Are the Poor Subsidizing the Rich via Credit Card Rewards? Don't Be So Sure

Do credit card reward systems subsidize the rich at the expense of the poor? There has been a lot of press over the last week about a new research report out of the Boston Federal Reserve that claims that credit card payments are a form of regressive wealth transfer, taking money from the poor in the form of higher prices and giving to the rich in the form of credit card rewards.

But before we all start crying foul and cutting up our credit cards, let's put down our pitchforks, take a deep breath, and think about this. It's not enough to just read the abstract and start writing eye-catching headlines; let's take a moment to understand how they came up with these numbers.

To start, let's just read the first couple of lines of the paper's abstract. 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.

A large part of this overestimation comes from their calculation of credit card transaction costs. In order to derive a number, the authors have to make assumptions about how much money Americans spend each year on products where this cash-to-card transfer happens. In Appendix A, they tell how they derived these consumption numbers from the 2007 Personal Consumption Expenditures: by subtracting out a number of categories that they assumed did not fit the bill, like health care, financial services, and "food produced and consumed on farms." This caused them to remove $2.66 trillion from the total PCE value of $9.83 trillion. However, they did not remove categories like housing and automobile purchases, which are generally not payable by credit card, either. These would have reduced their estimates by more than $2 trillion more and reduced their transfer results by more than 9%.

When asked why they left these categories in their calculations, their statement to us was, "In the case of mortgages and cars, the best course of action is to obtain data on actual credit card spending by merchant and product. We have not been able to get these data, however." As a simple check, we called the top three mortgage lenders and confirmed that they do not accept credit cards, and a 2008 CNW research report shows that only one in nineteen auto purchases involve any credit card transactions. So when faced with an absence of data, the authors arbitrarily decided to assume that everyone has the option to pay their mortgages and car payments with credit cards, which is clearly not the case.

Another assumption that I have to question is that merchants pass through the full merchant fee in their retail prices, on page 16. If we assume that every merchant passes on an equivalent fee, then we have to assume that high-income card users shop at the same places as low-income cash users, which I would argue is not true. Instead, I would think that few of the lower-income families spend their cash at Saks Fifth Avenue, and few of the higher-income families use their rewards cards at dollar stores. So the "rich" in this model drive up prices primarily at stores where they themselves bear the brunt, while the effect on the "poor" is mitigated.

Lastly, in order to make any argument about a credit card reward wealth transfer, the authors necessarily have to assume that the cost of handling cash is significantly lower than handling credit cards. However, the authors freely admit on page 12 that they are not taking into account any of the fixed costs of handling cash. While the costs of cash transactions are harder to quantify than a strict merchant fee, this does not mean they should be ignored. Some examples include the time cost of cash, which means needing more cashiers on hand, since credit card users rarely even need to sign anymore; the error cost of incorrect change, which is nearly zero for credit cards; the cost of cashier fraud, since it's a lot easier for your employees to steal your cash than your credit card transactions; and the cost of returned checks, since this is money you may never get back. While credit cards are subject to the costs of fraud, this is already baked into the merchant fee. But these costs as they apply to cash are ignored in the paper.

While it is a noble pursuit to help the poor in any way possible, I can't agree with the Boston Fed's approach. Even if you take their word as gospel, they claim that low-income households (≤$20k/yr) give up $23 each year so that the high-income households (≥$150k/yr) can earn $756. Therefore, the "poor" are transferring 0.1% of their incomes so that the "rich" can become up to 0.5% richer. These are hardly numbers that should incite calls for drastic policy changes. Plus, in Section 7 of this paper, they run a scenario analysis where they stress-test their own assumptions, bringing some of them more in line with my arguments above, and show that it would reduce their transfer estimates by 50% or so. If they had done so with all of the assumptions that I question above, the result would be even more drastic and would possibly even eliminate any claims of a transfer.

So when something like this hits the tape and creates a media feeding frenzy, I have to step back and wonder if the conclusions are bold or just absurd. Unfortunately, in this case, I think Mark Twain probably put it best:
"There are three types of lies: lies, damned lies, and statistics."
Tim Chen is the CEO/Founder of NerdWallet -- a credit card search site started with the intention of taking the headache out of finding a credit card online and saving users money in the process.
Do credit card reward systems subsidize the rich at the expense of the poor? There has been a lot of press over the last week about a new research report out of the Boston Federal Reserve that claims that credit card payments are a form of regressive wealth transfer, taking money from the poor in the form of higher prices and giving to the rich in the form of credit card rewards.

But before we all start crying foul and cutting up our credit cards, let's put down our pitchforks, take a deep breath, and think about this. It's not enough to just read the abstract and start writing eye-catching headlines; let's take a moment to understand how they came up with these numbers.

To start, let's just read the first couple of lines of the paper's abstract. 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.

A large part of this overestimation comes from their calculation of credit card transaction costs. In order to derive a number, the authors have to make assumptions about how much money Americans spend each year on products where this cash-to-card transfer happens. In Appendix A, they tell how they derived these consumption numbers from the 2007 Personal Consumption Expenditures: by subtracting out a number of categories that they assumed did not fit the bill, like health care, financial services, and "food produced and consumed on farms." This caused them to remove $2.66 trillion from the total PCE value of $9.83 trillion. However, they did not remove categories like housing and automobile purchases, which are generally not payable by credit card, either. These would have reduced their estimates by more than $2 trillion more and reduced their transfer results by more than 9%.

When asked why they left these categories in their calculations, their statement to us was, "In the case of mortgages and cars, the best course of action is to obtain data on actual credit card spending by merchant and product. We have not been able to get these data, however." As a simple check, we called the top three mortgage lenders and confirmed that they do not accept credit cards, and a 2008 CNW research report shows that only one in nineteen auto purchases involve any credit card transactions. So when faced with an absence of data, the authors arbitrarily decided to assume that everyone has the option to pay their mortgages and car payments with credit cards, which is clearly not the case.

Another assumption that I have to question is that merchants pass through the full merchant fee in their retail prices, on page 16. If we assume that every merchant passes on an equivalent fee, then we have to assume that high-income card users shop at the same places as low-income cash users, which I would argue is not true. Instead, I would think that few of the lower-income families spend their cash at Saks Fifth Avenue, and few of the higher-income families use their rewards cards at dollar stores. So the "rich" in this model drive up prices primarily at stores where they themselves bear the brunt, while the effect on the "poor" is mitigated.

Lastly, in order to make any argument about a credit card reward wealth transfer, the authors necessarily have to assume that the cost of handling cash is significantly lower than handling credit cards. However, the authors freely admit on page 12 that they are not taking into account any of the fixed costs of handling cash. While the costs of cash transactions are harder to quantify than a strict merchant fee, this does not mean they should be ignored. Some examples include the time cost of cash, which means needing more cashiers on hand, since credit card users rarely even need to sign anymore; the error cost of incorrect change, which is nearly zero for credit cards; the cost of cashier fraud, since it's a lot easier for your employees to steal your cash than your credit card transactions; and the cost of returned checks, since this is money you may never get back. While credit cards are subject to the costs of fraud, this is already baked into the merchant fee. But these costs as they apply to cash are ignored in the paper.

While it is a noble pursuit to help the poor in any way possible, I can't agree with the Boston Fed's approach. Even if you take their word as gospel, they claim that low-income households (≤$20k/yr) give up $23 each year so that the high-income households (≥$150k/yr) can earn $756. Therefore, the "poor" are transferring 0.1% of their incomes so that the "rich" can become up to 0.5% richer. These are hardly numbers that should incite calls for drastic policy changes. Plus, in Section 7 of this paper, they run a scenario analysis where they stress-test their own assumptions, bringing some of them more in line with my arguments above, and show that it would reduce their transfer estimates by 50% or so. If they had done so with all of the assumptions that I question above, the result would be even more drastic and would possibly even eliminate any claims of a transfer.

So when something like this hits the tape and creates a media feeding frenzy, I have to step back and wonder if the conclusions are bold or just absurd. Unfortunately, in this case, I think Mark Twain probably put it best:
"There are three types of lies: lies, damned lies, and statistics."
Tim Chen is the CEO/Founder of NerdWallet -- a credit card search site started with the intention of taking the headache out of finding a credit card online and saving users money in the process.