An Analysis of Credit Card Pricing Disparities Between Large and Small Issuers

The Consumer Financial Protection Bureau published a Data Spotlight on Feb. 16 indicating that small credit card issuers offer much lower interest rates compared to the largest issuers. The report makes use of pricing data from the recently expanded Terms of Credit Card Plans (TCCP) survey, which is conducted by the CFPB every six months.[1]

The added detail in the expanded TCCP survey improves the usefulness of the collected data by offering more granular information on credit card fees and terms and also by enabling a more controlled comparison of credit card terms across issuers. However, rather than make best use of these data, the CFPB conducts a crude analysis from which the agency draws simplistic, misleading conclusions about pricing disparities between large and small issuers. This note demonstrates that the reported findings do not hold up under a more rigorous analysis.

The CFPB estimates a huge difference in interest rates between large and small issuers based on a comparison of program median APRs within credit score bands.[2] The CFPB finds that the median of program median APRs by score band of the largest (top 25) card issuers exceeds that of smaller issuers by 8 to 10 percentage points, depending on the specified score band. The agency equates this to “an average savings of $400 to $500 a year for a consumer with an average balance of $5,000 using a small bank or credit union’s card.”

However, the CFPB’s analysis exhibits multiple flaws, hence falling well short of the standard of clarity and rigor one should expect from a regulatory agency. One important shortcoming is the failure to separate small banks and credit unions, obscuring an important distinction critical for understanding APR differences. Another is the failure to separate out specialty card programs—retail co-branded, travel and entertainment, credit builder, secured and so on—which are almost entirely within the purview of the top 25 issuers, again obscuring important distinctions. A third is the failure to control for the range of consumer credit scores that each program targets or spans, although this information is available in the TCCP data. Finally, the agency gives no consideration to how the credit limits offered to consumers, or other factors absent from the TCCP data, might affect reported APRs.

Using the most recent available data from the TCCP survey (the March 2023 release), we find that APR differences between the top 25 and smaller issuers shrink dramatically after accounting for effects due to credit unions and specialty card programs.[3] This finding is represented in Figure 1, which also applies a more controlled approach to comparing APRs across programs compared to the analysis conducted by the CFPB. Specifically, for Figure 1, we restrict the sample to plans with similarly situated risk profiles—in this case those spanning the highest score range (720 or greater) plus one or more lower score ranges, and we focus on minimum APRs (those charged to the highest credit score customers).[4] In addition to providing a more controlled comparison, this approach has the advantage that program minimum APRs are better populated in the data than median APRs by score band.

Figure 1: Median of Program Minimum APRs, by Issuer Size and Sample Specificity

fig 1 - median program min

Prior to separating out credit unions and specialty card programs, the median minimum APR is 21 percent for the largest issuers and 14.4 percent for smaller issuers, as shown by the first set of bars in Figure 1. This 6.4 percentage point gap is nearly as wide as those highlighted in the CFPB report.

On excluding credit unions (reflected in the second set of bars), the gap narrows to 4.0 percentage points, reflecting the relatively low APRs that credit unions offer to their members on credit card plans. With further refinement to address confounding factors—removal of specialty programs—the difference in minimum APRs shrinks to slightly over 1 percentage point.

The results in Figure 1 indicate that the difference in APRs for comparable credit card plans between the top 25 issuers and smaller issuers is much smaller than the CFPB’s report suggests. Therefore, this discrepancy is not indicative of a lack of market competition or “practices that imply anti-competitive behavior” as insinuated by the CFPB. It is more plausible that this difference reflects factors omitted from the analysis due to lack of information, such as differences in credit limits. More specifically, it is likely that the largest issuers are offering increased credit limits in return for higher APRs.

Consistent with this view, Figure 2 shows the median of program minimum APR by issuer size for three card program classifications characterized by increasingly strict credit standards.[5] Note that the first two categories (full risk spectrum, and 620 or greater) together yield the sample for Figure 1, while the third (720 or greater) corresponds to programs restricted to super-prime consumers. For programs that encompass the full credit risk spectrum, the largest issuers offer lower APRs compared to smaller issuers. As the programs become progressively more restrictive based on credit score, the largest issuers offer higher APRs, while smaller issuers offer lower APRs.

Consequently, for the more restrictive programs the largest issuers’ APRs substantially exceed the APRs of smaller issuers. The most intuitive explanation for why APRs rise as program credit standards tighten in the case of the largest issuers, causing them to exceed the APRs of smaller issuers, is that their higher APRs are accompanied by increased credit limits.

Figure 2: Median of Program Minimum APRs, by Issuer Size and Program Score Range

fig 2 - median program min

A final important point is that contrary to the CFPB’s unsubstantiated claim that the TCCP survey incorporates a “representative sample of smaller issuers,” there is reason to believe this is not the case. Indeed, there is reason to believe that the TCCP data underrepresents the population of small banks that rely on third parties to implement and manage their credit card programs and that consequently, the agency’s analysis likely overstates the degree to which lower-rate cards are available from smaller banks.

In the remainder of this note we elaborate on the analysis underlying Figures 1 and 2. In addition, we briefly examine median APRs by score range and the question of representativeness of the small issuer sample.

Why Credit Unions Are Not Comparable to Banks

Credit unions are a distinct category of financial institutions that serve market niches. As seen in Figure 1, the initial large gap observed between the APRs of the largest versus smaller issuers for the sample to a major degree reflects inclusion of credit unions in this comparison.

Credit card plans offered by credit unions comprise the majority of plans from smaller issuers in the TCCP sample, and typically offer much lower interest compared to plans offered by banks and consumer finance companies. The TCCP sample contains 83 credit card plans offered by credit unions, of which 15 are associated with three credit unions, Boeing Employees, Pentagon Federal and Navy Federal, that are among the top 25 issuers.[6]

There are multiple distinguishing aspects of credit unions that enable them to offer lower-rate credit cards and other consumer credit products. [7] To begin with, credit unions are not-for-profit organizations that are owned by their members, who have deposit accounts at their credit union. The profits generated by lending activities are exempt from income taxes and are returned to members, typically in the form of lower rates on loans and also through interest on deposits.

Members of a credit union generally have a common relationship as employees of the same company, government entity, branch of the military or other organization, or as residents of a local community served by the credit union. Membership is not open to outsiders.

The underlying common relationship of members in a credit union often implies that their incomes are comparatively predictable and verifiable. Credit unions can employ relationship-based underwriting, such as by restricting eligibility for a credit card to those who have maintained a stable deposit relationship (free of non-sufficient fund items, for example) for a minimum number of months.

These factors mitigate credit risk and may further enable credit unions to offer lower interest rates within any given range of credit score. Moreover, most credit unions are small, implying that they have the capacity to issue only cards with relatively small credit lines, which also may mitigate delinquency risk.[8]

Because of these factors, a credit card APR comparison between the largest issuers and smaller issuers that does not separate out credit unions, such as conducted by the CFPB, is misleading. Clearly, the fact that credit unions offer lower rates on credit cards when compared to the largest issuers does not imply that competition is lacking.

Distinguishing Among Types of Credit Card Plans

There are a variety of types of specialty credit card plans, including the two most familiar categories, retail co-branded and travel- or entertainment-related, and others such as secured cards that are intended for narrow population segments. Pricing can differ between these specialized card plans and generic cards, as well among the specialty plans, due to differences in the nature of the benefits provided or because of differences in consumer spending patterns and repayment behaviors that affect delinquency risk.

In some cases, the specialty plan categories contain too few observations from smaller issuers to support a comparison to the top 25. In other cases, the categories are small overall and populated by heterogeneous, customized programs that are not amenable to comparison across issuers. For these reasons, it is appropriate to exclude them from the analysis.

The retail co-branded category. Retail co-branded cards have the distinction of being issued in partnership with major retailers, automobile companies, gasoline companies, sports leagues or teams and wireless communications companies. These cards carry the name and logo of the company or organization affiliate and offer specialized rewards or other benefits (such as promotional interest rates) for purchases from the affiliate. The TCCP sample contains 124 retail co-branded cards, almost all of which are from the top 25 issuers.

Consumers who sign up for these customized programs might be adversely selected (tend to be higher-risk than is ex-ante evident based on credit scores or other observable measures). Alternatively, or additionally, consumers who have both a retail co-branded and a generic card may prioritize repayment of the latter if facing financial difficulties. Thus, these programs may tend to have higher delinquency rates and consequently be priced differently.[9]

Travel and entertainment cards. The TCCP sample contains 73 travel and entertainment cards, again nearly all from the top 25 issuers. These have the distinction of being issued in partnership with major airline or hotel companies or other major companies specializing in travel or entertainment. As in the case of retail co-branded cards, and for similar reasons, these programs may tend to have higher delinquency rates and consequently may be priced differently.

Secured and credit builder cards. One type of specialty card program is secured cards, which are intended for borrowers with elevated credit risk. They are backed by a security deposit and typically have a high APR. Other types of programs also target individuals with low or no credit scores, but with unsecured credit. These include student and “credit builder” type cards, which collectively we refer to as the credit builder category. The TCCP sample contains 103 secured cards and credit builder cards in total, all but two of which are issued by top 25 institutions.[10]

The bank partner co-branded category. The TCCP sample also contains 69 credit cards issued by First National Bank of Omaha, a top 25 issuer, that are offered through smaller banks, which we refer to as bank partner co-branded. Although the CFPB counts them among plans offered by the largest issuers, the bank partner co-branded cards are distinct. They are marketed by community and regional banks with their own name and logo, primarily to existing and potential new customers of these banks within the regions where they operate. Moreover, they artificially multiply the number of generic plans in the sample that are offered by First National Bank of Omaha, as many are replicates of the same card program. Thus, it is appropriate to separate them out from the analysis.

Other specialty programs. Additional specialty programs include cards offered only to select, existing customers of a bank (such as wealth management customers), and programs targeting membership organizations and professional line of work categories—most of which are for labor union members.[11] There are a relatively small number of such programs, which are selective or limited to narrowly defined customer bases and therefore are appropriate to separate out.[12]

Controlling for Program Credit Risk Profile

As noted previously, the CFPB failed to consider that there may be differences between card plans based on consumers’ credit risk profiles. The TCCP data provide information about the range of consumer credit scores that characterize each card program, allowing for a comparison of pricing across issuers with broadly similar risk profiles.

Specifically, the TCCP data indicate which of the following FICO® score range or combination of ranges consumers primarily are drawn from: No score, 619 or less, 620 to 719, 720 or higher. The data provide minimum and maximum APR for the plan as well as the median APR within each score range.

To construct Figure 1, we first select the card plans that span the highest score range (720 or greater) plus one or more lower score ranges. This ensures that the plan serves a diversified customer population that includes the highest score range. We then compare program minimum APRs between the largest and smaller issuers for this set of card plans.

One important advantage of this approach is that minimum APR are more fully populated in the TCCP database than median APRs by score range (fewer missing data points).  In addition, our approach overcomes an important limitation of the CFPB’s approach, which is that median APRs within broad score bands may not be comparable across card plans with different tiering structures.

For example, a card plan narrowly targeted to higher income consumers with super-prime credit scores—the 720 and above range—cannot be considered similarly situated to more expansive programs for a comparison of median APRs within this range. Plans that extend more broadly across the risk spectrum of consumers would likely have more customers concentrated at the lower end of this range, implying a lower median APR relative to the narrowly targeted program. Thus, it is preferable to compare APRs across programs with sufficiently similar target populations.

Moreover, the minimum APR for a program is the APR associated with the upper bound of the program’s score range. Thus, use of the minimum APR allows for a consistent comparison across similar programs. In contrast, median APRs by score range will vary with the credit quality distribution of consumers within each range, which may be less consistent across programs.

While in principle, the analysis of Figure 1 could be repeated for card programs distinguished by other risk profiles, such as programs limited to super-prime borrowers, this is not possible because of data limitations. The similarly situated programs selected for Figure 1 are the only ones containing a sufficient number of plans within each issuer size category to conduct a meaningful comparison.[13]

Drilling Down Further

The modest remaining gap shown in Figure 1 might reflect relatively large credit lines being offered by the largest banks, or other factors not observable in the data. That is, the largest banks may be providing higher credit limits or other expanded benefits, while offsetting the resulting, greater credit risk associated with larger balances (or the cost of the additional benefits) with higher interest rates.

We can shed light on this and a few other issues by drilling down further based on program credit risk profile; that is, based on the target range of consumer credit scores for each program as recorded in the TCCP data. Table 1 shows the median of program minimum APRs (after exclusion of credit unions and the specialty programs) and number of plan observations for the three risk profile categories that are sufficiently populated for reasonable comparisons across the categories.

Table 1:  Program pricing across program categories differentiated by risk profile

table 1 - program pricing

The first two plan categories in Table 1 together correspond to the sample for Figure 1, as they span the highest score range (720 or greater) plus one or more lower score ranges. The bottom category represents programs that focus on super-prime consumers. The three plan categories in Table 1 provide the data points for Figure 2.

Thus, we see that the remaining gap between the APRs of the top 25 and smaller issuers as of the final refinement applied in Figure 1 arises from the category with stricter credit standards, occupying the next to last row in Table 1. A yet wider differential is observed for the most restrictive category, occupying the bottom row and associated with super-prime consumers. The latter finding should be viewed with caution, however, since this category contains only three plans from smaller issuers.

Reiterating our earlier point, the most plausible explanation for why APRs rise as program credit standards tighten in the case of the largest issuers, causing them to exceed the APRs of smaller issuers, is that their higher APRs are accompanied by increased credit limits or other expanded benefits. Thus, the remaining APR gap between the top 25 and smaller issuers as of the final refinement applied in Figure 1 likely reflects higher credit limits or other benefits provided by the largest issuers.

A look at median APRs by risk tier. The CFPB analysis focused on a comparison of program median APRs by credit score range, whereas thus far our analysis has looked at program minimum APRs. Figure 3 extends our discussion to the case of program median APRs for the same two program categories populating Figure 1 (those spanning the full score spectrum those spanning 620 or greater) after exclusion of credit unions and specialty programs.[14]  Specifically, Figure 3 plots the median of program minimum APR along with the median of program median APRs by score range (720 or greater and 620 to 719) for each issuer category.  The sample is restricted to issuers that report all three APR measures, resulting in a smaller sample than used for Figure 1 (due to missing data on median APRs from some issuers).

Figure 3: Median APRs by credit risk tier

fig 3 - median APRs

Due to the dropped observations, the medians of program minimum APR in Figure 3 differ from those indicated by the last set of bars in Figure 1 (with no longer a gap between the largest and smaller issuers). That issue aside, Figure 3 shows that the APRs of the top 25 issuers are more risk-sensitive, rising more consistently from the program minimum APR through the risk segment medians.

This finding, however, has no implications for market competitiveness. Rather, it suggests that smaller issuers have less appetite for risk and apply stricter credit standards along dimensions other than score (look for offsetting factors in approving card applicants) in the lower score ranges. Alternatively, smaller issuers may reduce credit limits for consumers in the lower score ranges compared to what the largest banks are willing to offer.

Do the TCCP Data Incorporate a Representative Sample of Smaller Issuers?

A final important point to consider is the extent to which the TCCP’s sample of credit card plans of smaller banks and non-depository financial companies are representative of all such credit card programs in the U.S. Their presence within the sample is tiny compared to the overall number of regional and community banks in the U.S. If these programs are non-representative, then comparisons between them and those of the largest banks may not be generalizable. In particular, the observed APR gaps between the largest versus smaller issuers would be overstated to the degree that the TCCP’s sample contains overrepresents issuers with lower-rate programs.

One compelling reason to suspect that the sample may have such an overrepresentation is that numerous small banks—far more than are present in the TCCP sample—offer generic cards in partnership with third-party agents or platforms. The latter may be agent banks, as exemplified by the bank partner co-branded cards issued by First National Bank of Omaha. They also can be nonbank credit card service providers such as Elan Financial Services,  a company that claims to have partnerships with more than 1,200 financial institutions in addition to 250 credit unions “to provide an outsourced credit card program.”

Based on available information, it appears that the APRs associated with card programs of smaller issuers that are outsourced to Elan Financial Services roughly align with those offered by the largest banks. According to the webpages of a few small banks and credit unions identified by CreditKarma.com as offering credit card plans in partnership with Elan Financial Services, they all offer plans with minimum and maximum APRs of 18.24 and 29.24, respectively.

Summary

A recent analysis conducted by the CFPB using TCCP data finds a huge gap—8 to 10 percentage points depending on credit score tier—between the median APR of large (top 25) issuers versus smaller issuers. However, the CFPB’s findings do not hold up under a more rigorous examination that better controls for important factors neglected in the agency’s analysis.

When we control for differences due to credit unions, program type and program credit risk profile, we find only marginal gaps between the largest and smaller issuers. For the minimum APR across comparable card programs, a reasonably consistent comparison metric that mitigates pricing variation tied to differences in borrower credit risk, we observe a median across issuers of 18.2 percent for the 25 largest issuers and 17.0 percent for smaller issuers once these factors are accounted for.

These gaps are much narrower than reported by the CFPB and likely reflect more generous credit limits provided by the larger banks, or other factors on which information is lacking in the data. Moreover, there is good reason to believe that the TCCP data are non-representative of smaller issuers, such that the agency’s study likely overstates the degree to which lower-rate cards are available from smaller issuers.


[1] Beginning in July 2022, the CFPB enhanced the TCCP survey to collect much more detailed information, including on variation in card interest rates across consumer credit risk segments and on card availability (requirements for opening an account). The expanded set of data items collected has been incorporated into the two most recent public releases of the data.

[2] The TCCP data provide information about the range of consumer credit scores that characterize each card plan. Specifically, the TCCP data indicate which of the following FICO® score range or combination of ranges consumers primarily are drawn from: No score, 619 or less, 620 to 719, 720 or higher. The data indicate the median APR within each applicable score range and also provide the plan’s minimum and maximum APR.

[3] Our analysis excludes a few observations for which the reported APRs were deemed unreliable. These include three credit union programs with extremely low reported minimums that were inconsistent with the APRs currently posted on the issuers’ webpages. In addition, we dropped two cards from smaller issuers that are not credit unions and two from the top 25 due to reported minimum APRs that were inconsistent with other reported APR data for the same card (program maximum and medians by score range) and which deviated far from what was posted on the issuers’ webpages.

[4] In contrast, the CFPB compares median APRs by score range without controlling for a program’s risk profile, the third flaw among those enumerated above.

[5] The sample for Figure 2 retains all of the restrictions considered in Figure 1, except for relaxing the last one by including programs limited to super-prime customers (credit score of 720 or greater).

[6] Across card plans offered by credit unions the median of program minimum APRs is 13.5 percent, which compares to a median of 18 percent for program minimum APRs of plans offered by the group of smaller issuers. For federal credit unions, interest rates inclusive of all finance charges are capped by statute at 18 percent. However, the TCCP data show that the cap is usually not binding and even aside from the cap, credit unions are not comparable to other issuers for the reasons enumerated below.

[7] Credit unions also tend to offer lower rates on auto loans and unsecured personal loans. Comparisons of bank versus credit union loan and deposit interest rates are published quarterly by the National Credit Union Administration.

[8] One reason that smaller credit lines may mitigate delinquency risk is that consumers with multiple loan balances who are facing financial stress might prioritize repayment of their smaller loans. For evidence consistent with this notion, see Calem, P. S., & Sarama, R. F. (2017), “Why mortgage borrowers persevere: An explanation of first and second lien performance mismatch.” Real Estate Economics45(1), 28–74.

[9] The median of program minimum APR across the retail co-branded cards in the sample is 24.7.

[10] The median of program minimum APRs across these card plans is 28.0.

[11] This group also includes cards for other membership organizations; cards for nurses, educators and Japanese expats; and a medical credit card.

[12] There are 26 such credit card plans in total; all but two are from the top 25 issuers.

[13] There are only four plans that span the score ranges 619 or less or no score plus 620 to 719. There are only five plans specialized for consumers with scores ranging from 620 to 719, after excluding two for which the APR data were determined to be unreliable. Plans specialized for consumers with scores of 620 or less or no scores all are contained within the specialized, secured and credit builder categories.

[14] Recall from our previous discussion that other card programs—those distinguished by other risk profiles—are not amenable to the same analysis due to insufficient numbers of observations.