Adoption of Revolving Credit Terms, the Disappearance of the Salience of Credit Costs
and Overconsumption*
Mary Zaki
University of Maryland
and
Jaclyn Evans
US Government Accountability Office
First Draft: May 2016
Current Draft: March 2017
Abstract
We construct a new dataset of consumer credit costs from plans offered in mail-order catalogs
and find that interest rates became sticky in the 1960's once credit started being quoted in
revolving (credit card-style) terms rather than with a separate cash and credit price (installment
terms). We argue that previous explanations of credit card rate stickiness are not as applicable in
the mail-order catalog setting and instead propose that quoting credit in revolving terms
decreases the salience of credit costs in the price of goods, which discourages competition on
credit costs. This would also imply that the main way that we quote the cost of open-ended credit
leads to overconsumption of goods. We test this implication in an experiment in which liquidity
constrained participants are provided the option to use differently quoted but otherwise
equivalent credit plans to purchase goods. We find that participants who receive credit in
revolving terms consume more goods, are less sensitive to credit cost changes and tend to
underestimate the payments required to pay off the credit plan (despite spending more time on
the calculation) than their installment-terms counterparts. These results are persistent even if
revolving terms are quoted as Annual Percentage Rates (APR) or are unshrouded from the fine
print, suggesting the complexity of the cost calculation is driving the results.
I. Introduction
As is well documented in the literature, the extreme stickiness of credit card rates relative
to those of other forms of credit became very apparent in the 1980’s when most interest rates
*
Corresponding Author: Mary Zaki, [email protected]. We are thankful for Andrew Sweeting, Pamela
Jakiela, Neslihan Uler, Emel Filiz Ozbay, Lint Barrage and Gregory Veramendi for helpful guidance,
Jeffrey Hunt and Shirley Pon for excellent research assistance and Andrew Marder for graciously
retrieving some hard to find documents. The opinions expressed in this article are the authors’ own and do
not necessarily reflect the views of the US Government Accountability Office. This paper first appeared
with the title "Historical Cost of Consumer Credit, Interest Rate Stickiness and Salience: Evidence from
Mail-Order Catalogs" as a chapter in Jaclyn Evans' dissertation.
were rapidly dropping.1 For example, from 1981 to 1991, the yield on the 3-month Treasury Bill
decreased by 9 percentage points. Over the same period, average finance rates on 48-month new
auto loans and 24-month personal loans issued by commercial banks decreased by 5 and 3
percentage points, respectively. On the other hand, average rates on credit cards issued by
commercial banks stayed relatively flat at around 18% APR.2 The insensitivity of credit card
issuers to decreasing cost of funds as well as evidence of supranormal profits is particularly
puzzling given low entry barriers to and concentration of the industry.3 In light of these stylized
facts, researchers have proposed several explanations for credit card rate stickiness including
search costs, switching costs and adverse selection.
In this paper we look back in time in order to investigate the origins of rate stickiness to
see if consumer credit costs were always sticky, even before the prominence of credit cards.
Since cost and term information of consumer credit was not systematically collected in the
United States until the 1970’s,4 we created a dataset of credit terms from credit plans found in the
major U.S. mail order catalogs of the 20th century. By doing this we are able to construct a new
time series of finance rates offered by major creditors of small dollar loans from the late 1920’s
to the early 1990’s.
We find that there were two styles of credit offered in mail order catalogs. Initially, mail
order catalogs, as well as other retailers, offered credit in the form of “installment” plans in
which a one-time “carrying charge” was added to the cash price of the good and households paid
off this summed amount (often referred to as the “time price”) in equal payments over several
months. Then, between the late-1950’s and the early 1960’s, retailers started to offer a new form
1
For good overviews of credit card rates in the 1980’s, read Ausubel (1991) and Calem (1992).
Rates retrieved from FRED Economic Data from the Federal Reserve Bank of St. Louis and Federal
Reserve Bank Annual Statistical Digest, accessed through FRASER, Federal Reserve Archive.
3
See Ausubel (1991) for evidence of credit card issuer profits.
4
“Rates on Consumer Instalment Loans”, September 1973 Federal Reserve Bulletin.
2
of credit: “revolving” credit plans. These plans are very similar in form to today’s credit cards.
Goods purchased with these plans did not incur credit costs if the outstanding balances were paid
off within a specified number of days. After the end of this “grace period,” purchasers had to
make minimum monthly payments and incurred monthly interest charges, which were calculated
as a percentage of the outstanding credit balances.
For our analysis, we calculate a finance rate on all credit plans offered in the catalogs
based on the stream of monthly payments required to purchase a good of a certain cash price. We
find that before 1959, rates on installment plans that covered all goods in the catalogs ranged
from 10% annual percentage rate (APR) to 60% APR depending on the size of the underlying
loan (e.g. rates were only above 18% APR for small purchases and generally decreased as the
purchase price increased). Rates moved around through time and there were several instances of
decreases in the installment plan finance rates even without a major decrease in interest rates in
the general economy. However, upon adoption of a revolving credit plan, each retailer charged
18% APR and held that rate for at least a decade. This 18% APR credit rate was much higher
than that charged under installment plans for medium to large purchases.5 For several retailers,
the switch from installment to revolving credit also coincided with a surge in sales volume and
specifically sales of goods sold on credit and not of goods sold with cash. Hence, there is
evidence that retailers were more willing or more able to extend credit at a higher cost when
credit was quoted in revolving terms rather than in installment terms.
We investigate possible reasons revolving credit rates seemed sticker than their
installment credit counterparts. We argue that the afore mentioned theories used to explain the
credit-card rate stickiness of the 1980’s are not as applicable in explaining revolving credit rate
5
Revolving credit was initially cheaper than installment credit for small purchases; however, with
additions of minimum fees and changes in outstanding balance calculations, pricing between the two
types of credit became comparable.
stickiness in mail-order catalogs. We also argue that the sticky revolving credit rate was not
simply a story of market power or binding usury ceilings. Instead we propose that revolving
credit, as compared to installment credit, is more effective in diminishing the salience of credit
costs in the price of goods. Possible sources for this decreased salience may come from the
shrouding of credit costs in the fine print of revolving credit plans as well as the increased
complexity of the calculation itself. Under this hypothesis, rate stickiness can be explained by
theoretic models of shrouding prices or profitably deceptive equilibriums, in which it is optimal
for firms to hide or diminish costs rather than compete on them. Such models offer explanations
not only for the observed revolving credit rate stickiness in the mail-order catalog setting, but
also for retailer preference of revolving credit to installment credit and for the decision of some
retailers to concurrently offer installment and revolving credit.
We test our salience hypothesis in an experiment, on the Amazon Mechanical Turk
platform, in which participants are provided an income stream and the option to use differently
quoted but otherwise equivalent credit plans to purchase goods. We find that participants who
receive credit in revolving terms consume more goods, are less sensitive to credit cost changes
and tend to underestimate the payments required to pay off the credit plan (despite spending
more time on the calculation) than their installment-terms counterparts. These results are
persistent even if revolving terms are unshrouded from the fine print or quoted using the more
familiar APR (in addition to a monthly percentage rate). These findings highlight that consumers
find the revolving terms credit cost calculation to be complex, even when one abstracts from the
array of payment streams borrowers can choose from to pay off their credit balance. They also
may explain why we did not see almost any change in the finance rates of revolving credit plans
in catalogs in 1968 when creditors were required to disclose APR after the passage of the Truthin-Lending Act.
The key implication of these results is that for certain levels of credit costs,6 quoting
credit in revolving terms (which is the current standard for most open-ended credit) leads to
overconsumption of goods. Put in another way, it is quite possible that there are rational timeconsistent consumers who are purchasing goods on revolving credit that would not make these
purchases if their credit was quoted in installment, or more salient, terms.
This paper contributes to economic history literature by creating a new dataset of
consumer credit terms and documenting consumer credit costs over previously undocumented
periods of the 20th century. Using this dataset, we are able to highlight credit terms changes over
significant historical events, such as the Great Depression, World War II and the enactment of
the Truth-in-Lending Act. We are also able to identify the conspicuous change in rate behavior
upon the adoption of revolving credit terms. The paper specifically contributes to the strand of
literature that addresses credit card rate stickiness.7 We find that rate stickiness was occurring
decades before the 1980's and that rates were not always as sticky prior to the adoption of
revolving credit. We add credit cost salience to search costs, switching costs and adverse
selection as the leading explanations for credit card rate stickiness and document that stickiness
can occur even if we abstract from these other explanations. Our paper also complements the
theoretical literature on shrouding of price8 by providing both experimental evidence of the
channel for shrouding as well as historical documentation of firm behavior surrounding the
6
This "level of credit cost" could be the present value of credit plan payments made over time or a
combination of revolving credit's equivalent "one-time carrying charge" and the minimum monthly
payment requirements. Determining this "level of credit cost" exactly is beyond the scope of this paper.
Instead, we show an example, in an experimental setting, of overconsumption occurring as a result of
method of quoting credit costs.
7
Ausubel (1991) and Calem and Mester (1995).
8
Gabaix and Laibson (2006) and Heidhues, Koszegi and Murooka (2017).
adoption and exploitation of this channel over several decades. Finally our paper contributes to
consumer credit literature and especially the subset that relates to the mistakes that consumers
make with credit.9 However, unlike most of this literature, we focus on purchasing and credit
origination mistakes rather than credit allocation or repayment mistakes.10 We also are able to
more directly measure limited attention than in previous work as we have both measures of
inattention (accuracy and time spent on calculations of credit payments) and shocks to salience
of credit costs (method of quoting credit costs).11
Our conclusions parallel those found in Chetty et al. (2009) who examine the effects of
salience of taxes on demand. However, unlike taxes, credit is an optional feature whose costs do
not necessarily need to be incurred if the consumer decides to purchase through savings rather
than credit. The availability of credit increases the consumer's choice set to include not only the
(quantity of) goods that one could purchase but also their interaction with purchasing method
(savings vs. credit). Our study enables us to examine how the change in salience of credit costs
affects both the demand for credit as a purchasing method as well as the demand for the
underlying goods that could be purchased with credit. This distinction turns out to be relevant in
our study. Though we find that credit demand is sensitive to interest rates (a result also found in
Ponce et al., 2017 and Gross and Souleles, 2002), we do not find the same sensitivity for goods
purchased for certain ranges of interest rates.12 Hence, credit demand and purchases do not
necessarily follow a one-to-one relationship.
9
Ponce et al. (2017), Stango and Zinman (2015; 2014; 2009b), Agarwal et al. (2013; 2009), Gross and
Souleles (2002).
10
Several papers address if consumers optimally accept credit card offers by examining their post-offer
debt accumulation behavior (Shui and Ausubel, 2005; Agarwal et al., 2015b).
11
Stango and Zinman (2014) shock the salience of overdraft fees by use of surveys. Alan et al. (2015)
shock the salience of overdraft fees by text messages.
12
We find a big drop in demand for credit when rates go from 0% APR to 18% APR, but not a
corresponding drop in goods purchased over the same range of interest rates. The caveat is that goods
Our results also share similarities with those found in Bertrand and Morse (2011) in the
payday loan setting. Payday loans, in contrast to credit cards, quote credit costs in both dollar fee
terms and APR terms for the length of the loan. However, borrowers can also roll over payday
loans at maturity, which essentially restarts the loan and requires a charge of additional fees. The
authors find that payday loan borrowers will reduce subsequent take up of payday loans if they
are presented with information on the accumulated fees on a $300 loan for a variety of rollover
scenarios along with the equivalent fees that would be accumulated if the loan was made on a
credit card. In our research, we find that consumers will reduce demand if credit costs are
presented in "accumulated fee" terms rather than APR terms, even when interest costs are not as
exorbitant as those associated with payday loans (which can easily have rates over 400% APR)
and even if there is no option for rollovers or uncertainty concerning payoff streams.
The Remainder of the paper is organized as follows. Section II provides background on
the history of retailer involvement in the consumer credit industry. Section III describes the new
data set that we constructed. Section IV presents the stylized facts found in the data. Section V
reviews previous explanations for similar stylized facts, assesses their applicability to the mail
order setting and proposes a new explanation. Section VI tests our explanation in an
experimental setting. And Section VII discusses the implication of our findings and concludes.
II. Background
Unlike consumers in the late 20th century who could make small purchases on credit by
use of bank credit cards, consumers in the early 20th century depended on retailers for credit.
Retailers used credit to facilitate sales and promote customer loyalty (Mandell, 1990). Among
retailers that strongly embraced the use of credit were general merchandise mail-order catalogs
purchased without credit are always received sometime in the future. Furthermore, our experimental setup
commits non-credit purchasers to saving money to purchase the good at a designated future date.
such as Sears, Montgomery Ward and Spiegel.13 Examples of specific items sold on credit can be
found in the catalogs from the early 1900’s. Credit plans that offered credit for a larger category
of goods were published in catalogs in the late 1920’s. Eventually, mail-order catalog retailers
offered credit plans for all goods in their catalogs in the 1930’s. Many retailers, including mailorder, also offered customers access to charge accounts which allowed customers a window of
time (e.g. 30 days) to pay for a good without incurring any penalties. Often customers were
given charge plates or cards that contained the identifying information of a customer’s charge
account.
In the “installment” credit market as a whole, which included automobile financing and
personal loans, commercial banks remained the biggest lenders. However, retailers grew as
creditors, contributing to about one fifth of the total installment credit extended in the 1970’s
(See Appendix Figure A1). Retailers also were the dominant issuers of charge and “revolving”
credit as seen in Appendix Figure A2. At the beginning of the 1970’s retail stores had more cards
outstanding and more credit balances owed than any other type of credit card. In 1973 54% of
the credit cards in use were retail cards, while bank cards made up only 11% of the total credit
cards in use (Mandell, 1990). By 1981, three of the top five credit cards held by households were
from major retailers with Sears at the top, as seen in Appendix Table A1. Specifically, 57% of
U.S. households held at least one Sears credit card. By 1988 12.6% of total consumer spending
for all goods and services was charged to credit cards in the U.S., and the top card issuer in the
U.S. was Sears, followed by 3 other retailers (Mandell, 1990). Given the major involvement of
retailers in the consumer credit arena for much of the 20th century, it would be insightful to study
the terms of the credit plans that they offered to their customers.
13
Several of the mail-order catalog retailers also had physical retail stores and offered credit in those
stores as well.
III. Data
In this paper we construct a unique panel data set of credit terms from general mail order
catalogs from the U.S. spanning decades of the 20th century. Data was collected from Sears,
Alden’s, Montgomery Ward, Spiegel, and J.C. Penney’s catalogs and spans Spring 1928 through
Spring 1994, though no one company covers that entire range.14 We only collected credit data
from catalogs when credit covers a category of goods versus being specific for each individual
product. It was not uncommon for several credit plans to be offered in the same catalog.
Installment credit plans contained information about down payment, one-time carrying charge
and monthly payments for a given range of balance values; see Panel A in Figure 1. Revolving
credit plans contained information about minimum monthly payments and information about
monthly charges, which was typically found in fine print; see Panel B in Figure 1. Information
from these disclosures allowed us to construct the stream of payments required to pay off a
certain credit purchase. Since revolving credit allows for prepayment, we base our calculations
on a payment stream in which no more than the minimum required monthly payments are made.
From these payment streams, we construct a time series of our main variable of interest:
the annualized internal rate of return (IRR). The internal rate of return is defined as the interest
rate for which the net present value of the cash flows is zero; in other words, it is the interest rate
such that the cost of purchasing a good in cash is equal to the expected value of the monthly
payments under credit. The equation used to calculate the IRR is:
!
0 = NPV = −(Cash Price − Downpayment) +
!!!
𝑀𝑜𝑛𝑡ℎ𝑙𝑦 𝐶𝑟𝑒𝑑𝑖𝑡 𝑃𝑎𝑦𝑚𝑒𝑛𝑡!
1 + 𝐼𝑅𝑅!
where we multiply the IRR by 12 to annualize it.
14
Catalogs were retrieved from various libraries and archives as well as purchased on EBay.
Along with IRR, we also record a time series of a measure of the minimum monthly
payments required by credit plans. Specifically, the first minimum monthly payment will be
equal to all monthly payments in the case of installment credit. This is not necessarily true with
revolving credit as monthly payments are weakly decreasing with the outstanding balance.
However, the first minimum monthly payment will necessarily be the largest monthly payment
that a borrower will have to fulfill for the duration of the loan.
Finally, we record a measure of “carrying charge,” which we will refer to as “interest
cost.” Interest cost is equal to the carrying charge in the case of installment credit. To calculate
the interest cost for revolving credit, we add up all monthly payments in the payment stream and
subtract out the cash price of the good. For our time series variable of interest, we divide the
interest cost by the borrowed amount (cash price minus down payment).
IV. Historical Cost of Consumer Credit
Terms on mail-order catalog credit plans varied with purchase size, typically with higher
finance rates associated with smaller purchases. To simplify the exposition in this section, we
focus on a purchase amount of $200 in nominal dollars.15 Stylized facts in the section are
generalizable for most purchase amounts except for small purchases.16 Figure 2 presents the
annualized internal rate of return for credit plans offered by mail order catalogs.17 Installment
plans are in shades of red and revolving plans are in shades of blue. Plans that only covered
appliances, furniture and other specified durables are symbolized by diamonds, while plans that
covered all goods in the catalog are symbolized by circles. We see that costs of credit, during the
15
All values in this paper are nominal as APR is nominal. Real values can be provided on request.
Historical terms of credit for other purchase amounts can be provided by authors on request.
17
We drop credit plans for home modernization loans that covered the same items applicable for Federal
Housing Authority (FHA) backed Title 1 Loans.
16
low interest rate period of the 1930’s and 1940’s, is comparable in magnitude to those found on
credit cards during the equivalently low-interest rate period of the writing of this paper. For
example, in 2015, the average interest rate on credit cards outstanding was 12.09% APR while
the average 3-month Treasury yield was 0.05%. In comparison, the rates on installment plans
offered by Sears and Montgomery Ward were 12.86% APR and 13.33% APR, respectively,
when the 3-month Treasury yield was 0.05% in 1939.18 Montgomery Ward and Sears start
publishing installment credit tables in 1928. Rates quickly drop after the introduction of these
tables. From 1928 to 1940, rates charged by Sears and Montgomery Ward dropped from 17.52%
to 12.86% and 24.13% to 13.33%, respectively. In contrast, over the same time period, the 3month Treasury bill yield dropped by only 1.75 percentage points from 1.79% to 0.04%. We see
in Figure 3 and Figure 4, that the decreased installment credit rates are a result of retailers’
competing on monthly payment amounts, carrying charges or both. A decrease in either of these
two variables holding the other constant will lower the IRR.
During WWII, the Federal Reserve enacted restrictions on consumer credit that included
the setting of down payment floors and the limiting of the length of installment loans. As a
result, rates were high and firms did not offer loans above $220. When restrictions were lifted at
the end of the War, credit on more expensive items is again offered in catalogs and interest rates
on credit experience another falling episode. From 1945 to 1954 rates on installment plans in
Sears, Montgomery Ward, Aldens and Spiegel catalogs drop from around 16.52% to 13.00%,
18.59% to 13.00%, 18.59% to 12.67% and 16.44% to 13.33%, respectively. In comparison, over
the same time period, the yield on the 3-month Treasury bill increased from 0.38% to 1.73%.
For comparison, note that credit provided by mail order catalog retailers is secured by the underlying
goods while that provided by general purchase credit cards is typically unsecured.
18
We see in Figures 3 and 4 that the rate drops are, again, driven by a mixture of monthly payment
and carrying charge decreases.
In the early 1950’s, both Sears and Montgomery Ward started offering several plans on
appliances and other durable goods with lower monthly payments. For the majority of the 1950’s
neither Aldens nor Spiegel offered equivalently low monthly payment plans. Sears and
Montgomery Ward charged higher carrying charges on these more liberal credit plans.
Eventually, carrying charges increased on all plans as interest rates rose for the economy as a
whole.
In 1959 and 1960 Spiegel and Aldens replaced their installment credit plans with
revolving credit plans. As can be seen in Figure 4, upon adopting revolving credit, Spiegel and
Aldens started lowering the minimum monthly payment amounts to levels comparable to those
of Sears and Montgomery Ward liberal installment plans. As can be seen in Figure 2, the IRR
charged by these retailers strikingly jumped from around 15% APR to 18% APR upon adoption
of revolving credit (without a corresponding jump in cost of funds). In Figure 3 we see that
Aldens and Spiegel, correspondingly increased the potential amount of equivalent “carrying
charge” under revolving plans. Sears and Montgomery Ward also added revolving credit plans to
their suite of installment plans in 1959 and 1960 though they initially had very high minimum
monthly payment requirements relative to their other credit plans. In Figure 5 we see, from data
gathered from Company Annual reports, that Spiegel and Aldens sales surged upon switching
from installment to revolving credit. Specifically credit sales spiked much more than cash sales
during the same period. We do not, however, see a similar spike in sales among Sears and
Montgomery Ward upon their adoption of revolving.19
J.C. Penney entered the mail-order business in the early1960’s and when it started
offering credit it only offered revolving-style plans. By the early 1970’s all mail-order catalogs,
with the exception of Sears, only offered revolving credit. Sears retired installment plans in the
late 1970’s.
One striking feature of revolving credit is the stickiness of the rates. Mail order catalog
retailers published a rate of 1.5% a month (i.e. 18% APR) on their revolving credit plans upon
their adoption in 1959 and 1960. Revolving credit rates stay at 18% APR or higher for the
remainder of time the catalogs existed or credit terms appear in catalogs.20 In the case of Sears,
this was a period of more than 30 years that included more than a decade of decreasing interest
rates in the economy. In the next section we will go more into depth about possible reasons for
this rate stickiness.
To summarize, four observations were presented in this section:
1) Rates on installment plans decreased faster than the yield on 3-month Treasury bills
during two episodes in our period of study, leading to a decrease in the spread
between the two rates.
19
Annual reports report total retail sales and not only catalog sales. According to Annual Reports, both
Sears and Montgomery Ward offered revolving credit in their retail stores before introducing it in their
catalogs.
20
Alden’s charged a rates of 12% APR when outstanding balances were above $350. In the early 1960’s
and once in 1974 Montgomery Ward charged 12% APR when outstanding balances were above $500.
From 1965 to 1973 J.C. Penney offered a revolving credit plan for durables that charged 14.4% APR for
most outstanding balance amounts (though for a period they charged 18% APR when outstanding
balances were lower than $90). From 1969 to 1973 J.C. Penney charged 12% APR on both revolving
credit plans when outstanding balances were above $500. In a few states in the 1960’s and more states in
the 1970’s, retail store credit fell under usury laws. Thus, in some states, rates on revolving credit plans
were under 18% APR depending on the level of the state’s usury ceiling.
2) In the late 1950’s and early 1960’s, revolving credit was adopted by existing mailorder retailers and by the end of the 1970’s none of our retailers offered installment
credit plans.
3) Upon adoption of revolving credit, retailers charged an IRR that was greater than the
IRR charged on installment plans for equivalent borrowed amounts.
4) Upon adopting revolving credit, firms that had not previously offered liberal terms
under installment plans lowered their minimum monthly payment requirements to
more competitive levels. They also had a surge in credit sales without a
corresponding surge in cash sales.
5) Rates on revolving credit were very sticky, unlike their installment credit
counterparts. Rates on revolving credit were, generally, 18% APR or above even over
periods when most other interest rates in the economy were falling.
V. Explanations of Rate Stickiness
A. Search Costs, Switching Costs and Adverse Selection
Ausubel (1991) and Calem and Mester (1995) offer several possible explanations for the
observed credit card rate stickiness of the 1980’s. In this section we review these explanations
and examine their applicability for the mail order catalog setting. We also suggest another
explanation based on the observations in the previous section.
Ausubel (1991) proposes three reasons for credit card rate stickiness: 1) search costs, 2)
switching costs and 3) consumer overoptimism concerning balance repayment behavior
combined with adverse selection. Search costs could arise if it takes time and effort to find banks
issuing credit cards at different rates. Search costs should be relatively minimal in the mail order
catalog setting as credit plan comparison would only require opening to the credit sections of the
catalogs which themselves are received in the mail.
Switching costs can come from several sources such as the time and effort expended in
filling out an application, the emotional toll of receiving a rejection, the annual fee associated
with some credit cards and the perception that one builds a stronger credit rating or receives
higher credit limits by holding the same credit card longer. In several of these aspects, mail
order catalog credit switching costs are minimal. Credit applications in catalogs are typically
very short (usually encompassing half a page) and many times are located on the other side of
the product order form. Furthermore, no annual fees are associated with this type of credit.
Unlike general-purpose credit cards, retailer credit was exclusive to purchase of products sold by
the retailer and it was not unusual for households to have access to multiple retailer credit
accounts.21 Hence, there would be less of a concern that purchasing a good on credit from
different retailers would lead to a damaged credit record. Also, though we do not have
information on this, it is possible that retailers allotted greater “credit limits” to customers with
longer (good) histories.22 However, again, there would be no impetus for customers to close one
retail credit account, and with it a good credit relationship/history, in order to open another retail
account. On the other hand, like general-purpose credit cards, there is still a chance of being
denied credit by mail order catalog retailers, though this switching cost is applicable to both
revolving and installment plans.
21
In 1980 83 million Americans held 290.5 retail credit cards. That is on average 3.5 retail credit cards
per retail credit card holder. (Source: U.S. Bureau of the Census, 1989)
22
Since credit terms such as down payments (which were eventually equal to 0 in most catalogs) were the
same for all borrowers, it is more likely that catalog companies offered greater “credit limits” by either
completely approving the printed credit offer or completely denying it. At certain times, catalogs offered
several credit plans with varying liberal terms. Thus it is possible that catalog companies could deny
credit for one plan but offer the customer credit with more conservative terms.
Lastly, Ausubel proposes the existence of three types of credit card users: credit card users
who never incur interest with their usage because they pay off their balances before the end of
the “Grace Period,” low-default credit card users who think that they are the first type of credit
card users but end up borrowing and incurring interest and high-default credit card users who
know they will be borrowers and search for the lowest cost credit. The low-default credit card
borrowers ignore credit card interest rate information because they do not think they will incur
interest. In this setup, if credit card issuers compete on interest rates, they will disproportionately
attract the less desirable high-default type borrowers over the more desirable low-default type
borrowers. Hence credit card issuers do not compete on interest rates due to this adverse
selection problem. We saw in the previous section that mail order catalog retailers lowered
minimum monthly payment requirements on their revolving plans while maintaining the interest
rate of 18% over significant periods of time. This raises some doubt about the mentioned adverse
selection explanation in the mail order catalog setting as lowering minimum monthly payment
requirements could reasonably disproportionately attract more liquidity-constrained individuals.
However, Ausubel's story that retailers would lose income on revolving credit customers who
underestimate their probability of becoming borrowers by competing on interest rates can exist
in the catalog setting. However, by not lowering interest rates, retailers may also be discouraging
certain customers from borrowing (or even purchasing goods) who would have borrowed under
the cheaper (though less liberal) terms of installment plans. These customers would realize that
even if they pay the same (higher) monthly payments as required by the older installment plans,
they would take longer to pay off their loans and they would incur a higher equivalent of a
“carrying charge.” Hence, mail-order companies may or may not find it optimal to compete on
interest rates depending on how the losses from decreased interest income from irrational
revolving credit users compare to the gains from higher sales to low-default risk borrowers.
Calem and Mester (1995) add to these explanations three possibilities that involve search or
switch costs leading to adverse selection. They first posit that consumers with high search costs
also tend to hold high balances. This could occur if those who are unwilling to devote time for
search because of their preference for leisure also are impatient and value current consumption
over future consumption. Hence if credit card issuers lower interest rates, they would
disproportionately attract individuals with lower search costs and who also will hold less
profitable low credit card balances. Again, most explanations with search costs do not seem to
be as applicable in the mail order catalog setting as comparison between catalogs can be done
quite costlessly.
Second they propose that creditworthy customers are offered favorable credit limits by their
current creditors based on private information. Hence, these customers would face higher
switching costs than their less creditworthy counterparts. Credit card issuers who lower rates
would more easily attract the less desirable customers. This can possibly be applicable to mail
order catalogs in the sense that a customer who has built up a good credit history with one
catalog retailer may be able to purchase more expensive items on credit from that retailer than
from another with whom the customer has had no history. In this way, the customer stays more
loyal to that catalog. On the other hand, as previously mentioned, because retailer credit is
exclusive to a specific retailer, households may already have credit accounts with various
retailers at the same time, allowing them to more easily build credit histories than with generalpurpose credit cards. Furthermore, there are no indications in the catalogs that once a customer
has an approved open account (whether installment or revolving) that there could be individual
restrictions on the maximum amount purchased unless it would lead to an outstanding balance
that was above the maximum amounts printed in the catalog credit terms themselves.
Finally, Calem and Mester propose that individuals with high credit card balances have more
difficulty switching to credit cards because credit card issuers cannot identify if individuals are
simply moving their current balances to the new credit card or are planning to accumulate even
higher debts, making them higher risk customers. Due to this asymmetric information problem,
credit card issuers reject potentially more profitable higher balance individuals and accept less
profitable lower balance borrowers. Hence, credit card issuers are less motivated on competing
for customers, but keep rates high for their captive clientele. It is unclear in our study period
whether a customer’s credit balance information with one retailer was accessible to other
retailers.23 In fact, with less accessible credit information, retailers would face larger asymmetric
information obstacles than credit card issuers in the 1980’s. However, as mentioned previously,
retailers’ behavior of setting down payments to 0 and lowering minimum monthly payments
requirements would indicate that they were not trying to dissuade liquidity constrained (and
potentially higher risk) individuals from applying for credit. Finally, all Calem and Mester’s
explanations of rate stickiness should be just as applicable to installment credit as to revolving
credit.
To summarize, search costs should have been minimal in the mail order catalog setting.
Some types of switching costs possibly existed and could account for some of the credit cost
stickiness in catalogs, however since these costs would exists for both revolving and installment
credit plans they could not explain the variation of rate stickiness between the two credit types.
Adverse selection explanations for rate stickiness seem to be at odds with retailers’ offering of
23
Credit reporting agencies did not get computerized until the 1970’s, which is when they started to
aggregate information nationally (Furletti, 2002).
more liberal terms through time. On the other hand, consumer time-inconsistency or over
optimism concerning the level of debt that will be held with revolving credit can explain mailorder catalog retailer hesitance of competing on credit costs.
B. Salience of Revolving Credit Costs
Along with the previous explanations, we propose that revolving credit rates are stickier
than those on closed-ended credit because revolving credit more successfully reduces the
salience of credit costs in the price of goods. Theorists conjecture that a shrouded prices
equilibrium, in which firms do not compete on the total price of a good, can be supported if there
exist consumers who themselves neglect to consider the full price of a good. Gabaix and Laibson
(2006) model a setting in which base goods come with add-ons (e.g. a printer and printer
cartridges or a bank account and overdraft services). If some customers do not consider the cost
of these add-ons when making their purchase decision on the base good then it may be optimal
for firms to not compete on the price of the add-on feature. This would occur if unshrouding
leads newly educated customers to become less profitable to the unshrouding firm by their
avoidiance of the add-on feature (and potentially the base good all together). This model fits
nicely in our setting in which credit is an add-on to the merchandise purchased.
In a related model, Heidhues, Koszegi and Murooka (2017) describe a setting in which a
good’s total price is made up of upfront prices that are understood by all and other prices that
could be ignored by naïve consumers unless unshrouded (e.g. credit cards with upfront annual
fees but more shrouded interest rate costs). Firms compete on the upfront price, which has a price
floor. If the price floor is binding then firms will only compete on the shrouded prices if the
good is not socially wasteful (i.e. production costs of the good is below the value people have for
the good). However, firms will not unshroud if the good is socially wasteful because consumers
will not purchase the good once they are able to see the total price. So, perversely, socially
wasteful goods are always sold in a profitably deceptive (shrouded) equilibrium. The model can
also support an equilibrium in which sophisticated and naïve consumers each separate and
purchase superior and relatively inferior goods, respectively, and the existence of the superior
good reinforces a profitably deceptive equilibrium of the relatively inferior goods rather than
unravels it. This model presents an explanation of why mail-order catalog retailers who were not
offering the most competitive installment terms preferred switching to the more shrouded credit
innovation of revolving credit than offering more transparent installment credit with higher
carrying charges but lower monthly payments. If revolving credit plans were indeed relatively
inferior to installment credit or if they were socially wasteful, then the implied necessary
existence of a profitably deceptive equilibrium could explain why more firms adopted and
replaced their installment credit plans with revolving credit plans. Furthermore, the model can
also explain why Sears continued to offer both the superior installment plan, which could attract
sophisticated customers, as well as revolving credit plans.
Two sources of “shrouding” seem plausible under revolving credit terms. First, shrouding
could occur because cost information is less visible in the catalog. Initially, when installment
credit was introduced in catalogs, a cash price and a credit price were printed side by side (along
with other credit terms) near the advertised product. Later, installment tables clearly featured the
carrying charges that were to be added to order balances. In some early catalogs, customers had
to add carrying charges to cash prices themselves when filling out the catalog order forms. In all
these cases, credit costs were explicitly integrated into the price of purchased goods. Eventually,
order forms stopped including an area for customers to do the previously mentioned calculation
and instead only included a box that customers could check to indicate their desire to purchase
on credit. And under revolving credit, costs of credit on the main catalog pages that explained
credit plans became more obscured. Revolving credit tables in catalogs only emphasized “low”
minimum monthly payment requirements for associated outstanding balance amounts and
advertising in catalogs focused on the buying power of these plans. Information on costs,
specifically the monthly percentage charge on outstanding balances, was now located in fine
print of catalogs. If credit costs were mentioned in the main text in the catalog, they would not be
quantified but rather referred to as “a low monthly charge.”
The second source of "shrouding" could be coming from the increased cognitive effort
consumers must exert in order to translate the monthly or annual percentage charge into the total
price of a good.24 Researchers have found instances when consumers are more responsive to all
inclusive price postings than posts where parts of the price are quoted in percent form. This
occurs even when costs will be paid with certainty within minutes of the purchase decision
(Chetty et al., 2009). In the mail-order catalog setting, revolving credit posits a much more
complicated computation with greater levels of uncertainty, varying minimum payment
constraints and many possible streams of payments over longer periods of time. In fact, Gabaix
and Laibson (2006) assume that, “consumers show a relatively muted response to complex,
contingent, camouflaged, distant, or disaggregated costs.”
C. Other Explanations
There are other possible explanations for the rate stickiness of revolving credit in the
mail-order catalog setting. Unlike the bank credit card market, the mail-order catalog market did
not include as many players and did potentially have high barriers to entry. Hence, price
competition might be muted by market power. However, we point that enough competitive
24
This is somewhat the reverse of "payment/interest bias" as highlighted in Stango and Zinman (2009a)
in which people have the tendency to underestimate the interest rate charged on credit when they calculate
it only from the loan amount and repayment stream.
pressure existed in this setting to encourage some competition with installment credit prices.
Furthermore, though we see some of the mail-order retailers exiting the market in the 1980’s, it
is exactly the time when bank credit cards emerge as strong alternatives to retail credit. Hence,
we believe a simple market power explanation does not completely explain the observed rate
stickiness.
Another possible source of rate stickiness is binding usury ceilings. This certainly could
have been the case in the late 1970’s and early 1980’s when interest rates in the economy were
very high. However, the stickiness of revolving credit rates started almost 20 years earlier when
this type of credit first appeared in the catalogs and when usury regulation was only beginning to
be applied to it. As noted, Curran (1967), most states did not consider retail credit to be under
usury regulation before 1957. This is because the courts viewed goods sold on credit as being
sold at a different price than cash price, which was not illegal. However, with the prevalence of
credit use, laws suites emerged claiming usury violations that lead to the enactment of stricter
regulations on retail credit in some states throughout the 1960’s. If these usury ceilings were in
fact binding, then we would expect that rationing of credit would occur. Though we do not have
information on credit application rejections or credit account closures for the mail-order catalog
retailers, we do have information on credit terms. We see in Figure 4, that in the 1960’s, when
retail credit regulation became more stringent and interest rates were not decreasing, revolving
credit terms became more liberal while the posted rate stayed fixed at 18% APR. Since more
liberal terms should attract more liquidity-constrained individuals, it seems unlikely that firms
were facing binding rate ceilings. Alternatively, firms could have used the newly applicable
ceilings to facilitate tacit collusion (Knittel and Stango, 2003).
VI. Experiment
If revolving credit reduces the salience of credit costs in the price of goods then that
would imply that household purchasing decisions are less sensitive to credit cost changes if
credit is presented in revolving terms rather than installment terms. Alternatively, it would imply
that households make more purchases if credit is quoted in revolving terms rather than
installment terms for credit costs above a certain level. We tested these implications in an
experimental setting in which participants are offered differently quoted but otherwise equivalent
credit plans to purchase goods.
We recruited 1,477 participants through the online labor market Amazon Mechanical
Turk (MTurk) over three days in February 2017.25 We limited our study to participants located in
the United States who spoke English, were over the age of 18 and had at least a 90% MTurk task
completion rate.26,27 Participants were told that they would receive $2 for completion of the
study (which would take 15-20 minutes) as well as a 1/30 chance of receive up to a $77.50 worth
of bonus payments and Amazon Gift Cards.
In a tutorial section, participants learned that they would be presented with an income
stream of $5 for 13 months starting in a month and a set of specially priced Amazon Gift Cards28
that they could purchase with this income stream (see Figure 6). The timing and payment
amounts of the income stream essentially force participants to be liquidity constrained. All
participants were told that they could purchase a desired gift card by forgoing income stream
25
Several studies have found that MTurk workers are more diverse than student pools but tend to be
younger, more educated and have lower income than the general population (Berinsky et al., 2012 and
Paolacci et al., 2010).
26
In two instances, recruitment postings also added the criteria that participants needed to have completed
at least 100 previous MTurk tasks. This does not seem to be a binding condition among the majority of
our recruited participants. Furthermore, dropping responses of all participants who were recruited at the
time that this extra criteria was posted does not change the results presented in this section.
27
We also prevented people from taking the study more than once (through IP address and Worker ID
checks) and disqualified people who restarted the experimental instrument multiple times.
28
Amazon Gift Cards cost 80% of their value.
payments until they have “saved” up enough money for its purchase. After forgoing enough
payments, participants would receive a claim code to redeem the Amazon Gift Card. Remaining
income stream payments would then resume as scheduled.
Some participants were also presented with an additional purchasing method, “credit,”
that would allow them to receive their desired gift card relatively immediately.29 Participants in
this credit subset were presented with either an installment plan or one of three revolving credit
plans. Figure 7 presents an example of each of these credit plans. The installment plan in Panel A
instructs participants to add a one-time service fee to the price of the gift card to calculate the
opening credit balance. The balance would then be paid off with monthly payments
automatically withdrawn from the income stream. Panels B, C and D present different revolving
plan versions. The revolving credit plan in Panel B is our “base” plan that only includes a
monthly percentage rate (MPR) while the credit plan in Panel C also includes the APR, as was
required after 1968 by the Truth-in-Lending Act. Finally the credit plan in Panel D emphasizes
the service fee by using bold font, red coloring and underlining to unshroud it from the fine print.
What allows comparison between these credit plans is that, other than having different
presentation styles, they are essentially equivalent, requiring the exact same payments to pay off
for the same amount borrowed. Finally, the credit costs of these plans varied between
participants. In APR terms, participants received plans that either charged 0%, 18% or 42%
APR.30 Appendix Table A2 provides a link between revolving terms and installment fees for the
varying gift card prices. For ease of exposition in this section, we will be referring to cost of all
credit plans in APR terms regardless of how the credit plan was actually quoted.
29
Participants were informed that Amazon Gift Cards would be provided on the next business day.
Regardless of type of credit plan, credit cost or cost of desired gift card, all plans list the same required
monthly payment of $5.
30
A key aspect of the experimental setup is that prepayment of the credit balance is not
possible (due to the constraint of the income stream), which is unlike what could happens in the
catalog and credit card setting in which grace periods and the ability to payoff a balance
completely at any point in time exist. By not allowing prepayment, we abstract from the adverse
selection story presented by Ausubel (1991) as participants who choose to purchase on credit are
not irrationally expecting to payoff the balance before incurring interest costs.
After completing the tutorial section and answering questions that indicate
comprehension, participants were asked to make their purchasing decisions (See Figure 6): 1)
which gift card they would like to purchase, if any, and for participants with a credit option 2)
whether they would like to purchase the gift card by saving or credit. To motivate participants to
respond honestly, they were told that they had a 1 in 30 chance of receiving the income stream
and having their purchasing decision occur.31 In addition to this purchasing decision, participants
who received a credit option were also asked to calculate the number of months it would take for
the credit balance to reach 0 if a $40 Gift Card was purchased using their previously assigned
version of a credit plan with a credit cost equivalent to 42% APR (See Figure 8).32 Participants
were told that one participant from the pool of those who answered this question correctly would
win $10 through a random draw.
Results33
31
When a participant completed all questions in the study, he was prompted to select a number between 1
and 30, inclusively. Then a random number generator would do the same. If the numbers match, then the
participant’s purchasing decision would occur.
32
Those who were quoted credit in installment terms in their purchasing question saw the service fees in
the last 5 rows of column E in Appendix Table A2 in their calculation question. Those who were quoted
the base revolving or unshrouded revolving terms saw a monthly service fee of 3.5%. And those that were
quoted revolving APR terms saw both a monthly service fee of 3.5% and an annual percentage rate of
42%.
33
Results in this section do not include participants who spent less than 30 seconds examining their credit
plan in the tutorial section. To not bias participants to pay attention to credit terms that they would not
We first focus on the percentage of participants that decide to purchase a gift card. In
Panel A in Table 1 we see how that outcome variable changes as participants are offered more
costly credit plans. We see in column (1) that 79% of participants who were offered 0%
installment credit plans purchased a gift card. That percentage does not change significantly
when the cost of credit on installment plans increases to 18% APR. However, the percentage of
participants purchasing gift cards significantly drops to 61% when the installment plan credit
costs are equivalent to 42% APR. In comparison, 61% of participants who were not offered
credit at all decided to purchase a gift card. Hence, offering an installment plan with 42% APR
had an equivalent result on percentage of people purchasing gift cards as not offering credit all.
Turning to the revolving credit case34 in column (2) we see a different story: there is no
significant change in the percentage of participants who purchase a gift cards as rates increase
from 0% to 42% APR. In Panel B we see that the percentage of participants who purchase a gift
card when offered a 42% APR revolving credit plan is 12 percentage points higher than when
they are offered an equivalently priced installment plan. Hence we see that just be quoting credit
costs in revolving, instead of installment, terms we could induce more participants to make
purchases. We see similar results in Table 2 when we separate out the different revolving credit
plans. Again, there are no significant differences in the percentage of participants who purchase a
gift card with an installment plan versus any of the revolving plans when credit costs are less
than or equal to 18% APR. However, each revolving plan causes a higher percentage of
participants to purchase a gift card than an installment plan when the credit cost is 42% APR,
even if the revolving plan is quoting the APR or is unshrouding the cost from the fine print.
naturally pay attention to, we did not ask households specifics about credit plan costs in the tutorial
questions. Hence it is possible for participants to put little effort in understanding the credit plan to
expedite taking the study. Conclusions are not changed if we include these participants.
34
We pool all participants who received any kind of revolving plan in these estimates.
If we make the added assumption that the same percentage of people who purchase a gift
card when credit is not offered continue to purchase a gift card when it is (a.k.a. “always
purchasers”), then we can back out the percentage of people who always purchase by saving, the
percentage of people who purchase on credit when credit is available but purchase by saving
otherwise and the percentage of people who decide to purchase only because credit is offered.
Figure 9 presents the breakdown of these three groups of purchasers. As mentioned in the
previous paragraph, with installment plans, the percentage of participants who purchase only
because credit is available (marked in blue) disappear when they are charged 42% APR. Under
revolving plans, a larger percentage of these types of purchasers are retained when credit costs
are increased from 0% to 42% APR. We see that there exist, under both revolving and
installment plans, a percentage of “always purchaser” who never use credit (marked in dark
pink) even if the cost of credit is 0% APR. These might be people who distrust credit or who find
value in receiving the gift card in the future rather than immediately. Finally, the percentage of
“always purchasers” who decide to use credit (marked in light pink) decreases in both cases of
revolving and installment plans when credit increases from 0% to 18% APR. The fact that some
“always purchasers” decide to use credit brings to light another function of credit: credit is a
commitment to purchasing. This function is less relevant in this experimental setting as people
who decide to purchase through savings are committed to that choice. However, outside of the
experimental setting, households who intend to purchase through saving may not eventually do
so due to being unable to save up the needed amount or because of change of tastes. Hence,
offering a credit plan gives retailers a tool to guarantee a purchase.
Another outcome variable that we can look at is the price of the gift card that is
purchased. In Panel A in Figure 10 we see that the percentage of purchasers who purchase the
most expensive ($50) gift card drops from 73% and 75% when APR is 0% to 64% and 72%
when APR is 42% for installment and revolving credit terms respectively. However, neither of
these drops is statistically significant at the 15% level. Hence, the majority of the drops sales
revenue,35 found in Panel B of Figure 10, as a result of increases in credit costs are coming from
the extensive and not the intensive margin.
Potentially the decreased responsiveness to credit cost changes of people who receive
revolving credit plans compared to those who receive installment plans is actually due to
installment plan’s tendency to cause people to overestimate costs than revolving plan’s tendency
to underestimate them. This could occur because the credit price calculated from the installment
plan will always be greater than the net present value of a good purchased on credit (i.e. cash
price plus the present value of monthly credit payments). We test to see which credit plan leads
participants to most accurately determine the payments required to pay off their credit balance.
As mentioned earlier, all credit option participants were asked to determine the length of time it
would take to pay off a credit balance for purchasing a $40 gift card purchased with a 42% APR
credit plan. With a required payment of $5 per month, the correct response is 10 months. The
distribution of responses based on the credit plan that participants received can be found in Panel
A of Figure 11. We see immediately a very striking pattern. The most popular response among
any revolving plan is 8 months. This is a response that would indicate a calculation in which
there is 0 cost to using credit. Over 50% of participants who received the base or APR revolving
credit plans calculated 8 months. This percentage decreases for those who received the
unshrouded revolving credit plan, however only 20% of respondents calculated the correct
35
Revenue here is not including revenue from credit interest.
response of 10 months.36 On the other hand, 60% of the respondents who received an installment
plan calculated the correct response. In Panel B of Figure 11 we present the average time spent
on the calculation question. We see that participants who received revolving credit plans spent
significantly more time on the calculation than those who received installment plans, despite
having higher instances of incorrect answers. Installment plan participants spent an average of 65
seconds on the calculation question while revolving credit plan participants averaged between 97
and 145 seconds. These results indicate that participants are less sensitive to credit cost changes
when credit is quoted in revolving terms than when it is quoted in installment terms due to
underestimation of credit costs under revolving terms rather than overestimation of credit costs
under installment terms. Quoting revolving rates in the more familiar APR terms did not increase
accuracy of identifying the payments required to pay off a credit plan. Unshrouding the
revolving credit terms from the fine print did increase awareness of credit costs but did not
drastically increase accuracy of the responses. Putting all these results together points to
evidence that a lot more effort is required to calculate credit costs when credit is quoted in
revolving terms than when it is quoted in installment terms.
VII. Discussion and Conclusion
The experimental findings highlight the increased complexity of calculating credit costs
when they are quoted in revolving terms, even if there is no uncertainty in the timing of
prepayment. They also show us the consequences of decreased salience of credit costs through
revolving credit: overconsumption. Because credit cards and most other open-ended credit are
quoted in revolving terms, they are vulnerable to causing consumer to make purchases that they
otherwise would not have if terms were quoted in a more salient way. Thus, this
36
Soll et al. (2013) ask a similar calculation question using credit quoted in revolving terms, to adults
recruited from a national panel, and also find that participants underestimate the time it takes to pay off a
credit balance.
overconsumption due to credit can occur even when people are not present-biased or particularly
impatient.
Regulators who want to minimize these impacts need to increase the salience of credit
card costs. However, this task is not easy. For example, the Truth in Lending Act of 1968 (TILA)
mandated the disclosure of APR on all consumer credit. However, we see in Figure 2, that there
were no large changes to the IRR of revolving credit plans in catalogs when TILA went into
effect.37 This can partially be explained by the salience findings in our experiment.
The 2009 Credit Card Accountability Responsibility and Disclosure Act (CARD)
mandated the disclosure of interest rate costs for a credit card holder’s outstanding balance if
only the minimum payments are paid for the duration of the loan as well as the interest rate costs
if current balances are paid off in 36 months. This type of disclosure is addressing revolving
credit cost salience issues more directly than those in the TILA. Consumers can more clearly see
how interest rate costs add up in two examples on their statement.38 However, these disclosures
are presented after the purchase decisions have already been made. A potentially more effective
policy would mandate that credit card solicitation materials or statements include disclosures of
aggregated interest costs for a selection of starting balance amounts if only minimum payments
were made. In general, to abate overconsumption, the most effective policy measures would need
to increase salience of the cost of revolving credit at the point of purchase or earlier.
37
In fact, for smaller balances we see an increase in revolving credit costs at the time of TILA due to
adoption of minimum interest fees and changes in the way monthly outstanding balances are calculated.
38
Agarwal et al. (2015a) and Keys and Wang (2015) find that 36-month payment amount creates an
anchor effect. Keys and Wang present evidence that some consumers who were paying their full balance
end up paying the 36-month payment. Furthermore, this disclosure itself might be shrouded for credit
card customers who pay their bills and view their credit card activity online. These customers would
need to open an electronic version of their paper statement in order to view the disclosure as it is not
required to be displayed elsewhere.
References
Agarwal, Sumit, Souphala Chomsisengphet, Neale Mahoney, and Johannes Stroebel. 2015a.
“Regulating Consumer Financial Products: Evidence from Credit Cards.” Quarterly Journal of
Economics, 130.
Agarwal, Sumit, Souphala Chomsisengphet, Chunlin Liu, and N. Souleles, 2015b. “Do
Consumers Choose the Right Credit Card Contracts?” Review of Corporate Finance Studies.
Agarwal, Sumit, John Driscoll, Xavier Gabaix, and David Laibson. 2013. “Learning in the Credit
Card Market, ” American Economic Review, 5 (3), 681-703.
Agarwal, Sumit, Paige Skiba, and Jeremy Tobacman. 2009. “Payday Loans and Credit Cards:
New Liquidity and Credit Scoring Puzzles?” American Economic Review Papers and
Proceedings, 99 (2): 412–17.
Alan, Sule, Mehmet Cemalcılar, Karlan, Dean, and Jonathan Zinman. 2015. “Unshrouding
Effects on Demand for a Costly Add-on: Evidence from Bank Overdrafts in Turkey.”
Ausubel, Lawrence M. 1991. “The Failure of Competition in the Credit Card Market," American
Economic Review, 81 (1), 50-81.
Keys, Benjamin, and Jialan Wang. 2015. “Minimum Payments and Debt Paydown in Consumer
Credit Cards.”
Berinsky, A. J., Huber, G. A., & Lenz, G. S. 2012. Evaluating Online Labor Markets for
Experimental Research: Amazon.com's Mechanical Turk. Political Analysis, 20(3), 351-368,
doi:10.1093/pan/mpr057
Calem, Paul S. 1992. "The Strange Behavior of the Credit Card Market." Business Review,
Federal Reserve Bank of Philadelphia, January/February 1992, pp. 3-14.
Calem, Paul S., and Loretta J. Mester. 1995. “Consumer Behavior and the Stickiness of CreditCard Interest Rates.” American Economic Review, 85(5): 1327–1336.
Chetty, Raj, Adam Looney, and Kory Kroft. 2009. “Salience and Taxation: Theory and
Evidence.” American Economic Review, 99(4), 1145–1177
Curran, Barbara A. 1967. “Legislative Controls as a Response to Consumer-Credit Problems.”
Boston College Law Review, 8(3): 409
Gabaix, Xavier and David Laibson. 2006. “Shrouded Attributes, Consumer Myopia, and
Information Suppression in Competitive Markets,” Quarterly Journal of Economics, 121 (2),
505-540.
Gross, David B. and Nicholas S. Souleles. 2002. “Do Liquidity Constraints and Interest Rates
Matter for Consumer Behavior? Evidence from Credit Card Data,” Quarterly Journal of
Economics, 117 (1), 149-185.
Heidhues, Paul, Botond Koszegi, and Takeshi Murooka. 2017. “Inferior Products and Profitable
Deception.” Review of Economic Studies
Keys, Benjamin J. and Jialan Wang. 2014. “Perverse Nudges: Minimum Payments and Debt
Paydown in Consumer Credit Cards.”
Knittel, Christopher R. and Victor Stango. 2003. “Price Ceilings as Focal Points for Tacit
Collusion: Evidence from Credit Cards,” American Economic Review, 93 (5), 1703–1729.
Mandell, Lewis. 1990. “The Credit Card Industry: A History”. Boston: Twayne Publishers
Paolacci, G., Chandler, J., & Ipeirotis, P. G. 2010. “Running Experiments on Amazon
Mechanical Turk,” Judgment and Decision Making, 5(5), 411-419.
Ponce, Alejandro, Enrique Seira, and Guillermo Zamarripa. 2017. “Borrowing on the Wrong
Credit Card: Evidence from Mexico.” American Economic Review
Shui, Haiyan and Lawrence M. Ausubel. 2005. “Time Inconsistency in the Credit Card Market.”
Soll, Jack B, Ralph L Keeney, and Richard P Larrick. 2013. “Consumer Misunderstanding of
Credit Card Use, Payments, and Debt: Causes and Solutions.” Journal of Public Policy &
Marketing, 32 (1): 66–81.
Stango, Victor, and Jonathan Zinman. 2009a. “Exponential Growth Bias and Household
Finance,” Journal of Finance, 64: 2807-49.
Stango, Victor, and Jonathan Zinman. 2009b. “What Do Consumers Really Pay on Their
Checking and Credit Card Accounts? Explicit, Implicit, and Avoidable Costs.” American
Economic Review Papers and Proceedings, 99 (2): 424–29.
Stango, Victor, and Jonathan Zinman. 2014. “Limited and Varying Consumer Attention:
Evidence from Shocks to the Salience of Bank Overdraft Fees,” Review of Financial Studies, 27
(4): 990–1030.
Stango, Victor, and Jonathan Zinman. 2015. “Pecuniary Costs of Bounded Rationality: Evidence
from Credit Card Debt Allocation.”
U.S. Bureau of the Census. 1989. “Statistical Abstract of the United States: 1989.” Washington,
DC
Figure 1
Examples of Credit Terms in Mail-Order Catalogs
Panel A: Installment Credit Terms
Source: Montgomery Ward Fall 1954 Catalog
Panel B: Revolving Credit Terms
Source: Aldens 1963 Spring Sale Catalog
Figure 2
Internal Rate of Return for a $200 Purchase
Panel A: Spiegel
Panel B: Aldens
Panel C: Sears
Panel D: Montgomery Ward
Panel E: J.C. Penney
Legend
Notes: Solid black line represents the rate on the 3-month Treasury bill.
Figure 3
Interest as a Percentage of Loan Amount for a $200 Purchase
Panel A: Spiegel
Panel B: Aldens
Panel C: Sears
Panel D: Montgomery Ward
Panel E: J.C. Penney
Legend
Figure 4
First Minimum Monthly Payment for a $200 Purchase
Panel A: Spiegel
Panel B: Aldens
Panel C: Sears
Panel D: Montgomery Ward
Panel E: J.C. Penney
Legend
Figure 5
Sales Reported in Annual Reports
400
0
100
100
Millions of Dollars
200
300
Millions of Dollars
200
300
400
Panel A: Spiegel
1950
1955
Net Sales
1960
1965
1970
1950
1955
Net Sales Including Service Charges
1960
Credit Sales
1965
1970
Cash Sales
50
20
40
Millions of Dollars
100
150
Millions of Dollars
60
80
100
200
Panel B: Aldens
1945
1945
1950
1955
1960
1950
1965
1955
Installment Sales
Cash Sales
Net Sales
1960
1965
Credit Sales
10000
0
2000
5000
Millions of Dollars
10000
15000
Millions of Dollars
4000
6000
8000
20000
Panel C: Sears
1955
1960
1965
Net Sales
1970
1975
1955
1960
1965
Credit Sales
1970
1975
Cash Sales
Note: Dashed Vertical line Drawn in year immediately before adoption of revolving credit in retailer
catalog.
Figure 5 (con’t)
1500
0
1000
Millions of Dollars
500
1000
Millions of Dollars
1500
2000
2500
3000
Panel D: Montgomery Ward
1950
1955
1960
Net Sales
1965
1970
1950
1955
1960
Credit Sales
1965
1970
Cash Sales
Note: Dashed Vertical line Drawn in year immediately before adoption of revolving credit in retailer
catalog.
Figure 6
Example of Question Page in Experiment
Figure 7
18% APR Credit Plans
Panel A: Installment
Panel B: Revolving - Base
Panel C: Revolving - APR
Panel D: Revolving - Unshrouded
Figure 8
Calculation Question (Installment Credit Version)
Figure 9
Percent of People Making Purchases
Figure 10
Intensive Margin and Average Revenue
Panel B: Average Price of Gift Card Purchased
Per Person (Including Non-purchasers)
0.55
24
26
28
Dollars
30
32
Portion of Purchasers
0.60
0.65
0.70
34
0.75
Panel A: Percent of Purchasers who Purchase
the Most Expensive ($50) Gift Card
0%
18%
APR
Installment
Revolving
42%
No Credit
0%
18%
APR
Installment
Revolving
42%
No Credit
Figure 11
Calculation Question Results
Panel A: Calculation Responses
Note: The correct response is 10 months, which is designated by the red bar. 8 months, designated by the
black bar, would indicate a calculation in which there is 0 cost to using credit.
60
80
100
Seconds
120
140
160
Panel B: Mean Time Spent on Question
Revolving:
Base
Revolving:
APR
Revolving:
Unshrouded
Note: Bands represent 95% confidence intervals around estimate of mean.
Installment
Table 1
Dependent Variable: % of Respondents Who Make a Purchase
Panel A
(1)
Installment
0.01
(0.06)
(2)
Revolving: All
-0.02
(0.04)
42% APR
-0.18***
(0.06)
-0.04
(0.04)
Constant
0.79***
(0.05)
302
0.77***
(0.03)
886
18% APR
N
Note: Omitted category is 0% APR.
Revolving: All
Constant
N
(1)
APR: 0%
-0.02
(0.05)
0.79***
(0.04)
388
Panel B
(2)
APR: 18%
-0.05
(0.05)
0.79***
(0.04)
404
Note: Omitted category is Installment.
Standard errors are in parenthesis. + p<0.15, * p<0.1, ** p<0.05, *** p<0.01
(3)
APR: 42%
0.12**
(0.05)
0.60***
(0.04)
396
Table 2
Dependent Variable: % of Respondents Who Make a Purchase
(1)
(2)
(3)
APR: 0%
APR: 18%
APR: 42%
Revolving: APR
-0.04
-0.02
0.16**
(0.06)
(0.06)
(0.06)
Revolving: Unshrouded
0.00
(0.06)
-0.08
(0.06)
0.11*
(0.06)
Revolving: base
-0.01
(0.06)
-0.05
(0.06)
0.10+
(0.06)
Constant
0.79***
(0.04)
388
0.79***
(0.04)
404
0.60***
(0.04)
396
N
Note: Omitted category is Installment. Standard errors are in parenthesis.
+ p<0.15, * p<0.1, ** p<0.05, *** p<0.01
Appendix
Figure A1
Installment Credit Extended (millions of dollars)
80000
70000
60000
50000
40000
30000
20000
10000
0
1955 1956 1957 1958 1959 1960 1961 1967 1968 1969 1970 1971 1972 1973 1974
CommericalBanks
FinanceCompanies
RetailOutlets
OtherFinancialLenders
Source: Federal Reserve Bulletins
Figure A2
Charge Account Credit (millions of dollars)
12000
10000
8000
6000
4000
RetailOutlets
Source: Federal Reserve Bulletins
Non-retailcreditcards
1974
1973
1972
1971
1969
1970
1968
1967
1966
1965
1961
1960
1959
1958
1957
1956
1955
1954
1952
1953
1951
1950
1949
1948
1941
1945
0
1939
2000
Table A1
U.S. Households Holding At Least One of the Specified Credit Cards in 1981
Sears
57%
Visa
53%
MasterCard
47%
J.C. Penney
39%
Montgomery Ward
27%
Federated Dept. Stores
17%
American Express
11%
Source: Mandell (1990)
Table A2
Service Fees
Price of Gift
Card
(A)
Value of Gift
Card
(B)
$10
$20
$30
$40
$50
$12.50
$25.00
$37.00
$50.00
$62.50
Installment Service Fee
(C)
0% MPR/
0% APR
$0
$0
$0
$0
$0
(D)
1.5% MPR/
18% APR
$0.23
$0.78
$1.68
$2.94
$4.58
(E)
3.5% MPR/
42% APR
$0.56
$1.94
$4.27
$7.77
$12.63