The Impact of Credit Decisions on Shopping Behavior

James W. Gentry, Oklahoma State University
ABSTRACT - This study investigates the relationship between patronage decisions made by consumers and credit decisions mate by retailers. Part of the rationale behind the decision by retailers to maintain house accounts is that store loyalty is increased through the use of the store's or chain's own card. A survey of consumers' reactions to credit decisions yielded estimates of (1) the loss in patronage from current levels due to a rejection and (2) the opportunity loss in the increase in patronage that would have occurred had the application been accepted. The shifts in patronage were statistically significant, indicating that credit decisions do affect Patronage.
[ to cite ]:
James W. Gentry (1982) ,"The Impact of Credit Decisions on Shopping Behavior", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 385-388.

Advances in Consumer Research Volume 9, 1982      Pages 385-388

THE IMPACT OF CREDIT DECISIONS ON SHOPPING BEHAVIOR

James W. Gentry, Oklahoma State University

[The author would like to acknowledge the financial support of the Bureau of General Research, Kansas State University and the Dean's Excellence Funds, College of Business Administration, Oklahoma State University. Additionally, the author would like to express appreciation to Jack Chalfant and Daniel Braum for their help in the data collection and to Charles W. Caldwell, Raymond L. Smead, and U. Gary Simpson for their helpful comments.]

ABSTRACT -

This study investigates the relationship between patronage decisions made by consumers and credit decisions mate by retailers. Part of the rationale behind the decision by retailers to maintain house accounts is that store loyalty is increased through the use of the store's or chain's own card. A survey of consumers' reactions to credit decisions yielded estimates of (1) the loss in patronage from current levels due to a rejection and (2) the opportunity loss in the increase in patronage that would have occurred had the application been accepted. The shifts in patronage were statistically significant, indicating that credit decisions do affect Patronage.

Marketers have long realized that credit can be a promotional tool as well as a financial tool (Kaplan, 1967). Commonly stated reasons for the existence of a liberal retail credit policy include (1) customers could not afford to purchase many goods without credit; (2) consumers desire the convenience that credit cards offer; (3) credit tends to make shoppers less frugal and more impulsive; and (4) credit creates patronage loyalty (Cooper, 1977). In general, these reasons have been supported. For example, Hendrickson (1972, p. 49) summarizes a discussion with the President of Montgomery Ward, who estimated that sales of appliances, furniture, tires, and other items would drop 35 percent to 50 percent if credit were suddenly discontinued. The convenience aspect of credit cards is well documented (Cole, 1972, p. 70). While it has been found that credit customers make more purchases than cash customers (Bonk, 1975; Dunkelberg, 1974; Hendrickson, 1972; Stephenson and Willet, 1969), Bonk (1975) pointed out that credit sales may be merely former cash sales as credit is issued to higher income customers who are more likely to purchase anyway. Hirschman (1979) studied purchase volume by consumers of a department store chain, controlling for different levels of demographics. In several of the demographic control conditions, consumers that made at least one purchase using a bank card or a store-issued charge card did have a higher total dollar volume than strictly cash customers.

Store loyalty has been found to result from the availability of a charge account or an in-house credit card (Stephenson and Willet, 1969; Wong, 1969). The belief that store loyalty is fostered by a house account has been widely held by many department stores as evidenced by their refusal to accept bank cards. Hirschman, in an interview in Business Week (1979, p. 132), stated that a store charge builds a link to a consumer with a credit line that can be used there." On the other hand, Pessemier (1980, p. 39) points out that "legal restrictions and the ease with which competitors can match new credit services normally tend to limit the suitability of credit as a patronage generating competitive weapon." He further added that credit is frequently overlooked as a patronage tool by merchandising personnel since it is not controlled by the merchandising function. Furthermore, Hirschman (Business Week, p. 132) noted that customers are more mobile and less store loyal than previously. In order to attract these mobile shoppers, the retailer has to give them the method of payment that they prefer. Consequently, Penney's started to change their long held policy and started accepting Visa cards in 1979. Several other department store chains are contemplating similar changes. However, the chains are keeping their own credit cards and they are still actively pushing the use of the house carts. For example, the R. H. Holmes Co. in New Orleans pays sales clerks $2 for every bank card user whom they can convert to a Holmes charge customer (Business Week, 1979).

All things considered, marketers are generally supportive of a liberal credit policy. As such, they would like a credit extension process that does not foreclude potentially good customers. The most commonly used credit scoring approach is discriminant analysis, which yields the maximum amount of separation between "good" and "bad" credit risks. In typical usage, the cutoff score (the score that will be compared to a new credit applicant's score when his/her characteristics are put in the discriminant function) has been the mid-point between the two groups' centroids. The decision theoretic approach to discriminant analysis proposed by Walt (1944) adjusts the cutoff point by considering (1) the cost of misclassifying a bad risk as a good risk, (2) the cost of misclassifying a good risk as a bad risk, and (3) the relative prior probabilities of membership in the two groups. Those associated with the finance function have emphasized the first consideration, as bad debts are a very tangible cost consideration in the tradeoff between the costs of the credit operation and the finance charges generated. Further, the number of bad debts occurring serves as a basis for evaluating a credit manager. On the other hand, since there is rarely a follow-up made on rejected candidates, errors made in misclassifying a "good risk" are much more difficult to uncover.

Marketers emphasize the last two considerations. The relative prior probabilities of being a good as opposed to a bad risk favor a more liberal credit policy, as some researchers have suggested that this ratio may be as high as 30:1 or 40:1 (Myers, 1962). While marketers are strongly opposed to the denying of credit to good risks, the determination of this misclassification cost is a very complex task. Models have been proposed (Caldwell, Gentry, and Miller, 1980; Greer, 1967, 1968; Long, 1975; Metha, 1968, 1970) to estimate this cost. Critical inputs in these models are (1) the increase in patronage that occurs after accepting the good risk and (2) the decrease in patronage that occurs after rejecting the good risk. If there are no shifts in total dollar expenditures due to the credit decision, then the misclassification cost may well be negative (i.e., the firm is better of by rejecting all credit applicants), as Dunkelberg (1974) reported figures which indicate that credit extension costs may exceed the revenue obtained from finance charges.

Thus tools exist to help marketers operationalize their inputs into the credit scoring process. Whether credit scoring models should incorporate the means for determining misclassification errors depends on a more basic question: how significant are the effects of credit decisions on patronage behavior. This study will investigate the nature of patronage shifts due to credit decisions. More specifically, the hypotheses to be tested are as follows:

H1: It is expected that there will be an increase in relative patronage at a retail outlet when an individual's credit application is accepted.

H2: It is expected that there will be a decrease in relative patronage at a retail outlet when an individual's credit application is rejected.

RESEARCH METHODOLOGY

The research took the form of a questionnaire covering the respondent's credit history.

Population Sampled

The questionnaire was distributed in married student housing at a state university in a midwestern state, and 213 households completed the eleven-page instrument. A second smaller sample was taken of 94 households spread throughout the college town of 30,000; the purpose of this smaller sample was to provide a basis of comparison with the primary sample. There are several reasons for the choice of married students as the primary sample. They are easily accessible and are willing participants for the most part. Most important, though, is that this stage of the family life cycle, young married couples, is frequently the stage in which individuals first encounter the need for credit. Thus, they are more likely to be able to recall their experiences with acceptances and rejections and to recall their shopping behaviors. Further, they pose a challenging task for credit managers since the decision to extend credit to them has to be made on the basis of the credit application alone due to the fact that, for the most part, they do not have a credit record that can be obtained from a credit bureau. Also, in one sense they are "risks" due to their currently limited financial capacity while, at the same time, they are potentially very good customers if long run commitments develop due to their appreciation for the availability of credit early.

The choice of young married couples as opposed to non-college couples may limit the ability to generalize the results of this study since they are under-privileged members of a higher social class who, in the near future, are going to have a substantially increased income. It may be that their reaction to a credit rejection would be more severe than the general population's. The sample of townspeople was acquired with the intention of exploring this source of bias.

Sampling Plan

Every apartment in married housing was contacted at least once, but there was a fairly high level of non-response due to the residents not being at home. A second attempt was made to reach each household that was not successfullY contacted in the first wave of sampling. Non-response due to refusals was limited to two households; the respondents were paid a dollar for their participation.

The smaller survey of townspeople consisted of randomly selecting certain blocks within each of the city's census tracts. The quota for that tract was determined by the proportion of townspeople living in that tract. A verbal commitment to complete the questionnaire was obtained and it was left with the respondent The interviewer returned to pick up the completed questionnaire an hour or so later.

Questionnaire Content

The questionnaire contained a variety of questions, many of. which are not pertinent to the study at hand. For example, the respondents were asked about their reasons for seeking credit (or their reasons for not seeking credit if they had never applied), and whether they used various types of credit (oil companies, retail, bank cards) as convenience or intallment users. The questions of prime importance to this study dealt with their history of acceptances or rejections, and their recall of their patronage at the stores before and after the credit decision was made. An example of the question format is shown in Figure 1. Questions dealt with their first, second, most recent, and most frequently used credit cards (or accounts) for retail outlet and oil company acceptances, and with their first, second, and most recent rejections. Obviously, many of the questions were not applicable to all respondents.

FIGURE 1

EXAMPLE QUESTION RELATING PATRONAGE WITH CREDIT ACCEPTANCE

RESULTS

The percentage of goods purchased at a particular outlet or chain before and after credit acceptances are shown in Table 1. Table 2 reports similar percentages before and after credit rejections. As might be expected, the sample of married students had less experience with credit than had the sample of townspeople: only 30 percent of the married students had at least one credit card compared to 44 percent of the townspeople while 37 percent of the students had successfully obtained retail credit at least once compared to 59 percent of the town sample. In terms of rejections though, the student sample had a least as much experience: 21 percent of the student sample had experienced at least one retail rejection and 12 percent had been rejected by at least one oil company compared to 20 percent and 1 percent, respectively, for the town sample. The lower percentage of (recalled) rejections for the town sample may be due to (1) problems in recalling events that occurred many years ago (the acceptances; on the other hand frequently involve tangible reminders in the form of credit carts) ant/or (2) the fact that the initial credit applications were made at later stages in life due to the avoidance of credit by these people initially or to the fact the many forms of credit (bank carts, for example) did not become popular until recently.

As indicated in Table 1, Hypothesis 1 is supported as retail patronage does increase significantly (p < .05) after credit acceptances. For example, the aggregate results for retail acceptance indicate that about one-third of the respondent's purchases in a store's range of merchandise were made in the store or chain issuing credit after the acceptance as opposed to about one-fourth of the purchases prior to the credit application. In terms of oil company patronage, the approximate aggregate increase was from 35 percent to 45 percent. The mean percentage levels before and after a retail acceptance are fairly constant across the order of application, whereas the before and after patronage levels are much higher for the first gasoline credit card. The retail credit acceptances were more likely to be independent in terms of percentage patronage as there was a greater variety of product lines available (for example, discount store vs. department store vs. small specialty shop).

TABLE 1

PATRONAGE LEVELS BEFORE AND AFTER CREDIT ACCEPTANCE

Table 2 indicates that a significant decrease in patronage takes place after a credit rejection, thus supporting Hypothesis 2. The reported patronage levels prior to the rejected credit applications are very close to the reported patronage levels prior to the accepted credit applications, as one would expect. The aggregate results indicate a decrease in retail patronage from 23 percent to Table 2 indicates that a significant decrease in patronage takes place after a credit rejection, thus supporting Hypothesis 2. The reported patronage levels prior to the rejected credit applications are very close to the reported patronage levels prior to the accepted credit applications, as one would expect. The aggregate results indicate a decrease in retail patronage from 23 percent to The preceding analysis assumes that the patronage responses represented interval data. Recognizing that this may well be an overly optimistic assumption, the before and after distributions (obtained by aggregating over the responses to all relevant items; for example, all gasoline credit card acceptances) were compared using the Kolmogorov-Smirnov test. The responses were grouped into the following categories: 0-5 percent, 6-10 percent, 2140 percent, 41-80 percent and 81-100 percent. The changes in the student samples patronage patterns were not significant after a gasoline credit card acceptance but they were significantly (p < .001) changed by a retail or oil company rejection as well as a retail acceptance. The town sample's patterns were significantly different after a gasoline credit card and retail acceptance and after a retail rejection. Consequently, the non-parametric analysis generally supported the findings generated by Comparing the mean patronage levels.

TABLE 2

PATRONAGE LEVELS BEFORE AND AFTER CREDIT REJECTION

SUMMARY AND IMPLICATIONS

This study provides empirical support for a relationship between consumer patronage and credit decisions. Individuals rejected for credit indicated a significant decrease in patronage while those accepted for credit indicated a significant increase. These results certainly are not unpredictable. They do support the commonly held belief that credit decisions do affect store loyalty. As pointed out by one of the reviewers, store patronage is affected by a variety of factors, of which the availability of credit is only one. The retailer's credit policy is somewhat independent of other marketing variables, as it is less directly observable.

To the extent that the results (that respondents' recall of behavior indicated significant patronage shifts) of this study are generalizable, it would seem that credit availability does have a significant effect on patronage. The results indicate that a retail rejection may mean a patronage decrement from 34 percent of the individual's purchases (had he/she been accepted) to 15 percent and that an oil company rejection may mean a patronage decrement from 45 percent to 17 percent. The small sample size and the nature of the sample clearly limit the meaningfulness of these estimates, even though the results from the married student sample were very similar to those of the town sample.

These estimates fall well short of estimating the dollar amount associated with a credit rejection or acceptance, as no attempt was made to collect data on the respondents' level of spending at the different stores. In the study undertaken, this attempt would have had little or no value as the study dealt with the respondents' overall credit history with a variety of retail outlets. The sample size for any one outlet is not sufficient to provide a meaningful estimate.

An alternative approach would be to use panel data to measure shifts in patronage and to measure the dollar amounts spent at the various stores. One problem with the use of panel data, though, is that credit acceptances and especially credit rejections are not common occurrences for most households. Consequently, a large, in-place panel would not be a rich source of data on reactions to credit acceptance or rejections. A panel would need to be designed such that the members are at the stage of the family life cycle in which they are most likely to seek credit.

However, the approach used in this study could be modified by a retail store or chain in a large sample study of its own credit applicants to obtain estimates of the opportunity cost of rejecting a good risk. The amount of spending for that type of outlet would need to be obtained and could be combined with the knowledge of the shifts in patronage (the difference between the proportion that would have been purchased had the application been accepted and the proportion purchased after a rejection) and with the store percentage contribution margin to yield the estimate.

The cross-sectional nature of this study limited the evaluation of the respondents' "credit riskiness." The problem of determining whether rejected applicants are really bad risks needs much more attention in further research and will require a longitudinal study to investigate the credit riskiness of the rejected applicants. The results of this study, showing marked shifts in patronage after the credit decision, provide incentive for further efforts to better measure the relationship between patronage decisions made by consumers and credit decisions made by retailers.

REFERENCES

Bonk, Roger S. (1975), "Bank Cards: Addition or Erosion?" Stores, 5 (November), 5, 31, 34.

Business Week (1979), "Why Big Stores Are Taking Outside Credit Cards," (September 3), 132, 134.

Caldwell, Charles W., Gentry, James W., and Miller, James J. (1980), "Estimating the Cost of Misclassifying a Good Credit Risk," Proceedings, American Institute for the Decision Sciences, 279.

Cole, Robert H. (1972), Consumer and Commercial Credit Management, Homewood, Illinois: Richard D. Irwin.

Cooper, Daniel (1977), "For Retailers, New Credit Problems and Opportunities," Management Controls, 24 (November-December), 25-29.

Dunkelberg, William C. (1974), "A New Look at Credit Customer Profitability," paper presented at the Credit Management Conference, National Retail Merchants Association, Los Angeles.

Greer, Carl C. (1967), "The Optimal Credit Acceptance Policy," Journal of Financial and Quantitative Analysis, 2 (March), 399-415.

Greer, Carl C. (1968), "Deciding to Accept or Reject a Marginal Retail Credit Applicant," Journal of Retailing, 42 (No. 4, Winter). 44-53.

Hendrickson, Robert A. (1972), The Cashless Society, New York: Dodd, Mead and Co.

Hirschman, Elizabeth C. (1979), "Differences in Consumer Purchase Behavior by Credit Card Payment System," Journal of Consumer Research, 6 (June), 58-66.

Kaplan, Robert Mo (1967), "Credit Risks and Opportunities," Harvard Business Review, 45 (March-April), 83-88.

Long, Michael S. (1975), "Optimal Selection of Applicants for Closed-and Open-Ended Credit," Working Paper, Georgia Institute of Technology.

Metha, Dileep (1968), "The Formulation of Credit Policy Models," Management Science, 15 (October). B30-B50.

Metha, Dileep (1970), "Optimal Credit Policy Selection: A Dynamic Approach," Journal of Financial and Quantitative Analysis 9 5 (December), 421-444.

Myers, James H. (1962), "Numerical Scoring Systems for Retail Credit Evaluation," Credit World, 51 (April), 5-7.

Pessemier, Edgar A. (1980), "The Store Credit Operation as a Profit Center," in The Changing Universe of Retail Credit, New York: Institute of Retail Management, New York University.

Stephenson, Pc Ronald, and Willett, Ronald P. (1969), "Analysis of Consumers' Retail Patronage Strategies," Proceedings, American Marketing Association Fall Conference. 316-322.

Wald, A. (1944), "On a Statistical Problem Arising in the Classification of an Individual into One of Two Groups," Annuals of Mathematical Statistics, 15, 145-162.

Wong, Jim (1969), "An Analysis of Retail Credit Systems Their Relationships, Significance and Implications for the Small Retailer," Proceedings, American Marketing Association Fall Conference, 106-109.

----------------------------------------