Just after I went on holiday in July, the Boston Fed released a 57-page paper quantifying the subsidy from poor to rich that is the result of credit-card interchange fees. It was picked up by the likes of the NYT and the WSJ, and now Tim Chen, the CEO of NerdWallet, has decided to push back on the findings.
The numbers in the Boston Fed paper, says Chen, don’t pass the smell test:
The popular press had a field day with the idea that card-using households are earning $1,482 annually from cash users. But if we assume that the reward rate is 0.75% on rewards credit cards, as they mention on page 15, then the average card-carrying American has to spend $197,600 on credit card purchases each year. Even if we assume that card users receive the full 2% merchant fee, which is ridiculous, we’re talking about $74,100 in credit card spending. Keep in mind that this isn’t the number for “rich” card-carrying Americans; this is the average, and it doesn’t include any cash that these households might be spending, so something smells fishy.
This does a really good job of misrepresenting the Boston Fed paper. For one thing, the rich don’t just spend more money on credit cards, they spend more cash, too. So a lot of the cross-subsidy on this axis takes place from Americans to themselves: they take the money they overspend when they pay in cash, and get it back in terms of credit card rewards when they pay with plastic. The real cross-subsidy takes place between rich and poor:
On average, and after accounting for rewards paid to households by banks, when all households are divided into two income groups, each low-income household pays $9 to high-income households and each high-income household receives $434 from low-income households every year. The magnitude of this transfer is even greater when household income is divided into seven categories: on average, the lowest-income household (< $20,000 annually) pays a transfer of $23 and the highest-income household (≥ $150,000 annually) receives a subsidy of $756 every year.
Still, Chen’s calculation is mathematically correct: even if they’re getting the subsidy from themselves, card-using households are still getting a gross benefit of $1,482 per year, on average. That number seems high, and it’s derived from the same data used to generate the poor-to-rich cross-subsidy data. So is Chen right? Is the data flawed?
In fact, the paper says, on page 8, that the average household spends $1,190 a month on credit cards — that’s $14,280 a year. How can that level of expenditure generate such benefits of $1,482 a year? The answer is that the $1,482 number has nothing to do with the amount of money that credit-card users spend on credit cards, and it emphatically is not, as Chen implies, an estimate of the total value that card-users get back in the form of rewards: that’s just one part of the total calculation.
The real reason why the $1,482 figure is so large is that credit-card transactions account for only 17% of total expenditures, but raise prices for everyone. Everybody pays the same price, which is higher than it otherwise would be because merchants have to pay interchange fees to card companies. People using credit cards get some benefit from that, but people paying in cash just end up paying more than they otherwise would have to.
The numbers and formulas can be found on pages 17-18 of the paper, and they’re not easy to follow. But boil them down, and it comes to this: total merchant costs are $54 billion per year. Of that, $24 billion is accounted for by credit-card transactions, and $30 billion by cash transactions (which includes debit cards, in this paper, for the sake of keeping things simple). Cardholders also get $8.5 billion back in the form of rewards.
So people paying in cash end up paying 83% of the merchant costs, despite accounting for only 55% of merchant expenses. Meanwhile, people paying with credit cards pay only 17% of the merchant costs, despite accounting for 45% of merchant expenses, and they get $8.5 billion in credit card rewards back on top of that.
The Boston Fed study assumes that it would be fair if cash payers and credit-card payers paid for merchant expenses in proportion to the degree that they caused those expenses. The cost to cash payers is essentially the degree to which they help pay merchants’ credit-card expenses. And the benefit to credit-card holders is the degree to which merchants’ credit-card expenses are paid by cash payers, plus that $8.5 billion in rewards.
Chen has other problems with the Boston Fed study. He complains, for instance, that it includes housing and automobile expenses in the total expenditures split between cash and credit cards, despite the fact that precious few people use credit cards for either one. That’s a fair complaint. He also says that the poor and the rich shop at different places: if you assume that all poor people shop only at dollar stores and only pay cash, while all rich people shop only at Bergdorf’s and only pay with credit cards, then there’s no cross-subsidy at all. The Boston Fed study, it’s true, makes no allowance for the phenomenon of rich people shopping at rich-people shops.
Finally, Chen complains that the Boston Fed survey ignores a lot of fixed costs involved in handling cash, and then parades a long list of cash-related expenses, like the cost of incorrect change and the cost of returned checks, which he considers to be fixed costs rather than marginal costs. But the fact is that the paper does assume $30 billion in marginal costs of dealing with cash payments, and that number is big enough to encompass a lot of what Chen considers fixed costs. As the paper’s authors note, “the data do not distinguish well between fixed and marginal costs”.
Chen concludes that the paper’s numbers might well be “absurd”, and that if they took into account his quibbles, the cross-subsidies might disappear entirely. I’m far from convinced. Yes, the paper makes a number of simplifying assumptions. But for every assumption which might serve to ratchet up the size of the cross-subsidy, there’s another which serves to ratchet it down. Here’s the paper:
We have omitted from the benchmark transfer calculations two very important features of credit card markets—redistribution of bank profits and business credit card use — that most likely would increase the transfer estimates and by much more than the reductions reported in Table 12. In other words, we are confident that we have most likely understated the transfers rather than overstated them.
Redistribution of bank profits is basically a function of the fact that credit cards are a profit center for banks, and banks are owned by the rich, not by the poor. So when banks make money from their credit-card operations, that money ultimately benefits rich people with credit cards, rather than poor people who pay cash.
And the Boston Fed study looks only at individual credit card use, ignoring the huge market in corporate credit cards. Many rich people put a lot of their personal spending on corporate or business cards; I myself use a business card for nearly all my credit-card purchases, which dates back to my days as a self-employed freelance journalist. I’m therefore one of the lucky recipients of the cross-subsidy here, but I’m ignored in this study.
Overall, I’m much more persuaded by the Boston Fed study than I am by Chen’s attempted fisking of it. The numbers aren’t completely accurate — they can’t be. But the true cost of interchange fees is clearly paid by the many, with the bulk of the benefits going to the few, and therefore there’s bound to be a cross-subsidy there. And I don’t think that Chen is being intellectually honest: he’s looking only for aspects of the report which might overstate the subsidies, while ignoring everything which might understate them.
Finally, Daniel Indiviglio looked at the study, and concluded that although the costs are real, they’re worth paying:
Ultimately, the question comes down to whether the cost of placing an additional burden on the poor is worth the economic benefit that robust credit card usage provides. Unfortunately, due to the nature of the industry, it’s not clear that there’s a way to have both. In order to encourage more credit card use, the poor end up stuck with the bill.
I don’t agree with this. Of course you can have both: you can mandate lower interchange fees, for one thing. US interchange fees are the highest in the world, for no good reason. And you can pass a law allowing merchants to slap on a surcharge for credit-card transactions. Credit-card usage might decline, at the margin, if you did that, but it would still be “robust” — it just wouldn’t be excessive. And the poor wouldn’t end up paying billions of dollars for benefits which accrue overwhelmingly to the rich.
Update: Tim Chen responds in the comments. He makes some conciliatory noises: “I don’t necessarily disagree that interchange fees should be regulated,” he writes, adding that “cash users do bear the brunt” of the price increases that merchants slap on to be able to pay interchange fees. But overall he sticks to his guns as far as the numbers in the paper are concerned.
One of the biggest problems I have with the paper is the $1,482 benefit. I’m going to stick to this number for simplicity, even though it doesn’t account for income differences or cross-subsidies.
I think this is a silly stance to take, because the $1,482 benefit is entirely theoretical. It’s a benefit accruing to card holders which ultimately comes from those who pay cash — but in the real world, most of us do both. The authors aren’t saying that the average person with a credit card receives benefits which net out to $1,482 a year. If you want to see how the math works for individual people, then you have to look at the way that the authors slice the population according to income. The $1,482 number, rather, includes a lot of benefits that people essentially pay to themselves.
Let’s say I pay in cash with my left hand, and use my credit card with my right hand. Then the paper’s saying that if I’m an average American, my left hand is paying more than its fair share, while my right hand is paying less than its fair share. And the real problem isn’t so much that my right hand is getting benefits from my left hand — rather it’s that poor people are much more left-handed than rich people. And so they end up sending money from their left hands to rich people’s right hands.
The $1,482 number comes from looking at the money flowing unfairly from all left-handed spending, which ends up helping out all right-handed spending. It shouldn’t be confused with the average amount of right-handed spending that the average American engages in. Chen writes:
If I’m a card user spending $15,000/yr, how is it even mathematically possible for me to receive a benefit of $1,500, or 10%? Card companies are siphoning 2-4% off of every card transaction, so shouldn’t this serve as an upper bound on any benefit I can possibly receive?
But the point is that we’re talking about money which left-handed spenders are spending that they shouldn’t be. And that isn’t really a function of how much right-handed spending is going on. Indeed, if the share and amount of right-handed spending dropped from 17% to 10%, then the average benefit to right-handed spenders would go up, not down, because the left-handed spenders would be overpaying even more, to the benefit of ever fewer right-handed spenders.
Chen has another problem with the methodology:
Saying that a reduced burden on the part of the card users is the same thing as a benefit seems to me like double-counting. On one hand the authors are saying that cash payers are giving up $151, and on the other hand they are saying that this added burden is a benefit to card users. This seems logically inconsistent to me. If anything, the amount of rewards should be reduced by the premium that card users have to pay to derive their benefit, no?
But in fact the authors of the paper do just that. The amount of rewards is calculated as the difference between what card-holders should pay and what they do pay. It is reduced by the amount they’re paying to derive their benefit — but even after that reduction, the result is still large and positive.
Chen tries another tack:
Another way to look at it is that if all merchants started passing that fee entirely onto me, then I am receiving 0.75% in rewards and paying 2% in fees. In this case, wouldn’t the paper state that I’m theoretically still receiving a benefit from these rewards, even though I’m losing money in practice?
The paper is entirely consistent with a world where cardholders are “losing money in practice”. It just says that if they’re losing money, they’re losing less money than they should be. And people paying cash are losing more money than they should be.
Finally, Chen writes:
The only other thing I would point out is that the Durbin Amendment did give merchants the legal right to offer discounts to cash users, which is the same things as applying a surcharge to card purchases. And even the Boston Fed’s paper states that many merchant agreements allowed this practice beforehand as well. So I’ve always wondered, why don’t more merchants take advantage of it?
No, it isn’t the same thing at all. My business American express card comes with whopping great interchange fees — much more than my free Citibank Mastercard. If surcharges were legal, then a store could happily charge me more if I paid with my Amex than if I paid with my Mastercard. And that would be fair. But you can’t do that with cash discounts. The point about surcharges is that they would and should be used to discourage people from using the cards with the highest interchange fees. Merchants are perfectly happy to accept credit cards with very low interchange fees. And in order to be able to make the distinction, a cash discount isn’t good enough: you can’t offer me a cash discount for not using my Amex, if I don’t get a cash discount for not using my Mastercard. That’s why we need merchants to be able to impose surcharges, rather than just discounts.