Plastic Choices: Consumer Usage of Bank Cards vs. Proprietary Credit Cards Running Title: Plastic Choices Kenneth A. Carow, Ph.D. Assistant Professor Indiana University Kelley School of Business 801 West Michigan Street Indianapolis, IN 46202-5151 Michael E. Staten, Ph.D. Director, Credit Research Center Georgetown School of Business 1300 35 th Street NW Washington DC, 20007 April, 2000
30
Embed
Plastic Choices: Consumer Usage of Bank Cards vs. Proprietary ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Plastic Choices: Consumer Usage of Bank Cards vs. Proprietary Credit Cards
Running Title: Plastic Choices
Kenneth A. Carow, Ph.D. Assistant Professor Indiana University
Kelley School of Business 801 West Michigan Street
Indianapolis, IN 46202-5151
Michael E. Staten, Ph.D. Director, Credit Research Center Georgetown School of Business
1300 35th Street NW Washington DC, 20007
April, 2000
Plastic Choices: Consumer Usage of Bank Cards vs. Proprietary Credit Cards
Abstract Using survey data from retail and gasoline cardholders, we examine the substitution of general
purpose (bank) cards for proprietary cards and how issuers can predict which consumers are
most likely to substitute. Convenience and rebates are the primary reasons for using a bank
card. However, consumers use their proprietary gasoline cards to keep purchase records and
their proprietary retail cards to obtain better service. These results help explain the growth in
popularity of “co-branded” cards.
Keywords:
Credit card Consumer finance Credit
Plastic Choices: Consumer Usage of Bank Cards vs. Proprietary Credit Cards
Over the past decade, the consumer’s choice of how to pay for goods and services at the point of
sale has taken on great commercial significance.1 The cashless/checkless society has not yet arrived, but
the use of non-cash payment methods (credit cards, debit cards, stored-value cards) is growing rapidly.
Overall, cash and checks still account for about 75 percent of all retail transactions in the U.S.; however, the
number of transactions settled by plastic payment devices is sufficiently large to attract many new entrants
competing for market share. Entry has spurred dramatic developments in the features of plastic cards,
providing customers a far wider array of options than existed even as recently as 1990.
The growing popularity of general purpose (bank) credit cards and, more recently, debit cards is
causing many retailers to re-evaluate the purpose and relevance of their proprietary credit card programs (a
proprietary “store” or “gasoline” card is typically accepted only by the issuing merchant and its affiliates).
With nearly 400 million proprietary cards in circulation, retailers obviously have a significant interest in
customer preferences regarding how to pay at the point of sale. Will proprietary cards survive as a payment
alternative in the customer’s wallet, or has their time passed?
To date, no empirical work has analyzed why consumers choose to use one type of credit
card over another (proprietary versus bank card usage). Duca and Whitesell (1995) look at card
ownership. Martell and Fitts (1981) and Canner and Cyrnak (1985) compare the demographic
characteristics of credit card users to non-users. Neither study; however, controls for the
opportunity to use credit cards (i.e., card ownership). Controlling for card ownership, Lindley,
Rudolph, and Selby (1989) study card usage for several types of purchases, but do not differentiate
between the use of proprietary and bank cards.
This paper presents survey evidence on the reasons consumers choose to use a bank card vs. a
proprietary credit card. The results have important implications for both proprietary and bank card issuers
and explain the growing popularity of “co-branded” cards issued jointly by banks and merchants.
2
I. The Battle for Wallet Share
Charge volume statistics demonstrate the growing appeal of bank cards to consumers. U.S. charge
volume on credit totaled $1.1 trillion during 1999, accounting for over 20 percent of purchases. With nearly
800 million cards, revenues in excess of $80 billion, and a 2.2 percent average return on assets for the credit
card industry, competition is intense.1 With a total of over 400 million bank cards, 300 million “store” cards,
and 100 million gasoline cards, the array of payment options offered to customers has become an
increasingly important dimension of retailing.
There are three primary reasons associated with the increased importance of the payment
system. First, customers wish to minimize the transaction cost of paying for their purchases. For
example, oil companies have boosted gasoline sales per station by installing pump -based card
readers to speed up transactions, which reduces the transaction cost in terms of the customer’s
time. Second, payment devices can and are being used to build brand loyalty (repeat purchases).
Bank card issuers have found rebates are particularly effective at encouraging repeated card usage
across a wide variety of retail outlets. The power of rebate programs has not been lost on retailers:
many have agreed to jointly issue a “co-branded” Visa or MasterCard. Third, technological
advances have greatly increased the value of the information captured about customer preferences
when credit cards are used instead of cash or checks. Purchase patterns are key ingredients in
target marketing efforts to generate incremental sales. The card issuer captures the information,
elevating the importance of which card is used when a customer chooses to pay with plastic.
Merchants with proprietary card programs lose such valuable information if their cardholders
increasingly switch to bank cards.
With the proliferation of bank credit cards in wallets, and the increased acceptance of such cards
by merchants (including non-traditional outlets such as fast food restaurants and grocery stores), the
merchant’s proprietary plastic no longer provides a unique credit service. Fewer customers need the store
1 “Profits continue to grow in 1999, paced by fees,” Credit Card News, April 15, 2000, p. 1
3
card solely for its credit privileges, because they carry other credit cards that are also accepted. As the
number of retail outlets accepting bank cards grows, the customer’s reason for carrying a portfolio of cards
(bank and proprietary) diminishes. Consolidation of charge activity into a smaller number of cards in the
wallet is feasible. Consequently, the reasons that customers prefer one card over another interest both
proprietary and bank card issuers.
This paper utilizes two unique consumer surveys (one targeted at retail store cardholders, the other
at gasoline cardholders) to address the following question: For a sample of customers who own both a
proprietary card and a bank card, how often do they substitute the bank card in place of the proprietary
card, and why? The following sections identify and discuss:
a. Reasons for choosing to use the bank card instead of the proprietary card. b. Reasons for continued use of store/gasoline cards instead of bank cards. c. Characteristics of those customers who substitute more frequently vs. those who
continue to use their proprietary cards. d. Comparison of answers to these questions in two different retail contexts: gasoline
stations vs. other retail stores. e. Implications of the empirical answers to these questions for proprietary card issuers,
especially those considering co-branding with a bank card issuer.
II. Prior Research on Choice of Payment Method
Hirschman (1982) documents that consumers perceive different attributes, advantages and
disadvantages across five specific types of payment systems: cash, personal checks, bank credit cards,
proprietary store credit cards, and travel and entertainment cards (T&E cards, such as the American Express
green card). Through the use of focus groups, Hirschman finds that consumers identify the following
eleven distinct attributes of payment systems: 1) budgeting, 2) control spending, 3) documentation, 4)
social desirability/prestige, and 11) transfer time.
Among the credit card alternatives, consumers view bank cards as having greater leverage
potential and acceptability relative to retail store cards. For each of the nine remaining attributes, bank and
store cards are nearly identical. Assuming that consumers attempt to minimize the transaction costs of
4
supporting their expenditure stream, we can conclude that people who carry and use proprietary cards
either: 1) don’t have a bank card, 2) are credit constrained, (i.e., at or near the credit limits on their bank
cards), or 3) derive additional value from using the proprietary card for reasons not identified in the
Hirschman analysis. We address these possibilities in Section IV, which analyzes store card vs. bank card
usage at retail outlets.
While it is not possible to observe directly the consumer utility function, empirical analysis in this
area is related to a framework in which households maximize the utility of consumption, subject to the cost
of using a medium of exchange in any transaction. Duca and Whitesell (1995) find that households with
heads who are male or who live in rural areas are less likely to own a credit card, whereas households with
heads who are married, white, high school graduates, or college graduates are more likely to own a credit
card. Martell and Fitts (1981), Canner and Cyrnak (1985), and Boeschoten (1992) find that households with
greater education and higher income are more likely to use credit cards. Controlling for the opportunity to
use credit cards (i.e., card ownership), Lindley, Rudolph, and Selby (1989) find little relation between the use
of a credit card and the cardholder’s income, gender, age, or education. To date, no empirical work analyzes
why consumers choose to use one type of credit card over another (proprietary versus bank card usage).
III. Methodology and Data Collection
Methodology: Logit Estimation Model
As shown by Whitesell (1989, 1992) and Santomero and Seater (1996), the choice of debitable
accounts depends upon the costs of accessing and using these accounts. Similar to Hirschman (1982),
these costs can be thought of as the advantages and disadvantages derived from the attributes of each of
the payment systems. McFadden (1974) develops the general utility model where consumers determine the
choice with the highest level of utility by viewing each choice as a bundle of attributes. The consumer
evaluates each choice by combining the value of the choice’s attributes through a utility function.
5
In the retail setting, consumers have the option to pay via cash, proprietary card, or bank card.
This set of options is denoted as A and the decision to choose one payment option i over any other option
may be written as
( ) ( )Pr Pri A U Ui j= > , (1)
which is the probability that the utility of option i is greater than the utility of option j for all j, which are
elements of the array of options available in set A where j is not equal to i. We assume a utility function
( )V X b b xi i i k ikk
m
= +=
∑1
. (2)
( )V Xi i is the function which links the vector X i of measured attributes of option i to a summary indicator
of the overall value of option i. Using the common assumption that ε i and ε j are independently and
identically distributed, Hensher and Johnson (1981) and McFadden (1981) define the multinomial logit model
of choice as
( )( )
( )Pr i Ae
e
v X
j Av X
i
j=
∈∑. (3)
The measured attributes of each payment option are the level of importance placed on the reason for
utilizing a payment option. Fisher and Nagin (1981) also state that the utility model may include a series of
characteristics for each individual. Thus, the consumer’s choice of a specific payment system may vary not
only with respect to the attributes of the different payment systems, but also with the demographic and
credit characteristics of the individual.
Data Collection
We conducted a mail survey of retail credit cardholders in late 1990–early 1991. The Credit
Research Center (CRC) at Purdue University designed and provided questionnaires, cover letters, and
envelopes to six major national retailers, who mailed 7,200 questionnaires to random samples of their
6
proprietary cardholders.2 Questionnaires were returned directly to CRC. The overall response rate was 23.3
percent (1678 total surveys). We reduce the respondent group by 200 due to missing responses on the
demographic and credit questions. In order to concentrate the analysis on those individuals with the
opportunity to substitute a bank credit card for their retail credit card, we remove an additional 188
respondents who do not have both types of cards, leaving 1290 useable responses.
We sent a similar questionnaire to a random sample of gasoline cardholders during the spring and
summer of 1992. CRC provided questionnaires, cover letters, and envelopes to twelve participating oil
companies with proprietary credit card programs.3 The companies mailed 24,000 questionnaires to random
samples of their cardholder base. Questionnaires were returned directly to CRC. The overall response rate
was 25.9 percent (6451 total surveys). We reduce the respondent group by 965 due to incomplete
demographic and credit information. We remove an additional 678 respondents who do not have both a
gasoline credit card and a bank credit card, leaving a total of 4808 useable responses.
We did not reveal the retailer or the oil company in the questionnaire. Consequently, the
responses of the consumers relate to their general use of gasoline or retail credit cards and not to their
behavior with respect to a specific company (unless they owned only one retail or gasoline card).
IV. Analysis of the Decision to Use a Bank Card versus a Proprietary Card
The critical questions on which this section focuses are:
(gasoline cardholder survey) “Over the past year when you purchased gasoline at stations where you had a gasoline company card, did you ever use a bank credit card instead of your gasoline card?” (retail cardholder survey) “Over the past year when you purchased at stores where you had credit accounts, did you use a bank card instead of your store card?”
Those who responded “yes” were then asked how frequently they substituted and were given a choice of
“almost always”, “often”, “sometimes”, and “hardly ever”.
12.7% of retail cardholders surveyed and 5.9% of gasoline cardholders surveyed did not have a
bank card. The other 87.3% of the retail respondents and 94.1% of the gasoline respondents had the
7
opportunity to use a bank card. Of those who had the opportunity to substitute, 43.3% of retail survey
respondents and 25.4% of gasoline survey respondents said that they had used their bank card instead of
their proprietary card during the previous year. Those individuals who substituted “almost always” or
“often” are defined as “frequent substituters”. Those respondents who did not substitute, or substituted
“hardly ever” or “sometimes” are referred to as “infrequent substituters”. Of those who substitute, 51.1% in
the retail survey did so frequently, while 29.2% in the gasoline survey did so frequently. Two summary
points are apparent. First, the data show that a substantial number of proprietary cardholders use their bank
cards in place of proprietary cards. Second, it appears that consumers in a retail store context are much more
likely to utilize their bank cards, and do so more frequently, than consumers purchasing gasoline, an
interesting distinction explored more fully in the following sections.
The detailed analysis to follow is broken into three parts. In the first section, we analyze card
substitution relative to the specific reasons given by respondents in the survey. The analysis is similar to
that of Hirschman (1982) in translating reasons for use into a finite set of payment attributes. However,
information concerning consumer preferences regarding payment attributes is not easily observable (in the
absence of a customer survey). In the second section, we relate the attributes of the payment alternatives to
easily observable demographic and credit characteristics of the respondents who list them as important
factors. Finally in the third section, we analyze the consumer’s payment choice as a function of their
demographic and credit characteristics. Through this three part analysis, we gain an understanding of why
consumers substitute a bank card for their proprietary card, what demographic and credit characteristics are
related to these reasons, and an understanding of why these observable demographic and credit
characteristics are able to be used to identify frequent substituters.
Reasons for Choosing a Bank Card versus a Proprietary Card
To determine the reasons for their choice of card, respondents were asked, “When you choose to
use a bank credit card instead of your gasoline [retail] card, what are your reasons?” For each of the
8
choices offered, respondents were asked to rate the importance of the reason as “very important”,
“somewhat important”, “not very important”, or “not a factor”.
Table 1 presents estimates from a logit model that relates substitution behavior to the set of
reasons for using cards. For purchases at a gasoline station (store) which accepts both types of cards, the
dependent variable is equal to 1 if the respondent is a “frequent substituter”, and 0 otherwise.4 Each of the
independent variables is equal to 1 if the respondent answered that the reason was “very important”, 2 =
“somewhat important”, 3 = “not very important”, 4 = “not a factor”, and 5 = no response. 5
Overall, the predictive power of the models is high, as indicated by the relatively high concordant,
log-likelihood values and the Hosmer and Lemeshow Goodness-of-Fit Test. The Hosmer and Lemeshow
Goodness-of-Fit Test divides the observations into 10 equal groups. From the estimated regression
function, it calculates the estimated number of observations that should appear in each grouping. Using a
chi-square statistic, it tests if the pattern of actual to expected observations is a good fit. Thus, the test does
not reject the null hypothesis that the model provides a good fit to the actual data.
Consider first the reasons for using a bank card instead of a proprietary card. The equations reveal
that lower monthly payments, rebates and enhancements, and the ease of shopping with one card are the
most important reasons to use a bank card. For the retail survey, consolidating records is also significant.
As expected, consumers who ranked these factors as more important (value = 1 or 2) are significantly more
likely to frequently substitute a bank card for a proprietary card. On the other hand, gasoline cardholders
who thought obtaining a lower monthly payment was an important reason to substitute are significantly
more likely to be infrequent substituters. Put another way, when a customer who seldom uses a bank card
for a retail or gasoline purchase chooses to use one, the reason is to obtain a lower monthly payment.6
We also ask respondents to give their reasons for using proprietary cards instead of bank cards.
For gasoline cardholders who regularly use their gas cards (infrequent substituters), there are two
significant factors associated with choice of the gasoline card: 1) to keep gasoline purchase records
separate from other purchases,7 and 2) to keep credit available on the bank card.8 Consumers who view
these factors as important are significantly more likely to use their gasoline card.
9
Frequent store card users perceive that store card use makes it easier to return merchandise or get
sales notices, as compared to frequent substituters who either do not perceive such benefits or do not value
them. Thus, there is strong evidence that some consumers feel that they are more likely to be treated well by
retailers if they use the credit card from the retailer, and this is an important motivation to use the store card.
To summarize, consumers who frequently utilize a bank card in either a retail or gasoline outlet do
so to obtain rebates and shop with fewer cards. Consumers who frequently utilize their retail cards do so to
simplify merchandise returns, obtain sales notices, and generally be recognized as a loyal customer.
Consumers who frequently utilize their gas cards do so to maintain separate records of their gas purchases
and keep other credit lines open.9
Relationships of Reasons for Usage to Demographic and Credit Characteristics
Customer reasons for using a card provide important insight into the payment choice, but reasons
are not easily observable. In this section, we use customer demographic and credit characteristics to test
hypotheses concerning the observable characteristics of individuals who feel strongly about different
attributes of payment. In Tables 2 and 3, we only include reasons that are significant at the one percent
level in Table 1. For each reason, we construct a dependent variable equal to 1 if the respondent views the
reason as “very important” or “somewhat important”, and 0 otherwise.10 We estimate a logistic regression
relating each of these dependent variables to a set of demographic and credit variables.11
Reasons to Us e Bank Credit Cards: In Table 2 we analyze the perceptions of bank cards versus proprietary
cards in four areas: 1) rebates and credit card enhancements, 2) the ease of shopping with one card, 3)
consolidation of records, and 4) lower monthly payments. For both surveys, respondents with higher
annual incomes and more bank cards are more likely to use a bank card (instead of a proprietary card)
because of its rebate feature. Convenience users (those who typically pay the full amount on each month’s
card statement in the gasoline survey) are more likely to use a bank card because of its rebate feature.
Revolvers in both surveys are less likely to consider the rebate feature important to their card choice.
10
In both surveys, respondents over the age of 65 (and respondents under the age of 25 in the
gasoline survey) are less likely to view the ease of shopping with one card as an important reason to use a
bank card. The ease of shopping with one card matters less to those who own more proprietary cards, but is
important to those who own more bank cards. Similarly, convenience users are more likely to value
shopping with just one card than revolvers.12 Not surprisingly, factors related to the importance of
consolidating record keeping resemble those related to the importance of shopping with just one card.
Convenience users consider the lower monthly payments on bank cards unimportant. The
importance of lower monthly payments is significantly and negatively associated with income and age,
possibly because younger consumers and those with lower incomes are more likely to be liquidity
constrained (Duca and Rosenthal, 1993).13
Reasons to Use Proprietary Cards Instead of Bank Credit Cards: Table 3 analyzes the perception of
proprietary versus bank cards in three areas: 1) ease of returning merchandise, 2) ability to keep gasoline
purchases separate, and 3) keeping other credit lines open. Respondents who emphasize that the ease of
returning merchandise is an important reason for using the retail card are under the age of 25, over 55,
married, own more retail cards, or occasionally revolved their credit card balances. Gasoline survey
respondents who considered record keeping important have more education, are more likely to be female,
and own more gasoline cards. Convenience users and regular revolvers are significantly less likely to
consider separate record keeping important, relative to those who occasionally revolve. In general,
respondents in both surveys who want to conserve their bank lines tend to be younger, have lower
incomes, and to revolve their accounts. These results are consistent with expectations that younger
respondents with lower incomes are more likely to be credit constrained [Jappelli (1990) and Fissel and
Jappelli (1990)].
Substitution Frequency Using Bivariate Logit Model
11
Table 1 links the reasons provided by the respondents for choosing a particular card to the
frequency of card usage. Tables 2 and 3 show that many of the credit variables as well as some of the
demographic characteristics are significantly related to the importance of the card attributes (reasons for
choosing a particular card). Once a linkage to attribute value is established, we can explain the association
of demographic and credit characteristics with the substitution behavior. We use the logitistic regressions
displayed in Table 4 to test the relationships between the observable consumer characteristics and the
frequency of substitution. Since the goal is to utilize only observable characteristics, we do not include
attributes in this model.14 The dependent variable for the equation in Table 4 is the same dependent variable
used in Table 1. Specifically, for purchases at a retail outlet (store or gasoline station) accepting both the
respondent’s proprietary card as well as the respondent’s bank credit card, the dependent variable is equal
to 1 if the respondent is a “frequent substituter”, and 0 otherwise.
Of all the demographic variables in Table 4, only income is significantly associated with card choice
in both surveys. Low income respondents are less likely to use their bank cards. In Tables 2 and 3, income
was most highly correlated with a desire for lower monthly payments and to keep other credit lines open.
Throughout Tables 2 and 3, credit characteristics are more consistently associated with the
reasons for card usage than are the demographic characteristics. Similarly, the credit characteristics also
explain more of the variation in the dependent variables in Table 4.15 The more bank cards owned, the more
likely the individual is to use their bank card. Similarly, the more proprietary cards held, the more likely the
individual is to use their proprietary card. Thus, the respondents reveal their preference for a particular type
of card through the cards for which they apply, making current cardholdings a valuable, if obvious, signal of
willingness to take and use another card of a given type.
Previous results showed that consumers who are more likely to revolve their credit cards place a
low level of importance on payment consolidation and rebate features of bank cards, while placing a high
level of importance on keeping other credit lines open. The results from both the gasoline and the retail
credit card surveys indicate that revolvers are more likely to utilize their proprietary credit card at retail
12
outlets. Together, these findings are consistent with a credit constrained individual utilizing a wider variety
of credit cards in order to keep their line of credit open on their bank cards.
V. Summary and Conclusion
The primary question of interest has been the following: For a sample of customers who own both
a proprietary card and a bank card, how often do they substitute the bank card in place of the proprietary
card, and why? We have assembled evidence on customer choice of credit cards (proprietary vs. bank) in
two different retailing contexts: retail department stores and gasoline stations. Our empirical approach takes
the decision as to whether to acquire a proprietary card as given. In this research, we address which type of
card consumers choose in outlets that accept both. However, the results (particularly the reasons for using
one over the other) can probably be generalized to yield insights about who would acquire and use a
proprietary card in today’s market, and why.
Retail cardholders are less loyal to their proprietary cards than are gasoline cardholders. Over 40
percent of retail cardholders had used a bank card instead of their store card at least once in the previous 12
months, compared to only 27 percent of gasoline cardholders. Of those who did substitute, 51 percent of
the retail cardholders did so frequently compared to only 29 percent of gasoline cardholders.
Low income cardholders tend to use their proprietary cards more faithfully than individuals with
higher incomes. It could be that lower limits on their bank cards leave them “credit constrained,” such that
they utilize the store and gas card credit function more than other customers who use their proprietary
cards.
Customers who regularly revolve balances on their bank cards also do not substitute away from
their proprietary cards as often as those who only occasionally or never revolve. Again, the reason may be
similar to that for low income consumers: the fact that a cardholder revolves regularly suggests that he/she
is relatively more credit constrained than other cardholders, and wishes to utilize the credit function of
store/gas cards as opposed to other attributes. Given that revolvers generate finance charge income, this
may be an important insight for proprietary issuers in targeting active, profitable customers and may also
13
explain the continued profitability of proprietary card units through the 1990’s even as ownership and use of
co-branded bank cards grows. Caution should also be noted as credit constrained consumers have greater
risk of default than non-credit constrained consumers.
The most intriguing results of the study arise from the contrast across surveys in the reasons for
substituting a bank card (for those that do so frequently), and also for continuing to use the proprietary
card (for those that do not substitute frequently). For both retail and gasoline cardholders, the primary
reasons for using a bank card are the same: 1) the convenience and ease of consolidating purchases onto
one card, and 2) to obtain rebates.
However, the reasons for sticking with the proprietary plastic diverged sharply between the two
types of cardholders. Gasoline cardholders used their gas cards to keep their gasoline purchases separate
from other purchase records, and, secondarily, to keep credit available on their bank card. One implication
of this is that monthly statement format is likely to be important to gasoline cardholders. In contrast, store
cardholders use their store cards because they feel it brings them better treatment/service from the merchant
(easier to return merchandise, sales notices from the retailer, etc.). Apparently, retail store card users want
to be identified as loyal customers.
These results have particular relevance for the trend toward co-branding between merchants and
bank card issuers. Compare the impact of offering a co-branded, rebate card to traditional store or gasoline
cards. For store cardholders, a co-branded rebate card is likely to appeal to three distinct groups: 1) those
with a desire to consolidate cards carried, 2) those who are attracted by rebates, and 3) those who want to
be identified as loyal customers (affinity buyers). That is, the co-branded store card appeal cuts across the
set of reasons that characterize the behavior of both substituters and non-substituters in the retail store
environment.
On the other hand, the co-branded gasoline card prospects (for converting/preserving existing
cardholders) do not seem as encouraging. First, the frequency of gas card substituters is lower: there is a
smaller base of proprietary cardholders to whom consolidation via a co-branded relationship appeals, at
least based on revealed choice of plastic. Moreover, the co-mingling of purchase records which is typical
14
on bank card statements is at odds with the desires of those who like to use a gasoline card because it keeps
their gas purchases separate. This motivation is at the core of the loyal gas card-user population. Finally, a
co-branded card does not necessarily appeal to the secondary reason for gasoline cardholder loyalty:
utilization of the credit function of the gas card to keep lines open on their bank card.
To satisfy the rebate-oriented gas card customers, while at the same time generating a brand-name
presence in customer wallets (and repeat purchases), it seems an internal rebate program administered on
the proprietary gas card (rather than through a co-branded gasoline/bank card) might be relatively more
effective than such a program on a store card. Of course, this strategy risks losing cardholders altogether
to the industry-wide trend toward consolidation of plastic in wallets (i.e., exclusive use of a bank card). But,
the greater propensity of gasoline cardholders to use their cards should offer encouragement to those oil
companies debating the future of their proprietary card programs.
15
Notes
1 Flannery (1996) reviews many of the changes over the past two decades. Also see Berger and
Humphrey (1988), Horvitz (1996), and Mayer (1996) for predictions concerning the future of the payment system.
2 Participating retailers included: Federated Department Stores, Firestone, Hartmarx, The Limited, JC Penney, and Spiegel.
3 Participating companies included: Mobil, Total, United, Citgo, Phillips, Marathon, Cenex, Fina, Exxon, Unocal, Conoco, and Chevron.
4 Use of an ordered logit model with a dependent variable equal to 1 if the respondent substituted “almost always”, 2 = “often”, 3 = “sometimes”, and 4 = “hardly ever”, provides similar conclusions as the binomial logit model.
5 The results of the following logit regressions are not significantly affected by the inclusion or the exclusion of those respondents who did not respond to every one of these questions. Omitting the non-responses or including a series of indicator variables for each response does not significantly affect the conclusions from this survey. For a copy of these results, contact the authors.
6 Many retail cards have substantially higher minimum payments than bank cards. For example, a minimum payment of 10-20 percent of the outstanding balance is common on retail cards, but bank cards typically require only a 2-5 percent minimum payment. The difference is even more dramatic between bank cards and gasoline cards. Many gasoline cards do not permit fuel purchases to revolve, which means up to 100 percent minimum payment for many users.
7 This preference has not been lost on card issuers. The co-branded Shell MasterCard (introduced in 1992) does separate gasoline purchases from other purchases on the monthly statement.
8 Jappelli (1990) has shown that approximately 30% of the U.S. population is credit constrained (quantity of credit demanded exceeds quantity available). Individuals who are credit constrained may have incentive to utilize their proprietary cards more frequently, since it preserves a line of credit on their more widely accepted bank card. The relation between household demographics of our samp le and card choice is discussed below.
9 Overall, the predictive power of the models is high, as indicated by the relatively high concordant, log-likelihood values and the Hosmer and Lemeshow Goodness-of-Fit Test. The Hosmer and Lemeshow Goodness-of-Fit Test divides the observations into 10 equal groups. From the estimated regression function, it calculates the estimated number of observations that should appear in each grouping. Using a chi-square statistic, it tests if the pattern of actual to expected observations is a good fit. Thus, the test does not reject the null hypothesis that the model provides a good fit to the actual data.
10 Similar results are also obtained if an ordered logit model is estimated where 1 is equal to very important, 2 = somewhat important, 3 = not very important, 4 = not a factor, and 5 = no response. Similar results are also found if the non-responses are omitted from the analysis.
16
11 For the series of categorical variables (age, education, and income), the middle category is omitted.
The insignificance for each of the variables included in the model for age and education shows that there is not a significant difference between the omitted category and any of the other categories for these variables.
12 Revolvers may utilize several cards as a way to extend their total credit.
13 Some gasoline card programs do not allow fuel purchases to revolve, reinforcing the advantage of a bank card for lowering monthly payments. This may explain the significance in the gasoline survey, but not the retail survey.
14 As would be expected, inclusion of the attributes reduces (but does not eliminate) the significance of the demographic and credit characteristics.
15 We test this by comparing two separate logit models: one with only demographic characteristics and another with only credit characteristics.
17
References
Berger, Allen and David Humphrey (1988), Interstate Banking and the Payments System. Journal of Financial Services Research 1: 131-145. Boeschoten, Willem (1992), Currency Use and Payment Patterns. In Financial and Monetary Policy Studies, Vol. 23 (Dordrecht, The Netherlands: Kluwer Academic Publishers, pp. 4-41). Canner, Glenn and Anthony Cyrnak (1985), Recent Developments in Credit Card Holding and Use Patterns Among U.S. Families. Journal of Retail Banking 7: 63-74. Duca, John and Stuart Rosenthal (1993), Borrowing Constraints, Household Debt, and Racial Discrimination in Loan Markets. Journal of Financial Intermediation 3: 77-103. Duca, John and William Whitesell (1995), Credit Cards and Money Demand: A Cross-Sectional Study. Journal of Money, Credit, and Banking 27: 604-623. Fisher, Gregory and Daniel Nagin (1981), Fixed Versus Random Coefficient Quantal Choice Models. In Manski and McFadden (eds.), Structural Analysis of Discrete Data with Econometric Application (Cambridge, MA: MIT Press, pp. 273-304). Fissel, Gregory and Tullio Jappelli (1990), Do Liquidity Constraints Vary Over Time? Evidence from Survey and Panel Data. Journal of Money, Credit, and Banking 22: 253-261. Flannery, Mark (1996), Technology and Payments: Déjà Vu All Over Again? Journal of Money, Credit, and Banking 28: 965-970. Hensher, David and Lester Johnson (1981), Applied Discrete-Choice Modeling (New York: Wiley Publishing). Hirschman, Elizabeth (1982), Consumer Payment Systems: The Relationship of Attribute Structure to Preference and Usage. Journal of Business 55: 531-545. Horvitz, Paul (1996), Preserving Competition in Electronic Home Banking. Journal of Money, Credit, and Banking 28: 971-974. Jappelli, Tullio (1990), Who is Credit Constrained in the U.S. Economy. Quarterly Journal of Economics 105: 219-234. Lindley, James, Patricia Rudolph and Edward Selby Jr. (1989), Credit Card Possession and Use: Changes Over Time. Journal of Economics and Business 41: 127-142. Martell, Terrence and Robert Fitts (1981), A Quadratic Discriminate Analysis of Bank Credit Card User Characteristics. Journal of Economics and Business 33: 153-159. Mayer, Martin (1996), The Past is a Lousy Prologue: Payments Systems Innovations and the Fed. Journal of Money, Credit, and Banking 28: 975-979. McFadden, Daniel (1974), The Measurement of Urban Travel Demand. Journal of Public Economics 3: 303-328.
18
McFadden, Daniel (1981), Econometric Models of Probabilistic Choice. In C. Manski and D. McFadden (eds.), Structural Analysis of Discrete Data (Cambridge, Mass.: MIT Press). Santomero, Anthony and John Seater (1996), Alternative Monies and the Demand for Media of Exchange. Journal of Money, Credit, and Banking 28: 942-960. Whitesell, William (1989), The Demand for Currency Versus Debitable Accounts. Journal of Money, Credit, and Banking 21: 246-251. Whitesell, William (1992), Deposit Banks and the Market for Payment Media. Journal of Money, Credit, and Banking 24: 483-498.
19
Table 1
Logit Model Relating Frequency of Substitution to Reasons for Card Choice
The logistic regression analyzes the frequency with which respondents substitute a general purpose credit card in place of a proprietary credit card as it relates to respondents’ reasons for using cards. The dependent variable is equal to 1 if the respondent is a “frequent substituter” and 0 otherwise. Estimates are presented for both the gasoline cardholder and retail cardholder samples.
Gasoline Survey1 Retail Survey2 Parameter
Estimate
T-Statistic Parameter
Estimate
T- Statistic
Constant -1.29 (-4.16) ** -1.22 (-2.01) * Reasons for Using General Purpose Credit Cards:3
Lower Interest Rates 0.06 (0.95) -0.07 (-0.91) Lower Monthly Payment 0.29 (3.43) ** 0.18 (2.04) * Rebates/Enhancements -0.57 (-11.53) ** -0.20 (-2.90) ** Ease of Shopping with One Card -0.66 (-13.25) ** -0.48 (-5.84) ** Paid Fee; Use Card -0.11 (-1.66) -0.11 (-1.28) More Credit -0.09 (-1.11) Consolidate Records -0.46 (-6.62) **
Reasons for Using Proprietary Cards:3
Record Keeping 0.59 (11.58) ** Keep Account Active to Get Sales Notices -0.12 (-1.75) 0.14 (1.82) Keep Other Lines Open 0.20 (3.22) ** 0.20 (2.71) ** Not Sure if General Card is Accepted -0.04 (-0.82) Easier to Return Merchandise 0.47 (5.96) ** Better Service from Salespeople 0.06 (0.63) Don’t Own Acceptable General Purpose Card 0.15 (1.06)
1 The samples are limited to those respondents who have used a credit card within the past year for a gas purchase and who own
at least one general purpose credit card and one gasoline credit card. 2 The samples are limited to those respondents with at least one general purpose credit card and one retail credit card. 3 Respondents ranked the importance of each reason as: (1) very important, (2) somewhat important, (3) not very important, (4)
not a factor, and (5) no response. * Significant at the .05 level of confidence.
20
** Significant at the .01 level of confidence.
Table 2 Respondent Characteristics and Reasons for Using Bank Cards
The logistic regressions relate the demographic and credit characteristics of respondents to reasons for using a bank card instead of a proprietary card. The dependent variable for each logistic regression is equal to 1 if the respondent viewed the reason as “very important” or “somewhat important” and 0 otherwise. See the footnotes to Table 4 for a description of the independent variables.
Rebates and Enhancements Ease of Shopping with One Card Gasoline Survey Retail Survey Gasoline Survey Retail Survey Parameter Parameter Parameter Parameter Estimate T- Statistic Estimate T- Statistic Estimate T- Statistic Estimate T- Statistic
Intercept -2.26 (-8.90) ** -1.62 (-4.25) ** -0.80 (-4.15) ** 0.48 (1.43) Age of Respondent:
Male 0.14 (1.47) 0.02 (0.12) 0.31 (4.17) ** 0.08 (0.66) Home Owner 0.14 (1.01) -0.25 (-1.28) -0.03 (-0.28) -0.18 (-1.10)
Card Ownership:
Number of Bank Cards 0.16 (6.75) ** 0.23 (4.66) ** 0.06 (3.11) ** 0.08 (1.71) Number of Travel and Entertainment Cards 0.03 (0.32) 0.13 (1.45) -0.13 (-2.11) * 0.17 (1.95) Number of Gas Cards -0.05 (-2.15) * -0.11 (-5.69) ** Number of Retail Cards 0.03 (1.00) -0.07 (-2.71) **
Male -0.05 (-0.42) 0.29 (2.67) ** 0.12 (0.78) Home Owner -0.06 (-0.37) 0.13 (0.86) -0.13 (-0.71)
Card Ownership:
Number of Bank Cards 0.10 (2.16) * 0.02 (0.78) 0.14 (2.77) ** Number of Travel and Entertainment Cards 0.15 (1.82) 0.12 (1.24) 0.26 (2.66) ** Number of Gas Cards -0.12 (-3.86) ** Number of Retail Cards -0.03 (-1.27) 0.01 (0.27)
Table 3 Respondent Characteristics and Reasons for Using Proprietary Cards
The logistic regressions relate the demographic and credit characteristics of respondents to reasons for using a proprietary card instead of a bank card. The dependent variable for each logistic regression is equal to 1 if the respondent viewed the reason as “very important” or “somewhat important” and 0 otherwise. See the footnotes to Table 3 for a description of the independent variables.
Easier to Return Merchandise Keep Gas Purchases Separate Keep Other Lines Open Retail Survey Gasoline Survey Gasoline Survey Retail Survey Parameter Parameter Parameter Parameter Estimate T- Statistic Estimate T- Statistic Estimate T- Statistic Estimate T- Statistic
Intercept -0.63 (-1.91) 0.60 (3.31) ** -0.59 (-3.15) ** -0.06 (-0.16) Age of Respondent:
Male -0.18 (-1.41) -0.21 (-3.08) ** -0.05 (-0.71) -0.04 (-0.33) Home Owner 0.11 (0.66) 0.04 (0.38) -0.14 (-1.36) -0.06 (-0.35)
Card Ownership:
Number of Bank Cards 0.03 (0.61) -0.02 (-0.88) 0.03 (1.11) 0.11 (2.32) * Number of Travel and Entertainment Cards 0.00 (0.06) 0.10 (1.65) 0.05 (0.75) -0.01 (-0.06) Number of Gas Cards 0.08 (4.19) ** 0.05 (2.85) ** Number of Retail Cards 0.06 (2.35) * 0.02 (0.90)
Logit Model of Frequency of Substituting a Bank Card for a Proprietary Card
The logistic regression analyzes the characteristics of respondents who are likely to substitute bank credit cards for a proprietary credit card. The dependent variable is equal to 1 if the respondent is a “frequent substituter” and 0 otherwise.
Gasoline Survey1 Retail Survey2 Parameter
Estimate
T-Statistic Parameter
Estimate
T-Statistic
Constant -3.61 (-10.40) ** -0.56 (-1.39) Age of Respondent: 3
Married 0.00 (-0.02) -0.06 (-0.26) Divorced -0.32 (-1.26) -0.20 (-0.68)
Miscellaneous Demographics:6
Male 0.36 (2.66) ** -0.08 (-0.52) Home Owner -0.20 (-1.12) -0.40 (-1.96) *
Card Ownership:7
Number of Bank Cards 0.16 (5.12) ** 0.11 (1.93) Number of Travel and Entertainment Cards -0.16 (-1.41) 0.27 (2.69) ** Number of Gas Cards -0.09 (-2.59) **
Number of Retail Cards -0.13 (-3.99) **
Table 4 (continued)
Logit Model of Frequency of Substituting a Bank Card for a Proprietary Card
Gasoline Survey1 Retail Survey2 Parameter
Estimate
T-Value Parameter
Estimate
T-Value
Frequency of Paying Credit Card Balance:8 Pay Full Amount 1.41 (9.01) ** 0.46 (2.69) ** Hardly Ever Pay Full Amount -0.50 (-1.91) -0.75 (-3.65) **
1 The samples are limited to those respondents who have used a credit card within the past year for a gas purchase and who own
at least one bank credit card and one gasoline credit card. 2 The samples are limited to those respondents with at least one bank credit card and one retail credit card. 3 Respondents in the 35-44 age group are the omitted category. 4 Respondents with a high school diploma are the omitted category. 5 Respondents with $30,000-$49,999 income are the omitted category. 6 Indicator variables are equal to 1 if the respondent has the characteristic, and 0 otherwise. 7 Equal to the number of the type of card owned by the respondent. Values are from 1 to 10; respondents who answered they had
more than 10 cards are given the value of 10. 8 Respondents who stated that they sometimes paid their balance in full are the omitted category. * Significant at the .05 level of confidence. ** Significant at the .01 level of confidence.