www.responsiblelending.org

Joshua M. Frank, Senior Researcher

Center for Responsible Lending

February 16, 2011

Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Methods and Findings on Rates . . . . . . . . . . . . . . . . . . . . . . . . 5

Access to Credit: Evidence from Mail Volume . . . . . . . . . . . . . . . 9

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Endnotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

New credit card rules mandated by the Credit CARD Act of 2009 have resulted in significantly greater price transparency for consumers. This reverses a trend of increasingly unclear pricing that for years misled consumers into believing they would pay less for credit card debt than was true. Inaccurate pricing information likely caused many borrowers to take on more credit card debt than they otherwise would have.

The difference between the stated rate on credit card solicitations and the rate consumers actually paid widened to unprecedented levels by 2004 and stayed at those levels through 2008.

This difference narrowed markedly in the wake of reform, with stated prices on solicitations moving much closer to actual prices. This study also finds that, in the year since the CARD Act’s implementation, actual prices have remained stable and available credit has not tightened beyond what would be expected from the economic downturn. Because price transparency fosters competition, the long-term effect of the CARD Act is likely to be lower costs for consumers.

• New rules have reduced the difference between stated rates and actual rates paid on credit cards, resulting in more transparent pricing. An estimated $12.1 billion in previously obscure yearly charges are now stated more clearly in credit card offers.

• Once the economic downturn is taken into account, the actual rate consumers have paid on credit card debt has remained level.

• Direct-mail offers have been extended at a volume and pace consistent with economic conditions.

Critics of reform often argue that common-sense rules and oversight inevitably lead to significant and negative “unintended consequences” for consumers. The on-going crisis in the mortgage market is an example of the harm the absence of common-sense rules brings. This report shows that, prior to the CARD act, the credit card industry was another.

CRL examined five sets of data for this study. The first two, from

the Federal Reserve Board, track credit card rates, both as stated on

solicitations and as actually paid by consumers. The discrepancy

between the two measurements narrowed after the CARD Act

passed, with stated rates moving closer to actual rates paid.

The third source of information came from “Call Reports,” which

are statements of income and financial condition that commercial

banks file quarterly with bank regulators. Analysis of this data provided

further evidence that prices have become more transparent.

The fourth set, a private data base from Mintel Comperemedia, tracked the number of mail

solicitations over time. It showed that, once adjusted for the downturn, the number of solicitations

has held steady or even risen.

The fifth, from the website CreditCards.com, compared rates offered on all credit cards to those

offered on business credit cards, which are not subject to the CARD Act. The effective rate on

business cards increased relative to consumer cards, further evidence that reform did not cause

price increases.

18

16

14

12

10

Feb-00

Aug-00

Aug-01

Feb-01

Aug-02

Feb-02

Aug-03

Feb-03

Aug-04

Feb-04

Aug-05

Feb-05

Aug-06

Feb-06

Aug-07

Feb-07

Aug-08

Feb-08

Aug-09

Feb-09

Aug-10

Feb-10

Stated Rate (all accounts) Actual Rate Paid (accounts assessed interest)

Federal Reserve

Credit Card Rules

Finalized

CARD Act

Passed

(May 2009)

CARD Act

Implemented

(Feb 2010)

Source: Federal Reserve G.19.

Interest Rate (%)

Gap between Stated

and Actual

Contrary to credit card industry claims, the new rules have not caused prices to increase or access

to credit to fall. Instead, they have benefitted the public by making credit card pricing significantly

more transparent. Price transparency is likely to lower costs long term by spurring competition and

making it harder for issuers to manipulate or arbitrarily raise prices.

After passage of significant reforms like the Credit CARD Act of 2009 (CARD Act), it is important

to evaluate the consequences for consumers. This study examines both whether desired benefits in

the form of increased transparency have been achieved and whether new rules have caused credit to

be restricted, either through higher prices or reduced offers.

CRL examined five data sets in this study. The first two, from

the Federal Reserve Board (FRB), are data on credit card rates,

both as stated on solicitations and as actually paid by consumers.

Beyond rates, these data sets also provide insight into rate

transparency. The third data set came from revenue information

culled from information filed with federal bank regulators

in what are known as “Call Reports.” The fourth set was the

number of mail solicitations over time from a private data base,

Mintel Comperemedia. The fifth set, from CreditCards.com,

was stated offer rates on all credit cards compared to those for

business cards.

All of these data sources independently confirm that the new regulatory framework avoided

unintended consequences. Specifically, they show that new regulations did not constrict credit

either through price changes or through reduced offers. On the intended effects, however, the

data show that prices have become much more transparent under the new rules. For three different

regulatory event dates there was a decrease in the difference between the stated rates disclosed to

consumers and the actual rates paid.

• New rules have reduced the difference between stated rates and actual rates paid

on credit cards, resulting in more transparent pricing. An estimated $12.1 billion in

previously obscure yearly charges are now stated more clearly in credit card offers.

• Once the economic downturn is taken into account, the actual rate consumers have

paid on credit card debt has remained level.

• Direct-mail offers have been extended at a volume and pace consistent with

economic conditions.

actually pay at any given time than a third measurement, frequently cited in the media, of average

average open account, not just what consumers are offered in the mail. As pointed out recently by

law professor Bob Lawless, the FRB series titled “accounts assessed interest” paints a much more

benign picture of what has happened with interest rates relative to what has been reported in recent

The FRB tracks two data series on credit card charges. One is referred to as the interest rate on “all

accounts.” The second is the interest rate on “accounts assessed interest.” Given that multiple rates

can apply to any given account, it is this latter number which most accurately reflects the interest

rates that cardholders actually pay.

While the interest rate on “accounts assessed interest” rose briefly just after passage of the CARD

Act, it has declined since then to a level ¾ of a percentage point lower than it was at the time of

the law’s passage. There has been a slight (0.3 percentage points) increase in the rate for accounts

assessed interest when comparing the quarter when the Federal Reserve Rules (which included some

of the CARD Act’s reforms) were implemented to the most recent rate. In contrast, the rate on “all

accounts” has increased 1.4 percentage points over this same period. Between 2004 and 2008 there

was also a large and persistent gap between these two data series, with the “accounts assessed interest”

consistently showing a higher rate. This gap has narrowed considerably since the CARD Act

was passed. The most recent data shows a gap of 0.2 percentage points while, at its peak, the gap

was 2.3 percentage points.

What should we make of this? The key is in knowing what the difference between these two sets of

numbers represents. Despite the names of these data series, more important differences exist between

these two sets of measurements than whether they include accounts assessed no interest. In fact, the

two numbers have a completely different basis for calculation. The “all accounts” data set is based on

the stated purchase rate averaged by each reporting bank across all accounts, while the “accounts

assessed interest” data set is based on actual finance charges collected by each bank divided by the

balances that incurred finance charges. These finance charges do not include fees but do reflect

interest rates on other balances in addition to the purchase rate. This leads to a number of important

distinctions between the “all accounts” and “accounts assessed interest” as summarized in Table 1.

Measures rate in credit card

new purchases

Includes all accounts at

Measures actual charges

Includes cash advance and

Includes penalty rates and

Weighted by balances to

Card Comparison

Websites used in

Media Reports

FRB Stated Rate

(All Accounts)

FRB Actual Rate Paid

(Accounts Assessed Interest)

As Table 1 suggests, “accounts assessed interest” is a far more accurate gauge of the price consumers

pay to borrow on credit cards. It also implies an interpretation for the shifting gap between the

two rates is troubling because it would cause consumers to pay more than they had anticipated and,

likely, to take on more credit card debt than they would have assumed given accurate disclosures. To

keep the most important difference between the two data series clear, the FRB’s “all accounts” rate

will be referred to as the “stated rate” while the FRB’s “accounts assessed interest” series will be

referred to as the “actual rate paid.”

Figure 1 clearly shows that the actual rate paid has declined since the CARD Act. In addition, it

shows a large and persistent difference between the prices disclosed to consumers and those actually

paid in the time period from 2004 to 2008. Since the CARD Act, the two rates have moved much

closer together, suggesting that prices have become more transparent. In fact, just before the FRB

finished their new credit card rules, the difference in these two rates stood at 1.7 percentage points,

implying that consumers paid $16.3 billion more than their stated interest rates would suggest. Just

after implementation of the CARD Act, the difference was down to 0.4 percentage points. Using

recent figures on total credit card debt outstanding, this implies that consumers paid $12.1 billion

less annually in unanticipated finance charges compared to what they would have paid before credit

18

16

14

12

10

Feb-00

Aug-00

Aug-01

Feb-01

Aug-02

Feb-02

Aug-03

Feb-03

Aug-04

Feb-04

Aug-05

Feb-05

Aug-06

Feb-06

Aug-07

Feb-07

Aug-08

Feb-08

Aug-09

Feb-09

Aug-10

Feb-10

Stated Rate (all accounts) Actual Rate Paid (accounts assessed interest)

Federal Reserve

Credit Card Rules

Finalized

CARD Act

Passed

(May 2009)

CARD Act

Implemented

(Feb 2010)

Source: Federal Reserve G.19.

Interest Rate (%)

Gap between Stated

and Actual

To isolate the effects of reform from those of other economic

factors on interest rates, a regression methodology

are provided in more detail in Appendix A. The regression

results confirm that there is no evidence of a significant

price increase associated with credit card reforms. There is,

however, evidence consistent with improved transparency.

The regressions test for the impact of three key regulatory

events on interest rates. The first event is the announcement

of the FRB’s final credit card rules in December

2008. Although the CARD Act went further than the

FRB’s rules by adding and extending a number of rules, most of the key interest rate-related

provisions existed in some form in the rules. The second event is passage of the CARD Act in May

2009. The third event is the main CARD Act implementation date of February 2010, when most

of the Act’s provisions went into effect. None of the three dates led to significantly higher actual

rates paid, even though each event led to significantly higher stated rates. Even more importantly,

perhaps, all three events saw a decrease in the difference between the stated rates disclosed to

Across regression models, the results consistently indicate that there was no real change in interest

rates associated with any of the three regulatory events. In addition, the results are consistent with

the improved transparency in credit card pricing associated with reforms, but whether that change is

statistically significant depends on the regression model used.

Figure 2 summarizes these results. While stated rates show increases, actual rates paid remained more

figure) after all three regulatory dates. If there was a significant change in the stated rates, it appears

to be merely a better reflection of the rates that consumers were already paying.

Interest Rate Change Linked to Regulation

(percentage points)

2.5

2

1.5

1

0.5

0

-1

-1.5

-2

-2.5

-0.5

Fed Rules Finalized CARD Act Passed CARD Act Implemented

Stated Rate Rate Actually Paid

Dierence between Stated

and Actual Rate

Note: Results from regression analysis of interest rate level. All “Stated Rate” and “Dierence between Stated and

Actual Rate” changes are statistically signi-cant. None of the changes to “Rate Actually Paid” are statistically signi-cant.

While the results based on the two FRB data series explored above are convincing, it is useful to

corroborate those findings with an independent data set. To do so, this study includes an evaluation

two-thirds of outstanding credit card balances. The figure shows the credit card revenue for these top

4 issuers averaged and compared to the FRB data. There appears to be some evidence of increased

transparency among these particular credit card issuers. In the two years before the CARD Act’s

passage, the stated rate was on average 1.7 percentage points lower than average revenue for the

Top 4 issuers. However, in the two quarters since implementation, stated rate was 0.5 percentage

points higher than average revenue.

2007Q1

2007Q2

2007Q3

2007Q4

2008Q1

2008Q2

2008Q3

2008Q4

2009Q1

2009Q2

2009Q3

2009Q4

2010Q1

2010Q2

0

2

4

6 8

10

12

14

16

18

Average Revenue Top 4 Issuers Fed Stated Rate Fed Eective Rate

% of Balances

One way that issuers could have kept actual rates paid level over a reform period would have been to

tighten credit. We search for evidence of constrained credit by evaluating the volume of new direct

mail solicitations. Figure 4 shows mail volume using data from Mintel Comperemedia.

At the time the FRB made their credit card rules final, mail volume was already declining in tandem

have no connection with declining mail volume. In fact, before adjusting for economic conditions,

the Act’s implementation appears to coincide with a period of rising mail volume.

Because the results presented above are based on aggregate mail

volume, it could be argued that more and more offers are simply

going to fewer and fewer people. However, data from Mintel

(about the time of the FRB Rules), the number of people receiving

at least one credit card offer has increased from about 40% of

households (based on their sample) to about 60%. There was a

decline in the number of people receiving an offer before 2009;

however, the timing suggests that this was clearly driven by

economic conditions, not regulation. In fact, regulatory reform

coincided with a period when not only more offers were mailed,

but those offers were being sent to a broader group of consumers.

One could also argue that the increasing mail volume was less than it would have been in the

absence of reform. To test this hypothesis, we use regression models to control for outside factors.

for regulation’s impact are summarized in Figure 5, with more details provided in Appendix B. The

CARD Act’s passage and implementation is significantly linked to mail volume, but in the positive

direction. The law’s passage may have reduced uncertainty for issuers by clearly defining what the

new ground rules would be, leading to more offers. Before the law’s passage it is possible that mail

volume was reduced because of uncertainty in the regulatory environment in addition to even

greater uncertainty about the future of the economy. The reason implementation is linked to more

100

200

300

400

500

600

700

800

900

Jan-05

Apr-05

Jul-05

Oct-05

Jan-06

Apr-06

Jul-06

Oct-06

Jan-07

Apr-07

Jul-07

Oct-07

Jan-08

Apr-08

Jul-08

Oct-08

Jan-09

Apr-09

Jul-09

Oct-09

Jan-10

Apr-10

Jul-10

Federal Rules Finalized

CARD Act Passed

CARD Act

Implemented

Source: Mintel Comperemedia; CRL calculations.

Monthly Mail Volume (millions)

Another way to determine whether the CARD Act increased interest rates is to compare credit

card rates offered to consumers to those offered to businesses. Both the FRB rules and the CARD

Act apply only to consumers. Therefore, if the two customer groups show different rate patterns, this

may result from regulation that only affected consumers. Because the FRB data does not separate

business cards from consumer cards, we must rely on data from the credit card comparison and

Fed Rules Finalized CARD Act Passed CARD Act Implemented

Mail Volume (Millions of Pieces)

-50

50

100

150

200

250

300

0

Note: Results from regression analysis. Both positive changes linked to CARD Act passage and

implementation are statistically signicant. The decline linked to Federal Reserve Board Rule

Finalization is not statistically signicant.

Figure 6 shows the FRB rates previously presented alongside CreditCards.com data showing

the quarterly average offer rates to businesses and consumers. Offer rates on all credit cards

The offer rate rose on both business and consumer cards following the CARD Act. As shown in

Figure 7, the business card offer rate did rise somewhat less than credit card offers. However, if

consumer offer rates are becoming more transparent, this would be expected. Business cards still

are allowed to use the back-end rate increase tactics the CARD act outlawed for consumer cards.

Therefore, consumer card offer rates may have appeared to rise in part because offer rates more

in new offer rates overall to the change in the actual rates paid from the FRB, we can estimate the

size of this difference in transparency and adjust for it. As shown in Figure 6, when adjusting for the

estimated difference between offer rates and the rates actually paid (based on what consumers experienced),

business cards had a far larger actual rate increase than consumer cards, despite the fact that

they had no change in regulations. The rate increases shown in Figure 7 are at least in part the result

of economic conditions rather than rule changes. (Previously it was found that actual consumer rates

did not increase significantly after controlling for economic conditions.)

Stated Rate (Fed)

Aug-07

Oct-07

Dec-07

Feb-08

Apr-08

Jun-08

Aug-08

Oct-08

Dec-08

Feb-09

Apr-09

Jun-09

Aug-09

Oct-09

Dec-09

Feb-10

Apr-10

Interest Rate Paid (Fed)

Business Card Oers (creditcards.com) All Credit Card Oers (creditcards.com)

Interest Rate

8 9

10

11

12

13

14

15

16

Because business cards possibly are more sensitive than consumer cards to economic conditions,

regressions were run to evaluate whether such a difference existed. Neither the CARD Act’s passage

nor implementation had a statistically significant impact on the gap between business and consumer

card rates. The sign on the coefficient for these events was not even consistently positive, which it

would have to have been if the new law had caused consumer credit to constrict. More details on

regression results and methodology are provided in Appendix C.

Business Card Oer Rate

(creditcards.com)

All Cards Oer Rate

(creditcards.com)

Business Card

Estimated Actual Rate

Note: Change in rate is from period before Federal Reserve -nal card rules (3rd Quarter 2008) to the

most recent period (2nd Quarter 2010). Business Card Estimated Actual Rate is adjusted to take into

account the change in gap between actual rates and oer rates.

5

4.5

4

3.5

3

2.5

2

1.5

1

0.5

0

Change in Rate (Percentage Points)

The evidence from FRB interest rate data, businessversus-

consumer offer rates, and mail volume all

lead to one conclusion. Recent reforms have delivered

increased transparency without any evidence

of unintended consequences. In fact, there is no

evidence that the CARD Act or the prior FRB

credit card rules burdened consumers either with

led to more transparent disclosure of interest rates, with the stated rate more closely matching the

interest actually paid by consumers.

The public debate on regulation is often framed using a simplified caricature of economic theory,

portraying new rules or oversight as inevitably causing significant, negative “unintended consequences”

for consumers. While it is true that in a perfect market increased costs to businesses will be

passed on to consumers, most markets are far from perfect, and the credit card market is—and long

has been—more imperfect than most. Besides being highly concentrated, prior research has shown

that credit card issuers have relied consistently on behavioral biases and confusing, complex pricing

issuers that may be above what a competitive market would dictate. In 1991, Ausubel found that

credit card industry profits were three to five times the ordinary rates of return in the banking

comments from investors in these companies and analyst reports suggest that credit card issuers do

is important to note that any costs of regulation were likely imposed on consumers right away while

the benefits may continue to accrue over time.

Consumers gain greater transparency right away. For example, the new regulatory framework restricts

increases in interest rates for existing cardholder balances because of penalty rates, arbitrary rate

changes, and other methods to obscure the cost of credit. These methods include manipulating how

These immediate changes are important because they equip consumers with more accurate information

when weighing whether to take on additional credit card debt.

While transparency gives consumers immediate benefit, prices are likely to improve over time for

two reasons. First, prices may improve over time because of greater competition spurred by more

transparent pricing. Second, historically consumer rates tended to rise over time well after introductory

rates expired because of price manipulations and arbitrary rate hikes. Issuers are now greatly

limited in their ability to arbitrarily change prices to a consumer's disadvantage. Therefore, while

any costs to consumers from the new rules have already been put into place, the largest share of

The bottom line is that reform appears to have improved transparency, made the card market more

competitive and saved consumers money.

To isolate the price effect of reform from those of other economic factors, a regression

price increase associated with credit card reforms. There is, however, evidence consistent with

improved transparency.

As shown in Table 2, both the stated rate and actual rate paid significantly increased when

unemployment increased. This makes sense because the unemployment rate is closely linked to

credit card losses. In addition, both the stated rate and the actual rate significantly increase when

the prime rate increased.

The most important results shown here are the last three variables. They represent the impact of key

events on interest rates. The first represents announcement of the FRB’s final credit card rules in

December 2008. Although the CARD Act, which was passed in May 2009, went further than the

FRB’s rules by adding and extending a number of rules, most of the key interest rate-related provisions

existed in some form in the rules. The implementation date used here is February 2010, when

most of the Act’s provisions went into effect. If these events were impacting the actual rate paid, we

would expect to see a large, positive and statistically significant coefficient value for these dates.

Instead, we see small coefficients that are not at all close to statistical significance, showing that

while the prime rate and unemployment affect rates, these events did not. None of the three dates

led to significantly higher actual rates, even though each event led to significantly higher stated rates

on offers. Even more importantly, perhaps, all three events saw a decrease in the difference between

Dependent Stated Rate Actual Rate Paid Differences in Rates

Regression 1 2 3 4 5 6 7 8 9

R-square 0.90 0.89 0.88 0.83 0.83 0.83 0.60 0.57 0.56

Unemployment 0.07 0.25 0.45 0.44 0.44 0.49 0.37 0.19 0.05

44.2% 98.7% 100.0% 100.0% 100.0% 100.0% 98.4% 86.9% 37.3%

Prime Rate 0.41 0.44 0.50 0.60 0.60 0.62 0.20 0.16 0.12

100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 99.1% 97.2% 90.7%

Quarter -0.06 -0.05 -0.05 -0.004 -0.004 -0.003 0.05 0.05 0.04

100.0% 100.0% 100.0% 66.5% 68.1% 57.4% 100.0% 100.0% 100.0%

Fed Rules 2.18 --- --- 0.29 --- --- -1.89 --- ---

Finalized 100.0% 50.5% 100.0%

CARD Act --- 1.76 --- --- 0.39 --- --- -1.38 ---

Passed 100.0% 68.3% 98.5%

CARD Act --- --- 1.30 --- --- 0.19 --- --- -1.11

Implemented 99.5% 38.6% 95.9%

First number is regression coefficient. Second number is confidence level (One minus p-value).

Dependent Difference in Difference in Actual Change in Difference between

Regression 1 2 3 4 5 6 7 8 9

R-square 0.38 0.35 0.37 0.59 0.60 0.59 0.57 0.56 0.55

Difference in 0.10 0.18 0.21 0.17 0.12 0.16 0.07 -0.06 -0.05

Unemployment 38.4% 65.3% 72.8% 57.5% 43.6% 55.7% 27.9% 24.4% 19.7%

Difference in 0.21 0.23 0.23 0.30 0.30 0.30 0.09 0.07 0.07

Prime Rate 90.2% 92.4% 93.3% 97.0% 97.1% 96.9% 52.8% 41.9% 37.9%

Lagged

Difference in 0.24 0.40 0.39 0.42 0.45 0.37 0.18 0.05 -0.02

Unemployment 74.0% 95.4% 97.9% 92.3% 96.6% 93.7% 57.3% 20.6% 8.3%

Lagged

Difference in 0.25 0.30 0.30 0.28 0.28 0.27 0.03 -0.02 -0.03

Prime Rate 94.1% 97.6% 97.9% 94.5% 95.7% 94.3% 18.2% 9.0% 18.2%

2nd Quarter -0.16 -0.17 -0.17 0.70 0.71 0.70 0.87 0.87 0.87

Dummy 86.6% 86.2% 86.8% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%

3rd Quarter -0.06 -0.09 -0.08 0.47 0.47 0.48 0.53 0.56 0.55

Dummy 40.0% 56.7% 49.1% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%

4th Quarter -0.24 -0.25 -0.23 0.11 0.11 0.12 0.35 0.36 0.35

Dummy 96.3% 96.7% 95.6% 64.3% 65.0% 66.4% 99.6% 99.6% 99.5%

Fed Rules 0.272 --- --- -0.059 --- --- -0.331 --- ---

Finalized 88.2% 24.6% 92.7%

CARD Act --- 0.037 --- --- -0.178 --- --- -0.215 ---

Passed 17.4% 67.2% 76.7%

CARD Act --- --- 0.244 --- --- 0.05 --- --- -0.193

Implemented 72.2% 16.5% 57.8%

First number is regression coefficient. Second number is confidence level (One minus p-value).

However, the regression models underlying Table 2 did show evidence of autocorrelation—meaning

reasons, this can sometimes affect the regression results. Therefore, a separate set of models that

examine the first change in the level of a variable (first differences) was also run. Results from these

models are shown in Table 3. The results confirm that there was no significant change in interest

rate paid, we would expect to see a large, positive and statistically significant coefficient value for

these dates. Instead, we see small coefficients that are negative in two of three cases and not at all

close to statistical significance, showing that while the prime rate and unemployment affect rates,

these events did not. However, the stated rate also shows no significance difference, and the change

in the difference in rates during the reform events is only of borderline significance in one regression.

This could be because this second set of models is more accurate, but it could also be that those

Across both types of regression models, the results consistently indicate that there was no real

change in interest rates associated with any of the three regulatory events. In addition, the results

are consistent with the improved transparency in credit card pricing associated with reforms. But

whether that change is statistically significant depends on the regression model used.

Table 4 shows the results of regressions with mail volume as the dependent variable and economic

factors as well as regulatory events as the explanatory variables. Not surprisingly, the results in Table

4 show that higher unemployment is linked to fewer credit card solicitations. The prime rate has no

apparent connection to mail volume. The CARD Act’s passage and implementation is significantly

linked to mail volume, but in the positive direction. The law’s passage may have reduced uncertainty

for issuers by clearly defining what the new ground rules will be, leading to more offers. Before the

law’s passage it is possible that mail volume was reduced in part from uncertainty in the regulatory

environment, in addition to great uncertainty about the future course of the economy. The reason

implementation is linked to more offers is unclear. Shortly before the implementation date, the

Federal Reserve cleared up some possible ambiguities in the CARD Act’s interpretation by issuing

a set of rules. Therefore the jump could once again be due to reduced uncertainty.

First number is regression coefficient. Second number is confidence level

(One minus p-value).

Dependent Variable Mail Volume (millions of pieces)

Regression 1 2 3

R-square 0.81 0.87 0.87

Prime Rate 23.20 -16.68 11.29

82.6% 72.8% 58.4%

Unemployment -63.96 -14.98 -93.98

97.4% 100.0% 100.0%

Fed Rules Finalized -27.52 --- ---

24.4%

CARD Act --- 271.9 ---

Passed 100.0%

CARD Act --- --- 198.70

Implemented 100.0%

Dependent Variable Difference in Mail Volume (millions of pieces)

Regression 1 2 3

R-square 0.08 0.08 0.06

Difference in 24.60 20.71 22.10

Prime Rate 56.6% 49.2% 51.4%

Difference in -47.95 -34.90 -37.89

Unemployment 84.4% 70.0% 73.1%

Fed Rules 20.00 --- ---

Finalized 86.0%

CARD Act --- 22.8 ---

Passed 87.2%

CARD Act --- --- 16.04

Implemented 54.4%

First number is regression coefficient. Second number is confidence level

(One minus p-value).

As with the regressions on FRB interest rates, to compensate for autocorrelation, the analysis

was repeated looking at the first differences in the rate gap rather than the level. As with the

level model, there is no evidence that the CARD Act or FRB rules caused a constriction in

Table 6 shows regression results analyzing the relative offer prices for consumer and business cards.

Bankruptcy filings were used as explanatory variables since they allow business and consumer risk

to be differentiated. The dependent variable is the gap between rates offered on all credit cards and

that offered on business credit cards, so a higher number indicates that rates for consumer cards rose

The prime rate was significantly negative in both regressions, suggesting that an increase in the

result of the commercial Chapter 7 bankruptcies analysis was negative and statistically significant in

one of the two regressions, suggesting that these commercial bankruptcies cause business offer rates

to rise relative to consumer offer rates, which is consistent with what one would expect. Consumer

bankruptcies were not statistically significant, perhaps because offer rates on consumer and business

the CARD Act’s passage nor implementation had a statistically significant impact on the gap

between business and consumer card rates. The sign on the coefficient for these events was not

even consistently positive, which is the outcome one would expect if the law caused consumer

credit to constrict.

First number is regression coefficient. Second number is confidence level (One minus p-value).

Because autocorrelation was also present in these regressions, the analysis was repeated looking at

the difference in rate gap rather than the level. As with the level model, there is no evidence that

Dependent Variable All Products Credit Card Offer Business Card Offer Rate Gap

Regression 1 2

R-square 0.81 0.80

Prime Rate -0.87 -0.89

100.0% 100.0%

Chapter 7 Non-Commercial 0.000 0.000

Petitions 0.3% 28.6%

Chapter 7 Commercial -0.001 -0.001

Petitions 96.3% 91.9%

Chapter 11 Commercial 0.000 0.000

Petitions 10.2% 8.6%

Chapter 13 Non-Commercial 0.000 0.000

Petitions 75.5% 69.2%

Chapter 13 Commercial 0.004 0.004

Petitions 86.0% 84.0%

CARD Act 0.60 ---

Passed 77.6%

CARD Act --- -0.20

Implemented 29.6%

Dependent Difference in Gap between:

Variable All Products Credit Card Offer Business Card Offer Rate Gap

Regression 1 2

R-square 0.13 0.13

Prime -0.82 0.13

Rate 45.8% 8.5%

Difference in Chapter 7 0.000 0.000

Non-Commercial Petitions 52.9% 26.6%

Difference in Chapter 7 0.000 -0.001

Commercial Petitions 11.1% 37.8%

Difference in Chapter 11 0.000 -0.001

Commercial Petitions 20.8% 43.7%

Difference in Chapter 13 0.000 0.000

Non-Commercial Petitions 27.4% 1.8%

Difference in Chapter 13 0.006 0.005

Commercial Petitions 86.5% 79.6%

CARD Act 0.52 ---

Passed 67.9%

CARD Act --- -0.49

Implemented 37.1%

First number is regression coefficient. Second number is confidence level (One minus p-value).

1 For example, Ann Hardie, “Banks find ways to charge credit card users,” Atlanta Journal Constitution, September 18, 2010; “The Politics of

Plastic,” The Wall Street Journal, August 24, 2010; “Credit card law is driving up interest rates,” The Detroit News, August 25, 2010, available

at http://detnews.com/article/20100825/OPINION01/8250316#ixzz10NMKaWWa.

2 This is discussed further in: Joshua M. Frank, “Why Free Markets Can Sometimes Turn into ‘Peacock Markets’: The Evolution of Credit

Cards,” Journal of Economic Issues, June 2010; and Joshua M. Frank, “What Does the Credit Card Market Have In Common with a Peacock?”

24-40 The Lydian Payments Journal, 2009.

3 Both are available in the G.19 statistical release at http://www.federalreserve.gov/releases/g19/.

4 Typically, news sources use credit card comparison sites that create their own index as the source of an average credit card rate (such as creditcards.

com). While the averages used by these websites is valuable when no other public data source exists (such as for comparing business to

consumer card rates), they do not give as accurate an estimate of the rates consumers currently pay as Federal Reserve data when it is available.

5 Bob Lawless, “The Fed’s Apparently Nonexistent Interest,” Credit Slips, September 1, 2010, available at http://www.creditslips.org/creditslips/

2010/09/the-feds-apparently-nonexistent-interest.html#more.

6 The data for the G.19 interest rates on credit cards are collected on the Quarterly Report of Credit Card Interest Rates (FR2835a).The

language on the instructions given to banks is somewhat ambiguous regarding exactly which purchase rate to use if the purchase rate is subject

to change. Bank interpretations could vary causing some to not enter penalty rates.

7 Even when penalty rates are included, a simple average will tend to underweight penalty rates since active, high balance accounts will more

often be assessed these penalties.

8 The period before Federal Reserve rules was the 3rd quarter of 2008. The period after implementation of the main provisions of the Credit

CARD Act was the 2nd quarter of 2010. Aggregate balances used for the finance charge calculation was from revolving credit in the Federal

Reserve’s G.19 table. Both finance charge calculations used the same balance level in that they are based on revolving credit from 2008 which

was $957.5 billion.

9 Data used was quarterly and ran from the 4th quarter of 1994 to the 2nd quarter of 2010 resulting in 63 data points.

10 Since issuers need at least a month or two to change rates and there was significant uncertainty about the final composition of both the

Credit CARD Act and the Federal Reserve Rules even a month before the event date, the quarter after passage of the Credit CARD Act and

finalization of the Federal Reserve Rules are used as the first period after the event. However, in the case of Credit CARD Act implementation,

there was much more certainty for most of the quarter in question, and credit card issuers were changing terms before the final implementation

date. Therefore the quarter of implementation (the first quarter of 2010) is included in the period after the event since this is when changes

would likely first be seen. Models with all three events combined into a single regression were also tested. These led to very similar results,

with all coefficients for the stated rates and the difference between stated and actual paid rates being in the same direction as shown here,

and the first two event dates being statistically significant for both dependent variables. None of the events had a significant impact on interest

rates paid.

11 The actual rate went up slightly but the change was statistically insignificant. Results for difference model, while not shown in this figure,

were also statistically insignificant.

12 Top 4 issuers are Chase, Bank of America, Citibank, and Capital One.

13 Interest and fee revenue comes from SNL-compiled data originating from FDIC call reports (specifically, interest and fees is from line

RIADB485 of Schedule RI). This revenue may be higher than Federal Reserve reported interest rates since it includes some fee revenue.

However, this data excludes securitized assets, and the reported non-securitized assets may be younger and therefore include more promotional

rates, bringing interest rates down. Certain fee revenue sources may change when economic conditions vary substantially even if there has not

been a change in pricing. For example, due to rising delinquencies, late fee revenue should be higher during the recession.

14 The conclusion that the decline in mail volume is due to economic conditions is based on the significance of unemployment in driving mail

volume found in the regression analysis and the increase seen in unemployment at the same time that mail volume declined.

15 Mintel Comperemedia, “Q3 2010 Credit Card Acquisition Response Rate Tracker,” November, 2010.

16 Based on monthly data from January 2005 to July 2010.

17 Shortly before the implementation date, the Federal Reserve cleared up some ambiguities in the Credit CARD Act’s interpretation by

issuing a set of rules. So the jump could once again be due to reduced uncertainty.

18 The problems in using this type of data to understand the current rate paid by cardholders have already been noted. However, if the

distortion of rates follows similar patterns over time in business and consumer cards, then a comparison between the two can still be valid.

19 The business card series excludes data between the fourth quarter of 2008 and the second quarter of 2009 due to a lack of business card

offers at that time. Based on discussions with creditcards.com staff, it was determined that the rate is distorted during that time period and likely

unreliable due to a small sample size. Business card offers dried up during this time period. If anything this would suggest that credit was more

constrained for business cards than consumer cards. And in fact, the offers that are available for this time period show a jump of roughly six

percentage points in business card rates in the period just after finalization of the Federal Reserve credit card rules.

20 One or two issuers voluntarily improved their repricing rules for business cards, but most issuers did not.

21 More specifically, this report found no reduced access to credit, no increase in prices paid, and no decrease in mail volume or the scope of

consumers mailed to. While the aggregate level of credit card debt has declined recently, this appears to be due both to reductions in debt from

elevated credit losses as well as some intentional deleveraging by consumers.

22 The use of these tactics in the credit card market is discussed in detail in: Joshua M. Frank, “Why Free Markets Can Sometimes Turn into

‘Peacock Markets’: The Evolution of Credit Cards,” Journal of Economic Issues, June 2010; and Joshua M. Frank, “What Does the Credit Card

Market Have In Common with a Peacock?” 24-40 The Lydian Payments Journal, 2009.

23 Lawrence M. Ausubel, “The Failure of Competition in the Credit Card Market,” The American Economic Review, 1991, Volume 81(1),

pg 50-81.

24 Board of Governors of the Federal Reserve System, “Report to the Congress on the Profitability of Credit Card Operations of Depository

Institutions.” Even now and despite record credit losses, credit card banks are doing relatively well. In the latest quarter (2nd quarter of 2010),

the after tax return on assets for all banks has been 0.65% while the return on credit card banks has been more than double that at 1.41% (from

Federal Deposit Insurance Corporation, “Quarterly Banking Profile,” Second Quarter, 2010). In 2009, credit card banks did slightly worse than

banks over all. However, in 2008 they did much better. The performance of credit card banks during the period since the financial crisis is especially

telling because one of the rationalizations that has been put forward for why credit card banks “deserve” a higher return is that these unsecured

assets are much higher risk. However, that fact that during the period in which they had the highest credit losses ever credit card banks

perform about as well or better than other banks shows that cyclical economic risk is not higher for credit card lending, refuting that argument.

25 Some estimates of the impact of the Credit CARD Act on issuers are reported (among other places) in Andrew Johnson, “Regulations,

Unemployment Pose Challenges for CARD Issuers,” American Banker, July 21, 2010; Kevin Dobbs, “Wells says regulatory changes to cost it

$530M,” SNL Financial, August 9, 2010; and Greg Gemson, “Credit Card Statement: Last Round of CARD Act Rules Finalized – Issuers Will

Look to Offset Revenue Lost Due to ‘Reasonable and Proportional’ Fee Limits,” Moody’s, June 29, 2010.

26 It is possible that some of the decreased revenue to issuers is due to reduced volume rather than lower prices, which would not result in

consumer savings. However, the evidence presented here does not show a reduction in solicitation volume. In addition, the Credit CARD Act

still allows issuers to charge any interest rate they want (as long as they do so up front), so with no general price cap, there is no theoretical

reason that issuers need to limit the supply of credit. It is possible that more transparent rates on offers will result in a lower response rate, but

this would be a positive outcome.

27 The cost of manipulations to payment allocation is described in “What's Draining Your Wallet? The Real Cost of Credit Card Cash

Advances.” Center for Responsible Lending, December 16, 2008b. Available at: http://www.responsiblelending.org/credit-cards/research-analysis/

what-s-draining-your-wallet-the-real-cost-of-credit-card-cash-advances.html. The costs of variable rate manipulations such as the use of floors

set at the initial rate and manipulation of the index rate are described in Joshua M. Frank, “Dodging Reform: As Some Credit Card Abuses Are

Outlawed, New Ones Proliferate,” Center for Responsible Lending, December 2009, available at: http://www.responsiblelending.org/credit-cards/

research-analysis/Dodging-Reform-As-Some-Credit-Card-Abuses-Are-Outlawed-New-Ones-Proliferate.html.

28 In addition, some of the most important fee reductions in the CARD Act are either not reflected in the Federal Reserve data which tracks

interest rates only, or not yet in effect at the time of the last data point on fee plus interest income for the top issuers (fee rules went into effect

in August 2010, while the data ends at the second quarter).

29 Data used was quarterly and ran from the 4th quarter of 1994 to the 2nd quarter of 2010 resulting in 63 data points.

30 Since issuers need at least a month or two to change rates and there was significant uncertainty about the final composition of both the

Credit CARD Act and the Federal Reserve Rules even a month before the event date, the quarter after passage of the Credit CARD Act and

finalization of the Federal Reserve Rules are used as the first period after the event. However, in the case of Credit CARD Act implementation,

there was much more certainty for most of the quarter in question, and credit card issuers were changing terms before the final implementation

date. Therefore the quarter of implementation (the first quarter of 2010) is included in the period after the event since this is when changes

would likely first be seen. Models with all three events combined into a single regression were also tested. These led to very similar results, with

all coefficients for the stated rates and the difference between stated and actual paid rates being in the same direction as shown here, and the first

two event dates being statistically significant for both dependent variables. None of the events had a significant impact on interest rates paid.

31 These regressions tended to have a Durbin-Watson statistic below 1, which indicates the dependent variable is correlated with prior

values over time. However, the first differences model described next, had Durbin Watson statistics close to 2 +/- 0.5 indicating no evidence

of autocorrelation.

32 This result was robust to a variety of other specifications. An AR1 model (an alternative method of eliminating autoregression) yielded

similar results. Changing the dummy variable between a one-time shift in level at the time of the event and a long term change in the difference

model dependent variable coefficient yielded similar results. Using 4-quarter difference instead of first differences yielded similar results, as did

including levels for prime and unemployment or additional lagged values.

33 A one-time event causing a shift in the interest rate level may lose its statistical power when converted to a dummy variable in a differences

model. It is for this reason that we show the results of both models—regardless of the method used, real interest rates were not affected by

regulation using three different event dates.

34 Though not statistically significant in the differences model, the coefficients if anything are consistent with credit expansion after regulation.

Once again this result was robust to different regression models. Monthly seasonal dummies, which are not shown here for the sake of simplicity,

did not change this result.

35 Data was monthly and ran from June 2007 to August 2010. Bankruptcy data came from The Bankruptcy Database Project at Harvard (http://

bdp.law.harvard.edu) in cooperation with Automated Access to Court Electronic Records (AACER).

36 The prime rate has been stable from before the Credit CARD Act’s passage until the most recent data available.

37 Business losses depend on the health of consumers. Many business cards go to sole proprietors running their own small business and who may

behave much like consumers. In addition, many of these cards are linked to the business owner’s personal credit and appear on their personal

credit bureau. Therefore, it is likely that business losses are closely tied to consumer bankruptcies.

38 Once again this result was robust to different regression models. Monthly seasonal dummies, which are not shown here for the sake of

simplicity, did not change this result.

The Center for Responsible Lending is a nonprofit, nonpartisan research and policy

organization dedicated to protecting homeownership and family wealth by working to

eliminate abusive financial practices. CRL is affiliated with Self-Help, one of the nation’s

largest community development financial institutions.

302 West Main Street

Durham, NC 27701

Ph (919) 313-8500

Fax (919) 313-8595

1330 Broadway

Suite 604

Oakland, CA 94612

Ph (510) 379-5500

Fax (510) 893-9300

910 17th Street NW

Suite 500

Washington, DC 20006

Ph (202) 349-1850

Fax (202) 289-9009

Copyright © 2011 by Center for Responsible Lending