make money, credit cards have multiple different ways of earning money, but the most important one to this discussion is interchange. It is a fee ultimately paid by the card accepting business, which gets sliced up between various parties in the credit card ecosystem to incentivize them to put their logos in the wallets and on the phones of well-heeled customers and increase the amount they spend and the frequency with which they spend it. In industry, we sometimes distinguish interchange, which mostly goes to the issuing bank and scheme fees, which mostly go to the credit card brand itself, but as Interchange is much larger, let's just call them both Interchange for simplicity.

As an aside, another phrase you'll sometimes hear is the merchant discount rate.  

So if you're a grocery store and someone checks out with a hundred dollars of groceries, you notionally have a hundred dollars of accounts receivable at the point in which you have decided to give them their chicken. That customer has decided to give you some payments credential. And one way you could structure this transaction as the payments industry is to say, okay, you have a hundred dollars of accounts receivable. I will buy that from you for $98, the 2% representing a discount rate on the accounts receivable, which I'm buying from you, and we'll settle up tomorrow or the next day. How is that different than a 2% fee off the top that we call interchange? If you're not in this industry, you probably don't care about it. It's just a difference in nomenclature. 

So for simplicity, I'll call it interchange, but know that you might need to Google the words merchant discount rate or search for the words merchant discount rate when you're reading your own contracts.

Interchange is generally a percentage fee based on the final transaction size plus optionally a per transaction fee. You can just look up the rates, but I strongly recommend you don't, as you'll be reduced to gibbering madness. It took many smart people many years of work before Stripe could deterministically predict almost all interchange it was charged in advance of actually getting billed for it.

These words "look up the rates" are hyperlinked to the online resources, which show the rates for Visa and MasterCard. And if you wanted, you could find the rate schedule for Amex as well. [Patrick notes: I have to emphasize again: if you are not professionally interested in these documents, don’t try to learn anything but the general shape of the territory from them. These routinely befuddle specialists. You have been warned.]

These are very difficult to decode documents, for reasons. They don't matter to most users of credit cards either on the person who has the card in the wallet side or on the business that is accepting the usage of the credit card side. These rates are set within the financial industry / payments industry itself and cover transfers that are internal to the industry. The actual rate paid by the merchant at the end of the day does not necessarily one-to-one track the interchange rate within the payments industry.

[Patrick notes: One reason for the complexity of these documents is they represent multiparty industry-wide negotiated settlement between actors with cross-cutting preferences, including some actors who are simultaneously astride both side of the negotiation. Managing these stakeholders is a small part of what you pay Visa/Mastercard single digits bps on every transaction for. Someone needs to get Ogaki Kyoritsu Bank (disclaimer: no relationship aside from mutual affection for Ogaki, a small town in central Japan where I lived for a decade) and Paypal to the same negotiating table, and Visa/Mastercard did that work.]

Smaller merchants typically pay an all-in rate. [Patrick notes: For example, 2.9% bps and 29 cents per transaction, which was an anchor for many years in the United States after an upstart Internet payments processor popularized it. Another company later rounded it to 30 cents.] 

Larger merchants typically have negotiated bespoke pricing, which is often on what's called IC+ - interchange cost plus model, where they say, okay, I will pay whatever the interchange rate was at the end of the day, plus some number of basis points.

That “some number” will be sensitive to a number of things, including the acuity of the salesman, how merchant sensitive the merchant is, and ultimately of course the size of the deal and how competitive it is between the various players that could potentially service it.

To highlight something which is routinely surprising for non-specialists. Interchange rates are not constant and fixed. They're set based on quite a few factors. Gibbering madness intensifies, but most prominently based on the rank of card product you use, the more card product is pitched to socioeconomically well off people. The more expensive interchange is credit card issuers explicitly and directly charge the rest of the economy for the work involved in recruiting the most desirable customers.

As an aside that at many of the brands - and the word brand and scheme are interchangeable in credit card land, they mean the name of the logo that is most associated with card that is not the issuing bank. So, for example, if you have an American Express card, American Express is both the issuer of the card, but they are also the scheme. If you have a, let's see, Chase Sapphire Reserve, which happens to be a Visa at the moment, Chase is the issuing bank, Visa is the scheme or brand.

At some of the brands, the cost for interchange in the card is both determined by the sort of ranking of that card within their schema of perhaps five or six or 10 different ranks of cards could be at. And it's also based on the industry that the merchant is in. If you want to dive deep into the weeds, that will often be based on what's called the merchant classification code, MCC. And so groceries pay a different rate for credit card transactions, for the same customer on the same card, than a travel business might pay.

And there's other various ways that these things are different. For example, there is a discount if the card is present. [Patrick notes: This is based on the theory, and reality, that fraud is more common in card-not-present transactions. Banks will typically offload their liability for transactional fraud to the card-accepting business, but they would argue that they have to get the annoyed phone call and handhold an aggrieved innocent customer when it happens, and so they deserve something a bit extra for their trouble.]  

In a lot of circumstances, these pricing tables get extremely complex. And again, people whose only job is to understand the interchange pricing needed many smart people times many years to routinely predict all of it in advance.

The basic intuition underlying rewards cards as a product is that highly desirable customers have options in how they spend their money. You can directly influence them to use your rails by making those rails more lucrative, more fun, or both for the customer. And so card issuers and the networks compete with each other for so-called share of wallet by bidding with interchange. This is not quite as sophisticated as the system of dueling robots, which bids for your attention every time you open a page on the internet with an ad on it. The bids are generally speaking pretty static and made years in advance at or before the time a user signs up for a credit card product with relatively minor adjustments made over time. Program managers at card issuers are extremely, extremely sensitive to upsetting the apple cart and churning loyal users. And so they attempt to do this except when circumstances make it unavoidable.

As an aside, if you're a program manager at a card issuer, I apologize in advance for the following very hand-wavy explanation of how you live your life. But the thing that a program manager is attempting to do is create an appealing product offering to sell to users. They don't make that product offering change all that frequently, because customers will churn off of it.

They pay a high cost of acquisition for each customer that signs up for the product offering. Again, I quoted the number $200 earlier, and you generally want your customer to have many years of productive use of your product, to earn back that cost of customer acquisition and then get to actually earning you money. And so, you want your sort of target customer profiles to be relatively static over time for the particular product, and you want your appeal to that customer to be relatively static over time. So that you stay, quote unquote "top of wallet."

And so, your product proposition might be something like, okay, this card earns you cash back and it earns you more cash back than you would get from the basic offerings that you'll find at our competitors. But it's pretty set it and forget it. You don't have to think too hard about what categories you're using it on. And by the way, we have a beautiful website and you can call us 24 hours, seven days a week and et cetera, et cetera, et cetera. All the other things that you expect to get added to a card at this tier. And then you go about, you know, delivering on that promise to them and that makes you money. For ways we will discuss in a moment.

Why isn't every card a rewards card? Different regions have ended up with different equilibrium on the rewards game. In the United States, card acceptance is expensive and the rewards economy is robust. In Japan, card acceptance is expensive and the rewards economy is fairly muted due to effective collusion by issuers. In Europe, card acceptance is cheap by regulatory fiat, and so rewards are far less common or commonly lucrative than they are in the US. Similarly, debit card rewards used to be fairly common in the United States until the Dodd-Frank Act capped debit card interchange with a very important carve out for small banks in the Durbin Amendment. When interchange is regulated, the size of the pie isn't large enough to give the end user a tasty slice, and so they get nothing.

But we can see that even scoping to credit cards in the United States, not every card is a rewards card. What explains why Capital One, for example, offers rewards for its Quicksilver card, but not the Platinum MasterCard? Is mercury attempting to burnish its image after that horrible toxicity thing?

[Patrick notes: mercury is the name of a Capital One card product, and (unrelatedly) a fintech provider of banking services, which is not a bank. I use them for my LLC.]

You can find abundant evidence of this on any credit card issuer. If you want to go to their website, click the prominent link that says credit cards and see the wall of various offerings that they have for you. And those are designed so that some of them will be more attractive to you specifically, the listener of this podcast, and other of their offerings will appeal to different people in the market, but will be facially unappealing to you. Market segmentation is a fundamental thing that the industry is aiming to accomplish here.

Partly this is that different users have different jobs to be done for credit cards. Many users of credit cards, potentially including some readers of this column or listeners of this podcast, believe that they are the typical users of credit cards. No users of credit cards are typical.

As an aside, this is one of the things that the Atlantic article's more right on. There is actually bifurcation within the credit card market, although bifurcation is probably not the way I would describe it, because the camel has far more than two humps. It's a complex multi-dimensional space. [Patrick notes: Geeks are notoriously handwavy about claiming things are multi-dimensional spaces. Here’s four which you’ll be aggressively segmented on before you even get to the applications: credit quality, typical revolving balance, predicted spend per month, do-you-own-a-business? On the backend, it’s even more intense than that.

For example, do you have an intuition as to how the consumption patterns of a retiree and a young urban professional who each spend $2,500 a month on plastic differ from each other? Card issuers have an answer. That answer can cause them to seek compliant ways to sort those two users into two different products, because those products economic models will interact differently based on a user’s consumption basket.]

Does that seem like I'm pulling rank again against the Atlantic? Well, it's the truth. So what are you gonna do?

Credit cards are both a payment instrument and an access point for a revolving credit card line. That they're expensive way to borrow money is one of the first things said in any personal finance text. They're also one of the most accessible ways to borrow money, and that is their primary value proposition for many users, typically ones lower on the socioeconomic ladder. These users spend relatively little in a month compared to their revolving balance, which they continuously pay interest on. It trivially follows that most money earned on their accounts is earned via net interest margin and not from interchange.

As an ethnographic example of a thing that is very common at the lower parts of the socioeconomic ladder in the United States, we have as a nation made the choice that we would do some form of social benefits, not through traditional social benefits programs, what might be called Welfare or the dole in some countries, but by something which is called the earned income tax credit.

The way that most users of the earned income tax credit encounter it is that they have some amount of employment over the course of the year. They are charged payroll taxes and other deductions over the course of the year, including deductions against their federal income tax. They're awarded an earned income tax credit, which is a social decision from the policy of the United States of America to subsidize them.

And the form of the award of the earned income tax credit is that they will receive a refund from the IRS in excess of the amount that they paid for income tax over the course of the year. It might not be in excess the amount that they pay for payroll tax. [Patrick notes: Many people misunderstand payroll tax, since part of it is paid by the employer and may not be reflected on one’s paycheck. At the federal level, in the U.S., it is 15.3%, subject to some phaseouts/etc. ]

Some people do have net tax payments, considering payroll tax, which pays for Social Security [Patrick notes: and Medicare], but they do not have net tax payments with respect to income tax. Anyhow, when April rolls around, they get rebated a check or more commonly these days, an electronic transfer of a few thousand dollars.

There are personal finance-y ways to get that rebated over the course of the year. [Patrick notes: Adjust one’s “withholdings” with one’s employer. Unfortunately, the people for whom this advice is most relevant cannot easily action it, and the people for whom it is least relevant spend lots of time thinking about their Form W-4.] 

What many people on lower end of the socioeconomic spectrum do is borrow money from various sources over the course of the year, and then they pay down those debts after they have received their annual earned income tax credit, or other tax refund. [Patrick notes: “[O]ften considered the largest financial event of a low-income family’s year”, quoting a tax expert.]

So during the course of the year, if you just stylize and say, okay, at the end of the year or well, the end of the tax year, they will receive approximately $4,000 in EITC. That might cause them to spend, let's say, $300 or so a month on the credit card, and then pay that all down at the end of the year. And so the amount of interchange associated with $300 a month will be not a huge amount of money. It will be probably in the $6 a month range for the typical credit card.

And they will very likely incur much more than $6 a month in interest costs in borrowing money from the financial industry, from depositors, and similar and waiting to pay back that money with their earned income tax credit come April or May. [Patrick notes: Finger to wind? $30 a month. Average balance of $2k over the year, 20% APR.]

Back to the essay:

To the extent these users notice numerically defined features of their credit cards, which is somewhat dubious, it is the headline APR, the cost of credit and their credit limits. "Starter cards" and other products aimed at this user group typically have no rewards. They instead use interchange income to bid down the headline APR. Interchange functions as the subsidization of the cost of credit throughout the economy by businesses which want to sell things to people who would use credit to buy things. The limit case of this buy now pay later, where the cost of credit is subsidized straight to free is discussed elsewhere in Bits about Money.

The heaviest credit card spenders, and this fact is both uncontroversial and flies in the face of what many personal finance compliments believe, are wealthy and sophisticated. They use credit cards primarily as payment instruments. Issuers compete aggressively for their business, which is quite lucrative. This is not because they pay much interest because while they have higher headline APRs, they only rarely revolve balances. It is because "clipping the ticket" via interchange on high volume of transactions is an excellent business to be in.

How dramatic is this? Allow me to reproduce a graph from the paper "Regulating consumer financial products: Evidence from credit cards." This paper is something of the Rosetta Stone for credit card issuances a field both in that as a single source for understanding a huge range of human endeavor. And that it stands in for a very large literature that nobody else reaches for to be the first citation when the Rosetta stone already exists. If you see an unlabeled graph on Twitter about credit cards, it is probably lifted from this paper.

I'll say, as an aside, this paper dates back to 2014, but it's the best available in field, in my personal humble opinion, and I don't believe the conclusions of it have materially changed.

We're looking at figure D, from the paper interchange income minus reward expenses, graphed against FICO score. And the upshot of this graph is that there is no part of the FICO score spectrum at which issuers continually rebate more rewards to customers than they earn in interchange.

That's an extremely important fact about the world and, how to phrase this delicately... I don't disagree with it, on the basis of industry background knowledge. In fact, I've never met anyone in the industry who disagrees with this snapshot based on industry background knowledge. 

Which puts me in a somewhat awkward position where people who disagree with it on Twitter are things like, you know, writers for the Atlantic and professors at well-regarded universities.

And I feel like, okay, well: Welcome, amateurs who write for the Internet, to my neck of the world. [Patrick notes: You should read this with love for talented amateurs as a class, but also with a little bit of acid, because society is entitled to have high standards for those whom it has granted positions of trust.] Those of us who actually do this professionally feel pretty confident about what the ground truth is here.

[Patrick notes: If I could highlight a few choice Twitter comments, here are three.

]

As you can see as a percentage of average daily balance, ADB, even after a rewards expense interchange gets sharply more lucrative at the top end of the credit score distribution, 740 plus, which is roughly 10% of accounts. The difference is actually larger than you see here since credit lines and ADB also increase with credit score. For predictable reasons, rich people consume more than poor people on an absolute basis. Film at 11.

I'll drop a link in the notes to a graph of that, which is produced by the Fed. [Patrick notes: Graph below produced by me, on the basis of the Consumer Expenditure Survey from, apologies, the Bureau of Labor Statistics, which is not the Fed.] But your intuition here is exactly right. When you graph people by income levels in the United States, the top deciles of income consume a multiple of what the lower deciles of income do. The graph between the lowest deciles of income is less skewed than you would expect. One of the reasons for that is that there are some people who are higher wealth, but lower income such as retirees.

And then one reason is, you know, the entire point of having a subsidization regime is so that people who are lower in income get to consume more than they would in absence of a subsidization regime. [Patrick notes: This is one reason the above graph is so close to flat on deciles 1 to 3. Another is that people can pull forward consumption using debt and/or fund consumption from savings which are not directly reflected in their income number. And still another reason, underappreciated by economists IMHO, is that there are many people whose wealth and income are not well-reflected by summary statistics of their household’s economic life for this year.]

The United States polity generally votes for some level of subsidization of people who are at the lowest levels of income, with respect to consuming particular kinds of goods and services in particular, through different mechanisms.

One mechanism is the EITC, which is often not counted as income depending on the instrument that is used to measure it. One mechanism is the program that is popularly known as food stamps [Patrick notes: the Supplemental Nutrition Assistance Program (SNAP)], which I discussed with Dave Guarino in another episode of this podcast. And of course, there are other mechanisms.

The complexity spectrum of rewards products. The simplest rewards product are straight cashback. The issuer totals up all of one's net purchases, all purchases, less refunds in a statement period. It then credits the user with a particular percentage of that for each statement, either automatically or periodically at the user's request. It transforms some of that notional bank credit into a statement credit decreasing the amount the user needs to pay to cover their purchases in the current month.

This is as simple as it gets, and we're already necessarily hand waving away libraries worth of complexity. For example, calculation of net purchases needs to be fairly robust against adversarial collaboration of users and businesses accepting the cards or the issuer will get turned into a money pump within a matter of days, and will not likely be able to detect or reverse this condition for at least several weeks. This has happened many, many times.

As an aside, we heard some stories last week. [Patrick notes: For example, when a credit card issuer and airline mutually decided to subsidize donations to a charity up to a cap by rewarding donors with airline miles… but forgot to actually impose the cap over a weekend. Gary asserts that individuals have reported to him receiving high six figures USD-equivalent of value associated with that decision. On general industry knowledge, that strikes me as in-no-way-implausible.

I’d also note that Gary cites an explicit example of collusion between a credit card accepting business and users who have other-than-economic reasons to use that business, which is something I point to to demonstrate an empirical example of collusion happening, not to endorse it as a tactic.]

Credit card issuers, when they screw this up, lose millions of dollars and dry their tears on money.

[Patrick notes: Why do I say that? Because issuing credit cards is a very lucrative business to be in. Operational losses, credit losses, paying for smart people, and marketing spend all comes out of the wash. Many people think spending a million dollars on an operational loss is somehow worse than spending a million dollars on poorly calibrated marketing campaign. This is a preference the finance industry tries, with varying levels of success, to train out of its employees.] 

Anyhow, in the mists of history, that percentage started fairly flat, typically 1%. The math supporting this is typically fairly simple. Take in 140 bps as revenue, pay the user a hundred bips as effectively a cost of customer acquisition. Keep some portion of the 40 bps as one's margin. In nations other than ones with effective cartel like behavior by issuers, this equilibrium was not stable because competing issuers could bid up to, for example, 1.25% cashback on the same underlying economics and compete for share of wallet. This happened extremely robustly for decades.

One iteration of the game was cashback categories, particularly post-internet. Certain cohorts of customers were most interested in the headline cashback. Issuers began to design products that were much more complicated, such as 1% cashback in principle. However, 1.5% cashback at gas stations.

A fun rabbit hole about credit card acceptance. This rabbit hole, by the way, extremely relevant to the credit card issuing programs which are co-branded with an airline or hotel partner, which we discussed extensively last week. A huge percentage of all credit cards are quote unquote co-branded. They're issued by a financial institution, but bear the name of some other institution, which inspires a lot of loyalty, a teeny tiny percentage of co-branded cards name, for example, tertiary educational institutions, most name a business that the customer has an intense ongoing relationship with: Costco, their airline of choice, or other examples.

Co-branded cards are extremely big business. One subtlety about them: A co-branded card will often have a special rewards tier for the business named on the card. This is partly that business choosing as a marketing expense to split some of their core margin with their most loyal customers who are overwhelmingly the target audience for the co-branded card, but surprisingly to many non-specialists, this is not the sole source of margin to reward card holders with.

This is because at the scale of the largest businesses in the world, financial services are cross-sold and structurally interconnected. There is one team in a bank, it happens to be Chase for what it's worth, attempting to capture Starbucks card processing business. There is another team that wants to convince Starbucks to co-brand Chase card products. Those two teams can talk to each other prior to making proposals to Starbucks.

And so without knowing anything about the payments industry, you could speculate with a very high degree of confidence that Starbucks has received a complicated spreadsheet saying that Chase will charge it for every conceivable type of credit and debit card that wants a latte. A prominent negotiated line on that spreadsheet includes effectively a major discount for what are called on-US transactions when Chase's left hand needs to move money from Chase's right hand because someone wants to pay Starbucks money for their Starbucks using their Starbucks branded money.

This is very easy to conflate with and is separate from Starbucks' extremely impressive stored value product that probably mostly cannibalizes their credit card product, which, alright, we have to stop the levels of digression here. We'll be here all day.

Because on-us is structurally cheaper transaction, literally travels over fewer rails than therefore there are fewer mouths to feed in the transaction. And because it is incentive compatible for all parties to subsidize these transactions. As a card program designer, you have a relatively easy time being generous with rewards to this particular cell of your spreadsheet.

As an aside, these programs are not just co-branded, they're co-developed. The people in charge of making the program at the large issuer, the bank and the people at the large company that is bringing the engaged audience - those teams will be sitting in rooms together to craft the marketing message around this. And also in what is a very, very interesting poker game where nobody wants to show all of their cards. Because there are hugely, hugely consequential commercial negotiations happening, where like one basis point times many billions of dollars is real money.

But without showing all of our cards and without telling all of the secrets we know about our own business, can we give our partners some indication of, say, which market segments are strategic priorities over the course of the next couple of years? And can our partners say, oh, well those actually jive with our strategic priorities, at least in part?

So I'm gonna draw a Venn diagram here. The people in the middle of the Venn diagram, we'll go halfsies with you. Let's say, could we give them five basis points more in terms of rewards? Okay. Well, our marketing department... Well, maybe indicatively, we're good for two basis points of it. Would you like to give them three? Okay. Halfsies. 2.5, 2.5. All right, we can do that. Moving on.

[Patrick notes: Please treat these as indicative descriptions rather than as transcriptions of what the conversation sounds in the room. Often, the participants will need to back it out to interchange numbers, and there will be uncertainty. For example: “Our rack rate for bag fees is $55. We mutually want no-bag-fees to apply to 150,000 people who would otherwise pay them. We will charge you, the issuer, $15 per occurrence for waived bag fees, which we model as costing you approximately …  Do we have a deal?” And then a business analyst for the issuer will make an Excel model and say “Oh, that’s a 2 bps ask on 5% of the portfolio; here’s my modeled economics for that subsection. Recommendations: definitely want the free bags, should agree to 2 bps, best to concentrate our negotiating effort on other line items.”

I will reiterate that if I had exact knowledge of a specific negotiation that I would be unable to tell you, and will say that I am providing a hypothetical conversation for color purposes. I have very high confidence that, if you found people who were actually in the room, they would say a) I wasn’t in the room and b) I have a pretty good model of the thought processes of people who were in the room.]

And these conversations take months and months. This is a very in-depth process to develop these products.

Back to more complicated cases: So imagine you're the card program designer for a card that, like most cards, does not have a particular named business happily subsidizing your users. You desire to quote a headline cashback number much higher than 1%, but you'd still like to keep some margin from interchange. What could you do? One is you make the headline number larger, but contingent on something. Say for example, the card rebates 1.5%, but only for bookstores, any bookstore. For other transactions, it's 1%.

Patrick, commenting on audio: As an aside, Matt Levine pointed to this column and thought that I was making an implicit point about Amazon. I would not make that implicit point about Amazon. Indeed, if you were a card program designer and you said, I think we should make a card that you can use at bookstores, surely that is risk free? Somebody smarter than you in the department will say, well, Amazon might code as a bookstore and it would be very, very bad to give people a high rate on Amazon specifically, unless we already have Amazon's processing business.

And so you will see, the best cashback you can get at Amazon is from the Chase Amazon card, where Chase also has a large share of Amazon's processing business. [Patrick notes: … to the best of my knowledge and belief, using only publicly available data sources.]

Anyhow, less I can say about that, the better. So shifting it to the side, but pretend I said hobby stores or something and made a better decision here. We're just taking a broad class of merchants that might particularly excite a user that they feel a particular emotional attachment with.

The thing you would love with this offering is to preferentially attack people who are very emotionally invested in being readers and who spend very little of her card wallet on books. The emotional investment in the story of the card offers brings the customer in. The blended cost to acquire the customer is closer to the 1% industry standard and not the 1.5% headline number.

By the way, as an aside, I used the feminine pronoun for this user because women buy the vast majority of books in the United States, which is a useful thing to know if you work in marketing towards this field.

Money is fungible, money is fungible, money is fungible, but many people don't actually orient their lives as if this were true. And so the financial industry meets them where they are and then charges them for the privilege. The user values a dollar more when it is a books dollar than a food dollar. You as a credit card program manager can math out a way to get her as many book dollars as she is interested in.

This is in spirit similar to how Byrne Hobart describes airline frequent flyer programs as working or again, how Gary Leff did in last week's episode. One key difference that credit card program managers have to understand: a source of advantage for frequent flyer miles as a pseudo currency is that they can turn very low marginal cost inputs (unsold seats) into very high perceived marginal benefit outputs (free vacations). Books dollars may very well be worth more than a dollar to our target user here, but books dollars are very difficult to manufacture in quantity for less than about oh, 98 cents or so.

In principle, you could offer more than the direct interchange revenue as the headline number if you were very, very sure that your typical user would not preferentially use your card only to buy books and use a competitor's card to buy groceries, gasoline medicine, and similar.

Now, unfortunately, remember what we said about the typical credit card users. That's right, they don't exist. Very many of your users will do what you want them to and use the card in a perfectly acceptable, but not exactly optimal fashion, and you will have a blended cost very near 1% for them, and very many of your users will do exactly what you most don't want, and use the card to only buy books. That is far less incentive compatible for you, particularly if you decided that the business of manufacturing books dollars was so lucrative that you could rebate more than the direct interchange revenue given mix effects.

These users will have blended costs very close to your headline number, not to your modeled blended costs. These users will even band into tribes, find each other on the internet and swap tips for exploiting poor defenseless credit card program managers like yourself. The tribal elders will eventually run businesses with names like The Points Guy or Bankrate, which eventually get quietly acquired by very sophisticated private equity firms.

Those PE firms are betting that you'll continue paying generous per signup affiliate commissions to internet properties, which will send you new card users. You bet you'll be paying tens of millions of dollars annually to frequently adversely selected newcomers.com and Reddit's bet that they will continue shorting that they have pooled one over on you because ha ha, you're not nearly as good as they are at fourth grade math.

They're keeping spreadsheets. The biggest difference between you and a redditor is not ability to do fourth grade math or ability to do spreadsheets. Redditors are frequently sophisticated with their spreadsheets. Many of them could clearly earn three orders of magnitude more from the financial industry if they stopped thinking that the right way to monetize spreadsheet skill wasn't gaming credit card signup bonuses. The biggest difference is that you're optimizing over portfolios over time, and redditors are largely playing in single player mode and frequently over short horizons.

As an aside in the credit card churning community, single player mode is when you are optimizing your credit card usage only for yourself. Two player mode is when you have a person, typically but not always a romantic partner, that you can heavily influence with respect to their credit card purchasing decisions, and their actual use of commerce in the real world, which is often necessary to unlock some of the bonuses that the community wants to access.

You only care about single or dual player mode to the extent that you avoid obviously degenerate offerings where adverse selection selecting single players quickly dominate your entire book of business. This is very much a risk, which is why the bank has smart people like you keeping spreadsheets. The redditors think failures for the bank sounds like Pudding Guy.

We discussed Pudding Guy last week on the episode with Gary Leff. 

Pudding Guy was of course one of the highest ROI buys in the history of capitalism. [Patrick notes: I did not get the impression in listening to Gary’s explanation that he would agree with this statement, FWIW, but I still do.]

If I can say a little bit more about portfolio math here, people seem to have an intuitive understanding that portfolio math matters for banks when they're originating debts, with the intention to earn an interest rate spread on debt. And so, for example, you might understand that, okay, if you are lending to a pool of customers with, let's say low FICO scores, these are folks with socioeconomic challenges. You might expect there to be ex-ante and ex-post risks in this pool.

And I'm going to quickly make a decided in my own head division of these two: an ex-ante risk is you're lending to someone who you believe is a real, legitimate user. But this person is not actually a real legitimate user—they intend to commit fraud. Either in their own name or in the name of someone's identity that they've stolen or in the name of the identity of a person who does not actually exist in the fiscal universe. They intend to immediately abscond with the money.

An ex-post risk, on the other hand, is simply someone who has benign intent at the moment they sign up for the credit card product. But they are simply a worse credit risk than some other people in the economy. And between today and some point in the future, maybe it's three months in the future, maybe it's three years in the future, maybe it's after they have an accident, maybe it's after the next economic downturn—they might be unable to make good on their promises to the credit card company.

So you as a bank are signing up a portfolio of people and both the ex-ante and ex-post risks in that portfolio have to ultimately be paid by the interest rate that you charge to the portfolio. But within a particular FICO range, et cetera, et cetera, you don't know who are going to be the people that don't pay. Life would be so much easier if you knew just who wasn't going to pay you back. I mean, you wouldn't have to do much work at all. But since you don't know who in the portfolio is ultimately not going to pay, you have to offer them very similar terms.

Part of what all users of the debt product are paying for is for those risks. In a counterfactual world where the bank was better at underwriting or where there was just less risk in the economy to purchase, they would pay lower interest rates.

This is quite similar to credit cards in an interchange model.

The bank does not know ex-ante who is going to use their card in a way which generates lots of interchange margin for them that the bank is allowed to keep, and who will use their cards in a nearly optimal fashion in ways that might be nearly optimal for a particular card product, and who will use cards in, let's say, a degenerate fashion. [Patrick notes: “Degenerate” has a particular nuance in some parts of the world which does not necessarily comport to how it reads elsewhere. This is downstream of technical usage in mathematics.] 

And I'm saying degenerate here to mean borderline abusive, borderline fraudulent uses of the card. [Patrick notes: There is a rich history of people debating the ethics of various forms of actions against credit card issuers. I could not usefully elaborate on it without an essay to work in, except to say that if you were to ask the issuers themselves, they would say that they happily have some degenerate users on the books. There is a difference between Goldman Sachs and SBF! In 2025, if you’re in finance, you probably still want to transact with one of them in the future.]

I won't point any fingers in particular, but let's say that that exists in the world.

And you as the bank are offering very similar offers in terms of the mechanics of the card to these various subpopulations. And you are trying to model out what the blended effect of that will be over your entire customer base for the card. And indeed, if you could create structural ways to have less risk of abusive use of cards, to have less risk of customers making quote unquote "perfect decisions" with card, that would give you more margin to play with.

And you might do things to increase your marketing wins, such as moving up your headline rates for rewards rebates, for example.

[Patrick notes: One reason you care about the thing I call your “headline” rate is because you’re doing SEO for Google searches by way of your affiliate marketers. Your affiliate marketers know that an influential subset of searchers really cares about the headline rates.

They accordingly boost the prominence of the cards with the best headline rates on their sites. This gets lots of Google traffic to those affiliates, and also might even boost your bank’s site’s rankings on many queries of interest to you. And so you might be bidding margin not to influence the purchaser but to influence Google.

The world is fractal and mysterious, is it not.]

And indeed, banks do this. For example, if you spend a lot of time in the churning forums, you'll know that the big blue bank has a rule that they call 5/24. The mechanisms of that you can read in other places [Patrick notes: as you’d probably guess, that site is a credit card affiliate, because who else would write a well-SEOed article about a narrow credit card servicing topic]. Essentially it says that there is a tripwire for someone's use of the United States credit card issuing economy, after which we no longer really want their business.

The reason that bank chose that is they did some level of internal data science and said, okay, up to a point, we kind of like having sophisticated customers who run lots of money through our cards. It's great up to a point. Past this point, they're more headache than they're worth, both in terms of like the actual operational headache and that they frequently expose us to adverse selection. [Patrick notes: No non-public knowledge used to make this claim; I am surmising.]

And so we have a hard and fast rule, which we enforce across most of our product lines, that says, if you are past this trip wire, we no longer want your new business. We'll continue to keep your old business. We will honor our contracts. And indeed the thing that banks are increasingly doing is because the general story with regards to bank has gotten better, they can do near real-time de-risking for certain abusive uses of certain products.

For example, if you historically had a credit line associated with your credit card and let's say $20,000, and on the first day that you got that card in your hand, you put $20,000 worth of transactions through, the bank would kind of be stuck with that. They would not be able to in real time say, well, we gave you $20,000 worth of limit, you used it, but we didn't intend that to happen.

These days, banks can make minute per minute decisions. And so if you go out to Best Buy and spend $5,000 on a flat screen TV and then go out to another company and immediately want to spend another $5,000 on something on the first day, despite you nominally having limit as-yet unused, your bank is likely to decline that second transaction. They want to de-risk, because their history of people who spend $20,000 in a day on the credit card is just not that great.

And that's because that history blends people who are imprudent users of credit. It includes people who are doing fraudulent use of credit. Iit also includes people where a deal was struck and they're keeping up their end of it. And so the bank says, we're altering the deal just a tiny bit. Pray we don't alter it further.

[Patrick notes: Is this abusive on behalf of the banks? I tend to think not so much. The contract almost certainly gave them the we-can-alter-limits-at-any-time-to-manage-risk rights, and the spirit of the agreement was not “Let me charge $20k on the first day.”, and so a user with a quixotic preference there is I think entitled to the bank’s decision whether the bank approves or disapproves their usage of the card on the first day.] 

Further refinements in cat and mouse games. Just like lotteries have to keep re-skinning the random number generators or the games get stale and ticket sales go down, credit card program managers have to periodically shake things up for something to cut through their customer's built-in resistance to marketing and massive, massive marketing budgets.

One innovation now 20 plus years old was a rotating favorite category for cashback. So instead of it being 1.5% for bookstores, 1% to all else, it would be 1.5% for groceries in Q1, for gas in Q2, et cetera, et cetera. The theory behind this was pretty simple. Many customers attracted by the headline number could be brought in the door, but would be fairly inattentive after a card was their new top of wallet, which is the default card they use for spending over time. A combination of inattentiveness, changing willingness to spend time playing the game and rewards caps would bring the portfolio's cost of rewards closer to the baseline and not to the headline rewards number.

I introduced a couple of concepts in this paragraph. 

One: top of wallet. So top of wallet hearkens back decades in the credit card industry, and it means, especially for your most valuable customers, they might have many payment methods available in their wallet, but there's literally one that is at the top. And if they're not making a considered decision on a transaction by transaction basis, if they've just been on a date and are more concerned with the lovely person staring across the table at them than they are concerned with which card comes out of their wallet, you want to be the default one that they pick, the one that is on the top of the wallet, because you're going to capture most of the transactions over time.

[Patrick notes: This is, evocatively, known as “share of wallet.” We also include things that aren’t actually in your wallet in share of wallet, of course, which is why “Cash App competes with debit cards for share of wallet” is a sentence you can say.] 

Patrick, commenting on audio: And again, capturing the transactions of well-heeled people is an excellent business to be in. So top of wallet has become more important over time rather than less due to the ergonomics of using credit cards, particularly for online transactions and through electronically facilitated transactions like, for example, the card you have in Apple Pay.

And so, just using the convenience sample of myself, I have probably 40 credit cards and debit cards and other various pieces of plastic running around. [Patrick notes: The combination of having weird hobbies, which eventually might have lead to working in industry, and—importantly—having this count doubled once by needing cards in two nations and once more by needing cards for the business, too.]

Patrick, commenting on audio: But I only have two in Apple Pay at the moment. One for dollars for my personal spending and one for dollars for my business spending. And so, if there was hypothetically another card that was in my wallet and my wallet is at home, that issuer is never going to get the interchange revenue if I'm transacting on my phone. And this behavior is something that you see, not just with yours truly, but a lot through the market segment that I'm a small member of.

(I clarified US dollars, by the way, because, well, there might be some people who listen to this who don't know. For many years I lived in Japan. I continue to maintain part of my life over there and part of my business over there. And as a result I have some cards which are Japan specific as opposed to dollar specific.)

And the word rewards cap is also similar. So a rewards cap function is something like this. We say that you will get 1.5% on a revolving category on the first $2,500 of spend in the category in a quarter. If you spend $10,000 on the category, then mechanically, even if you are getting 1.5% on the first 2,500, your blended cost on that category alone is going to get pushed much closer to 1%. It's finger to the wind mental math here, it's probably gonna be like somewhere in the 1.15% ish region.

[Patrick notes: 1.125%. Apologies for imprecision; this is embarrassing to me, as a former member of the math club, but my ability to do mental math while speaking is near zero.] 

Patrick, commenting on audio: Like many innovations in credit card program design, caps are targeted against a specific subset of the population, which is, they care a lot about the headline number and their highest spending, but they don't care enough to ridiculously optimize their use of cards and to switch out a card after it has hit its $1,500 cap in Q1. Many people, even people who are quite financially sophisticated, don't feel like it's sufficiently worth their time to keep a spreadsheet of every dollar of transactions they have made in Q1, and then update that spreadsheet when Q2 rolls around and a new cap is in place.

Does this represent the financial industry pulling something off on the customer? No. It's financial industry offering services. It describes those services fulsomely on the marketing page, and then delivers all of them. But it does represent the financial industry making a bet that a particular offering is going to sound very attractive to a customer. And then that customer's use of that offering will be profitable for the issuer of the card.

And the financial industry will not apologize for that. Like, we're all capitalists here. You know what no business under capitalism has to apologize for? We have a product. It does what it says on the tin. We deliver it to you, and then we make money on it. Take pizzerias. They charge a markup on cheese, pepperoni and tomatoes, labor, rent, et cetera. You get a pizza, the pizza is wonderful. Hopefully you'll come back next week. [Patrick notes: And then the pizzeria owner makes some cheddar.]

Back to the essay:

If for some reason playing this game interests you, one prominent product is called the Chase Freedom Card. It should not be an attractive product for you, given plausible assumptions about the readership of this column or listeners of this podcast, but it is an attractive enough product that probably millions of Americans use it.

Giving the customer choices more frequently.

If you were hypothetically to have spent the last few years dining with relatively well off people in the San Francisco Bay Area, when it came time to pay for dinner, you'd see two credit card products with a combined 90% plus share, the Chase Sapphire Reserve, and an interchangeable American Express card. This includes many diners who are professionally connected to upstart payment methods.

American Express for a very long time had an almost mortal lock on the top end of the credit card market. Chase attempted to disrupt that in 2016 by very overtly attempting to buy away their core customer. You can read lots in other places about the original promotion, generally discussing about how absurdly lucrative it was for customers and how it caused Chase to have a temporary loss associated with their cards business. Personally, I think it was one of the most interesting strategic moves in the last 20 years of retail banking, but a full essay on that would require some other than public knowledge as to the size of the tsunami that happened after it.

One thing that is public but not well appreciated: Chase didn't just decide to create an extremely lucrative for the customer offering out of goodness of their hearts and out of their own profit and loss. No, they pitched Visa on this idea. For too long, Visa, you have watched your competitor, American Express, outcompete every issuer in the Visa system for the best wallets in the world. They can do that because they can afford to. Because American Express charges systematically higher interchange rates than Visa does, even at its top most tier.

Visa, you should create a new tier where your not exactly chosen champions can try to spend those interchange dollars to give American Express a run for their money, and lo, Visa did create a new tier. You can expose yourself to Gibbering Madness if you want to know what the name for it is. But Chase got Visa to authorize Chase charging almost the entire economy more for credit card acceptance, with the specific goal of outcompeting American Express for the most lucrative, highest monthly spend, virtually never revolve a balance credit card users. This was part of what made the numbers work.

Another part is that the CSR offers a fairly complicated reward scheme with a lot of opportunities for people to pick things which feel great, but are not optimal. For example, Chase will let you cash out CSR points on a one-to-one or worse basis at either Amazon or Apple integrated directly into the checkout flows. Those feel great.

And so a thing that you'll find if you love using your Kindle like I do, is a Kindle book might be priced at $3 and literally on the physical device of the Kindle, if you have a Chase card with CSR points attached to it, attached to your Amazon account, the Kindle will prompt you: Would you like to spend 375 Chase reward points on this?

You probably shouldn't do that. But it feels great in the moment. Ooh, Chase is giving me free books.

[Patrick notes: I saw an even more compelling interstitial than the below depiction of a store page, but could not conveniently reproduce the interstitial at press time. So instead I have the store page for Dwarkesh's Scaling Era, available through Stripe Press or at many fine bookstores. Disclaimer: that is my former employer, I received a review copy and blurbed the book, and I was on Dwarkesh's podcast once.]

In the article, I have an example of the Amazon checkout, which is on the web, not on the physical Kindle device. [Patrick notes: Or a Daylight Computer, which is a superior Kindle reading experience. If they could make it just 20% smaller/lighter it would be perfect.]

But it's the same story. Chase will allow you to cash out 1.25 points for $1. That is not the optimal way to use it. 

People love Amazon and Apple, and they love free Amazon and Free Apple even more. This is true even among a portion of very sophisticated, wealthy CSR users who love this idea so much that they click a button designed for suckers.

Why is that a sucker’s checkout button? Because CSR also includes a feature called Pay Yourself Back, which in the past prominently and today, a bit less prominently, lets you cash out rewards at better than one-to-one. You get a 25% bonus if you pay yourself back by nominating past transactions. At time of this writing Grocery stores. 10,000 points gets you $125 in statement credits.

Stating the obvious: Chase knows what a computer is. It does not require you to actually identify your purchases at grocery stores. This is a product decision to both a) force you to use your card to grocery stores and b) force yourself to say, chase is getting me free groceries. How nice of them on a transaction by transaction level, once a month to use your card optimally.

There are many, many other sucker buttons. But because the card is fundamentally targeted at rich, sophisticated people, Chase really does pay out a shedload of rewards. The 3% headline rate for travel and dining, plus the 25% Pay Yourself Back kicker means that there is a sustained and trivial pathway to get 375 basis points out, which is one of the very, very few places in the industry where there is a sustained, trivial and uncapped way to get more than the direct cost of interchange.

Why did this persist? Partly it is due to the standard credit card portfolio strategy. Every time someone uses a CSR on Amazon and not, for example, the Amazon rewards card also from Chase or uses CSR points to buy high margin, white plastic, Chase's contribution margin for the portfolio goes up and at scale, quite a large percentage of customers do this.

Partly it is due to Chase thinking that this will be so good for the target customer that their target customer will not bother playing the game optimally. This would require the customer carefully maintaining a portfolio of credit relationships and having seven cards saved on their iPhone and not simply the CSR.

Partly, it is due to a really interesting strategic reason for having CSR available to a much larger degree than American Express: Chase's diversified financial services empire. The CSR was effectively designed as a wedge product to get something Chase branded into the hands of affluent up and coming young urban professionals with the goal of eventually getting them not to just move their wallet, but their entire financial existence and potentially their current or future entities financial existence onto Chase. The Sapphire mini brand was so loved that they reused it for a bog standard premium checking account. It was named the Chase Sapphire premium checking account in a decision which probably consumed several tens of millions of dollars of professional effort. And I mean that absolutely descriptively and not as a criticism, I was not in that meeting, but I've been in that kind of meeting during my career. Those who have been in it know that it is not in any way a single meeting, single decision maker sort of call.

Patrick, commenting on audio: A thing I did not mention during the writing of this essay last year was that it could also be subsidized by charging poor people lots of interest because that fundamentally isn't true. That's not how Chase thinks about it. That's not how Chase describes it in their annual reports, in their reports to regulators. That's not how internal people at Chase will discuss it with you over dinner. And believe me, if you find the right people, they will blab all sorts of things over dinner. 

[Patrick notes: You will note that I am using Chase as a synecdoche here for the wider set of credit card issuers, which is a thing that real people will be really annoyed about, and so I immediately corrected the impression.] 

Not saying anything about people at Chase specifically, just making some comments on general industry knowledge. Anyhow, the narrative is utterly false. Ah, I get a little frustrated sometimes. I'm sorry.

More directions to go in. The blessing and a curse of essays is that they have to end somewhere, then pick up a new somewhere else. Hopefully the above gives you a bit more context on what is in your wallet. In the future, we'll likely discuss the complicated iterated game played by the credit card networks and the rest of society regarding interchange rates, how there are high interchange, high reward equilibria, and low interchange, low reward equilibrium, the recent settlement where the networks agreed to temporarily decrease interchange and other topics.

Patrick, commenting on audio: And that's the end of the essay. But before I leave you, I want to give you a brief, preemptive answer to a question. Isn't the cost of interchange just shared by all customers? And so, some customers typically those at the socioeconomically higher ends of the scale, get something back for interchange in the form of credit card rewards. And some customers typically at the socioeconomically lower end of the scale, paying with either cash or a debit card, don't get something back for interchange. Isn't this unfair?

It's a hand wavy version of an argument that very many people have made, including the Atlantic, including two economists. One of those economists called me incompetent at this field, which… opinions may vary. Credit card issuers are, across their various portfolios, largely competent at portfolio construction.

It is unlikely that there is a material sized portfolio or sub-portfolio where they are truly rebating more rewards to users than they are earning in the form of interchange. And so, the thing that you need to show in the world for this to happen is like, there is a net greater transfer to the user in terms of rewards than the amount that businesses pay for their custom. If there isn't, you can just immediately discard this argument.

[Patrick notes: I made an abbreviated version of this argument on Twitter, and will write it down more formally sometime. It is counterintuitive to many people. I do not think, on considered reflection, that it is false. I think the Atlantic article’s thesis is incontrovertibly false. It specifically alleges widespread cross-subsidization through interchange. This just doesn’t happen, not at the frequency or scale alleged.]  

Patrick, commenting on audio: A thing I use to help people visualize this is consider a case where you don't have the confusing notion that you are paying a percentage out of every transaction. Consider the case where there is a fixed cost. The grocery store has to spend some amount of money every day on electricity or even the cost of baggers. The cost of electricity or the wages owed by people who are doing the bagging is relatively constant, no matter whether the grocery has a busy day or a less busy day.

So how would you allocate this across all users of the grocery store? Well, if you tell me I would allocate this evenly across all users of the grocery store. If there are a hundred people that bought groceries that day, you allocate the cost of electricity evenly to the hundred people. That's an argument. If you make that argument, all costs fall disproportionately on poor people. No matter what.

No matter what you pay for electricity, no matter how you decide to give bonuses to your baggers, no matter whether interchange is percentage based, is capped by regulation, et cetera, et cetera, you have built into your assumption about the world that you are using in your toy model that all costs fall most prominently on poor people.

So here's my counter proposal for toy model: we allocate costs rateably among users based on what percentage of the store's business they represent. Now, true for groceries, true for most parts in the economy, people who are higher up on the socioeconomic spectrum just consume more. And so when you allocate the cost of electricity, the people who are higher on the socioeconomic spectrum pay more of the cost of electricity through their grocery store than poor people do.

If you make the argument that poor people are subsidizing rich people's use of electricity at grocery stores? Straightforwardly false, right?

If you want to make the argument work that poor people are subsidizing the store's cost of payments, a necessary sub-argument is: okay, I agree that we're going to have to allocate most of the cost of payments to rich people. [Patrick notes: I can have a productive discussion with someone who agrees with this, and still believes the following paragraph. If you don’t agree with it, I don’t know whether math works for you the same way it works in most of the universe, and prefer to spend my limited time discussing things with people for whom 1+1 reliably does not equal 3.]

But, you would continue, I think that rich people get even more than that aggregate cost of payments to them, through rewards. [Patrick notes: The way this is usually phrased is “Well, cash payers and debit card users suffer high prices, but only rich rewards card users get rewards. Therefore, the poor subsidize the rich. QED.”

This does not work if the rich are most of the spend at the store, because obviously they’re paying the interchange (through their own higher prices). And even if you did not have access to the store’s processor reports, you’d have wildly abundant evidence if we were in this world, because most “rich people” would earn greater than zero rewards net of their own cost of interchange. Which does not happen.] 

And that… just doesn't follow from the math. And you would see that more clearly in a format that made it easier to talk about math.

[Patrick notes: Hope springs eternal, among advocates and perhaps some scholars, that one can find a math equation which would prove this to be true. It’s a tidy just-so story.

For example, some of my favorite authors (some on the team wrote the paper I describe as the Rosetta stone) took a whack at the topic in Who Pays For Your Rewards? Redistribution in the Credit Card Market. The authors would, I think it is fair to say, strongly prefer a world in which they found the standard redistribution narrative.

They looked and… found something to publish.

We next investigate whether differences in net rewards across FICO scores are driven by underlying differences in cardholders’ income, which would suggest a redistribution from poor to rich consumers. If FICO scores are positively correlated with income and high-income consumers spend more money, then they will earn higher rewards. Indeed, in the financial press, credit card rewards are often framed as a reverse Robin Hood mechanism in which the “poor foot much of the bill for credit card points, miles, and cashback” (Vox, 2023).

Our results, however, show that this explanation is at best incomplete.

The authors eventually find, and please accept this as a short gloss that encourages you to actually read this paper, that redistribution happens if you’re allowed to say redistribution is mediated by savviness in execution. Because they develop inconvenient results like, quoting verbatim:

High-income consumers with high FICO scores benefit the most from reward credit cards compared to mid- and low-income consumers with high FICO scores. At the lower end of the FICO distribution, however, this pattern is reversed. On average, net rewards are far more negative for high-income consumers with low FICO scores than for middle- and low-income consumers with low FICO scores.

Or, to put that another way: if there is redistribution happening, it necessarily includes redistribution from unsophisticated high income customers to sophisticated low income customers.

This is an inconvenient result if you want to sell the untrue story that the Atlantic (and the NYT, and Vox, and etc etc) want to sell you.

]

Patrick, commenting on audio: But of course, rewards aren't the only thing that interchange pays for. I think this is an important thing for people who are greatly concerned about the justice that we offer people in terms of payments, to be concerned about. One of the things that interchange pays for is finding someone in society that is willing to hold certain varieties of risks.

[Patrick notes: There are many other things interchange pays for, of course. If you’re of a certain age, you remember when your local pharmacy and department store maintained their own credit accounts, by the hundreds or thousands, mostly by hand. This has virtually disappeared from the world, because the financial sector was able to do it more efficiently, with lower costs, in a way more palatable to businesses and customers. 

Those clerks weren’t free! The math did not math itself! The Post Office charged for stamps in those days, too! The customer offended by her saleswoman saying “You’re late on payments and so no communion dress for your daughter” did not forget that conversation come Christmas!

Interchange replaced all of that, with a more legible cost. Some people really, really hate legible costs. Their feelings will not make businesses return to the old world, because that world sucked.

]

Patrick, commenting on audio: The reason that most interchange flows to the credit card issuer, and not to, for example, Visa or someone else involved in the transaction has historically been that they say, well, of interchange not ultimately remitted to the customer in the form of rewards, we are the ones who do most of the operational work. We did most of the sales work. We hold most of the risk.

[Patrick notes: The marketing and sales work that banks do on behalf of credit cards is extraordinary. It is the largest source of funding for most sites on the Internet which do financial education, for example. Thoroughly incentive compatible, right? Congratulations on Googling what a credit report is, we’d be happy to inform you, BTW have you considered getting a starter credit card? We have an entire journey built out to support you past the landing page.

I’m not being acidly sarcastic here! There is genuine social purpose in extending credit! That implies that someone needs to e.g. go out to universities and hand out T-shirts and explain what APR stands for! The people who do most of that work are paid by banks to do it.

] 

At the end of the day, something that the grocery store is paying when a low income consumer who does have a credit card comes in and pays with the credit card is: in the future, if the user can't make good on their payments to the credit card, that doesn't matter. The grocery store still gets paid about two business days after the card is presented, no matter what the future behavior of that user is. [Patrick notes: Assuming the user doesn’t dispute the transaction. The business still bears fraud risk, but it doesn’t bear credit risk. They have sold that risk to the issuer and the interchange included the price.]

Equitably apportioning non-payment risk in society

[Patrick notes: Many people are surprised about the below representations. If you’ll permit me to be so bold, this knowledge is class-gated and most people who read about financial infrastructure on a professional or hobby basis have not belonged to a social class which needs to be expert in the consequences of bouncing checks.

I direct people seeking confirmation to reputable reporting, such as this piece by the NYT in 2012 or by the Chicago Tribune earlier, or to a legal expert of your choice. A quote from the NYT:

[Collusion between retailers and the state] shocked Angela Yartz when she was threatened with conviction over a $47.95 check to Walmart. A single mother in San Mateo, Calif., Ms. Yartz said she learned the check had bounced only when she opened a letter in February, signed by the Alameda County district attorney, informing her that unless she paid $280.05 — including $180 for a “financial accountability” class — she could be jailed for up to one year.

“I was so worried driving my kid to and from school that if I failed to signal, they would cart me off to jail,” Ms. Yartz said.

Note that Alameda County is doing a kinder, gentler diversion from criminal prosecution here. Many jurisdictions treat uttering like they do simple theft. They simply prosecute and convict you. (These decisions are often heavily influenced by… look, you’ve read about the U.S. justice system, right? All the things.)

Curious as to what diversion programs look like from the perspective of an aggrieved grocery store? Florida will happily explain it to you.

] 

I want to talk about something I've discussed in The Long Shadow of Checks because there is a different way that society can allocate the relative risks of nonpayment of a particular user.

I want you to think of someone who sounds very sympathetic.

I'm going to indicatively say maybe she's a single mother. She has children at home. Her children are hungry. She goes to the grocery store and at this exact moment in time, she knows that she is behind on the rent. She knows that her income situation this month does not look great. The child support did not come through. She knows she is unlikely to be able to pay for the basket of groceries that she has today.

In the case where she pays by a check, having the knowledge at the time that she writes the check that she'll be unable to pay for the groceries, this user is committing a crime. The crime is called uttering in many states in the United States.

People go to jail for uttering. The cost of sending them to jail is ultimately borne by the taxpayer.

Of course, the true cost, the cost society must be most morally concerned with, is the cost to the mother, the cost to her children. But in terms of dollars and cents costs, even if you discounted the social value of her staying with her children to zero, there is a huge cost associated with arresting, investigating, prosecuting, and incarcerating her. And again, we investigate, arrest, prosecute, and incarcerate people for uttering every day across the United States.

If she uses a credit card to buy a hundred dollars of groceries that she ultimately can't afford, the cost to society is… a bank takes a minor hit to their credit losses budget. There are countermeasures against that and consequences for it. The bank will report to a credit reporting agency that she did not pay her bill. But she doesn't go to jail. Her kids don't lose their mother. 

That seems like obviously a better equilibrium for society than the "Well, we are going to make checks good for use as money, by jailing some portion of the people who use checks."

So when you think: what does interchange pay for? Why do we charge grocery stores interchange? Why can't they just process check payments for 25 cents each? Why do they have to pay $1.50 to do a typical size grocery basket on a card that offers no rewards? One of the things that $1 and 25 cents pays for is for those children to be able to continue seeing their mother. 

I think that's cheap at the price.

And see you next week on Complex Systems.