A recent working paper from the Federal Reserve has been increasingly cited by critics of credit cards – particularly credit cards that offer rewards. The authors, who include an economist at the Federal Reserve, use expansive and bold language, claiming that their analysis “contributes to the literature that highlights the role of the financial system in driving wealth inequality.” Lawmakers and news outlets have, accordingly, been relying on the paper’s headline-grabbing abstract to bolster arguments for increased price regulation in credit card interchange markets.

• The paper’s authors apparently felt a need to “dress up” their findings by drawing correlations to other demographic characteristics of borrowers. But, their paper is explicit that rewards cards do not redistribute funds from low-income consumers to high-income consumers.

• By their own fine print, the authors are clear that at the end of the day, their primary finding is that consumers with good credit pay less to borrow and earn more rewards than consumers with bad credit. What is interesting is that the authors demonstrate an actual “Robin Hood” effect in the credit card market. (NOT a “Reverse Robin Hood” effect, to be clear.)

This blog dives into the recent credit card working paper, to separate the facts from the fiction and emphasize a key finding of the author’s research: the vast majority of money that’s paid into rewards cards comes from high-income consumers – with the benefits of rewards cards spread out more evenly across cardholders of all incomes.

The Abstract’s Claims… 

The big headlines around the working paper come from the authors’ claims in the paper’s abstract that “[w]e estimate an aggregate annual redistribution of $15 billion from less to more educated, poorer to richer, and high to low minority areas, widening existing disparities.”

If you were to just read the selection from the abstract above, you could understandably think the paper gives evidence that rewards cards benefit rich consumers at the expense of the poor. But, in reality, the paper tells a very different story.

…What the Paper Actually Says

Notwithstanding the abstract’s bold claims about redistribution from the poor to the rich, the text of the paper specifically refutes the notion of a “Reverse Robin Hood effect.” (pp. 3-4 & 30):

“While credit card rewards are often framed as a “reverse Robin Hood” mechanism in which the poor subsidize the rich, our results show that this explanation is at best incomplete.”

The authors go on to explain that “our results are not driven by income, as they hold within the sub-samples of low-, middle- and high-income individuals.”

Indeed, if you strip out the hyperbolic language from the paper, its core findings are pretty straightforward (p. 30):

• “Super-prime and prime consumers spend more money and thus earn higher rewards, but they also pay back their balances in time and thus incur lower interest payments.”

• “Conversely, sub-prime and near-prime consumers earn lower rewards and incur higher interest payments due to higher outstanding balances on rewards cards.”

In other words,

• Consumers that pay their bills on time pose less repayment risk to lenders and, accordingly, benefit from cheaper interest rates.

• Consumers that don’t pay their bills on time pose a higher risk that they won’t pay back their loans. Therefore, lenders charge them higher interest rates.

Don’t Confuse Correlation with Causation

The authors apparently felt the need to “dress up” their findings about credit scores, by noting correlations between credit scores, on the one hand, and income, race, and geography, on the other hand.

However, the paper ultimately concedes that none of these demographics are responsible for consumers’ different outcomes with credit cards.

For instance, if you read the footnotes (footnote 24, specifically), the authors concede that FICO and income and are only moderately correlated – a point that other Board researchers demonstrated several years ago.

The authors similarly concede that “all coefficients become statistically insignificant and close to zero in magnitude when controlling for a ZIP code’s average FICO score, indicating that differences in financial sophistication are the underlying mechanism driving our geographical results.”

“A rose, by any other name…”

The authors try to re-define what it means to have a good or bad credit score by distinguishing consumers that are “financially sophisticated” and “naïve.” However, the only measure of sophistication or naiveness the authors point to is FICO score.

Scholars have written volumes of papers exploring the relationship between credit scores and financial literacy – as credit scores ultimately reflect a combination of consumer behaviors that roughly reflect both an ability to pay with a willingness to pay. Relevant to this discussion, though, is that the authors made no effort to try to model or otherwise pull apart their relationships. If you look at their data tables, they are simply using FICO scores. Accordingly, they may call credit scores a different name, but that’s all they’ve done in this paper.

No “Reverse” Robin Hood Effect Here – But Evidence of a “Regular” Robin Hood Effect

The authors explain that high-income people with high credit scores get the most value out of their rewards cards, earning on average $20.10 a year. (These consumers earn a lot of rewards on purchases and are less likely to carry a balance, much less pay late.TE)

What’s more surprising is that high-income people with low credit scores pay the most into rewards cards, losing on average $12.75 a year.

The researchers themselves put it plainly, stating (p.4):

“Thus, high-income consumers with high FICO scores benefit from rewards credit cards largely at the expense of high-income consumers with low FICO scores.”

As for low-income consumers:

• Low-income consumers with high credit scores gain, on average, about $9.71 in value each year.

• Low-income consumers with poor credit scores ultimately net negative with their rewards credit cards. But it’s proportionately a much smaller average annual loss – $2.56 a year.

We’ve created a table, using the paper’s data.