Coupon Proneness: Impact of Individual Difference ...

Coupon Proneness: Impact of Individual Difference Variables and Category Specific Variables

Kapil Bawa City University of New York Srinivasan Swaminathan Drexel University

Please address all correspondence to: Dr. Srinivasan Swaminathan Department of Marketing College of Business and Administration Drexel University 32nd and Chestnut Streets Philadelphia, PA 19104 Ph: (215) 895 2137 [email protected]

June 30, 2004


ABSTRACT In seeking to enhance the effectiveness of coupon promotions, researchers have long sought to identify "coupon prone" consumers. The measurement of coupon proneness in previous research, however, has tended to ignore the confounding effects of coupon attractiveness and the variations in coupon usage across categories. This study proposes a framework for measuring coupon proneness based on a model of coupon redemption that explicitly accounts for coupon attractiveness as a determinant of redemption behavior which is estimated at the category level. The empirical test of the model shows that the category-specific measure of coupon proneness can provide rich insight into how coupon usage varies across categories and yield better predictions of consumer response to coupons. A test of the model's predictive validity shows that it correctly predicts 89% of redemption intentions for the consumers in the sample. The authors find that propensity to redeem coupons in a category is related to category-specific factors such as brand loyalty and perceived coupon availability in the category, as well as to individual characteristics such as general coupon proneness, value consciousness and price consciousness. These results also show that propensity to redeem coupons is correlated across categories and that the degree of correlation depends on similarity between the categories. Specifically, the correlation within products and within services is higher than between products and services.


Coupons continue to be an important promotional tool in many consumer product categories, with 248 billion coupons distributed in the United States in 2002 with an average redemption rate of 1.5% (Promo 2003).

In the year 2002 consumers saved more than $ 3 billion, or $0.80

per coupon redeemed on average. (NCH marketing services, 2003). As the manufacturers and researchers seek to enhance the effectiveness of coupon promotions, they have focused considerable attention on the identification of "coupon prone" consumers and on the measurement of "coupon proneness" (e.g. Teel, Williams and Bearden 1980; Narasimhan 1984; Levedahl 1988; Bawa and Shoemaker 1987a; Lichtenstein, Netemeyer, and Burton 1990). While many important and useful insights about consumer response to coupons have emerged from these studies, two important concerns remain unaddressed in the literature on coupon proneness. First, previous studies have typically measured coupon proneness in terms of the individual's (observed or self-reported) coupon usage without taking coupon characteristics into account. An individual's coupon usage behavior depends both on his or her inherent coupon proneness, or desire to use coupons, but also on the attractiveness of the coupons encountered. For example, a consumer may be inclined to use coupons but exhibit low coupon usage if he or she fails to find coupons that are sufficiently attractive (i.e., coupons with high face values or for a preferred brand). Thus the failure to consider coupon attractiveness as a motivator of coupon usage might yield an inaccurate measure of coupon proneness. Second, although consumers' coupon redemption behavior varies substantially from one product category to another (Bawa and Shoemaker 1987a; NCH 1996a, 1996b), no attempt appears to have been made to study consumers' propensity to redeem coupons at the category 1

level and to examine its relationship with individual and category-specific factors such as coupon availability and brand loyalty. This makes it difficult for managers to predict how individuals will respond to a specific coupon for a specific type of product. Blattberg and Neslin (1990, p. 290) emphasize the need for managers to study coupon proneness at the product category level before formulating promotional strategy. The purpose of this study is to investigate how consumer's propensity to redeem coupons varies across categories. The paper seeks to contribute to the literature in three ways. First, it provides a method and basis for the measurement of coupon proneness at the product category level. The proposed method explicitly takes into account coupon attractiveness and the propensity of the individual to redeem coupons at the product category level. Second, it seeks to provide a better understanding of the extent to which category specific coupon proneness are correlated. Findings of our research reveal that the category specific coupon proneness are neither orthogonal nor perfectly correlated. Third, it seeks to understand how ‘category specific coupon proneness’ vary depending upon individual and category-related characteristics such as brand loyalty, usage rate, perception of coupon availability and price consciousness. Our research is particularly useful from a managerial standpoint because it allows managers to predict consumer response to different coupons and to identify categories in which joint couponing strategies can be most successfully utilized.

PRIOR RESEARCH Broadly speaking, two types of approaches appear to have been used in the past to measure coupon proneness. One approach, exemplified by Lichteinstein, Netemeyer and Burton (1990), measures coupon proneness using multi-item scales. A second approach uses behavioral measures of coupon proneness such as the observed percentage of purchases made with a

2

coupon, or coupon usage relative to that of the population (e.g., Narasimhan 1984, Bawa and Shoemaker 1987a). Lichtenstein et al. (1990) define coupon proneness as the increased propensity to respond to a purchase offer because the coupon form of the offer positively affects purchase evaluations. While reasonable, this definition overlooks the possibility that coupons may not have the same impact on a consumer across all product categories. For example, a consumer who is prone to using coupons in general may choose to disregard coupons for all but his favorite brand in a category where he is highly brand loyal. In a different category, where he is less loyal to any particular brand, he may switch brands to take advantage of a coupon offer. This suggests that if we are to understand and predict consumer response to specific coupons, we need to measure coupon proneness at the level of the product category. For ease of exposition we term the Lichtenstein et al. measure "general coupon proneness" hereafter to distinguish it from the category-specific measure proposed in this study. The multi-item approach proposed by Lichtenstein et al. is rooted in Classical Measurement Theory (CMT) and so is also subject to the limitations that characterize CMT-based measures in general: namely, item dependence and sample dependence (Hambleton and Swaminathan 1985). The alternative, behavior-based, approach used in past research to measure coupon proneness also has a potential difficulty in that it fails to take coupon characteristics into account. Past research has identified a number of coupon characteristics that have an effect on redemption behavior. For example, Reibstein and Traver’s (1982) analysis of 140 coupon drops found that coupon face value and distribution media were significantly related to redemption rates. Ward and Davis (1978) also found similar effects for face value and distribution media in their analysis of purchases of concentrated orange juice. Studies by Shoemaker and Tibrewala (1985) and Bawa and Shoemaker (1987b) showed that redemption intentions depend on coupon face value and the consumers' prior probability of purchasing the couponed brand. Research by Raghubir 3

and Corfman (1999) reveal that past promotional pattern and consumer’s expertise moderate the impact of sales promotions. A useful framework for understanding the determinants of coupon usage is one proposed by Bawa and Shoemaker (1987a). Their framework suggests that the net benefits of using a coupon have three components: economic and psychic benefits, substitution costs and effort costs. Economic and psychic benefits refer to face value savings and to non-financial benefits such as the smart-shopper phenomenon. Substitution costs are the disutility the consumer incurs by possibly purchasing brands he or she would not otherwise have bought, and effort costs are due to the consumers searching for coupons and handling them. To the extent that the coupon has a higher face value, requires less effort, and is for a preferred brand, the coupon is likely to be perceived as more attractive and hence more likely to be redeemed. This suggests that the measurement of coupon proneness must take into account the attractiveness of the coupons encountered. Failure to control for coupon attractiveness might result in misleading estimates of coupon proneness; for example, consumers who are not usually coupon prone will appear to be more so if they find an attractive coupon and decide to redeem it. Coupon attractiveness must therefore be accounted for in the estimation of coupon proneness. The measurement of coupon proneness must also consider variation across product categories. Bawa and Shoemaker (1987a) found that while households’ coupon usage across categories is correlated, there is substantial variation from one category to another. The major implication of their research is that households may differ in their propensity to redeem coupons across categories. Hence, we need to distinguish between the general coupon proneness of an individual and his or her propensity to redeem coupons in different categories. Blattberg and Neslin (1990) also note that category redemption rates vary widely and hence managers need to conduct research on category-specific propensity to redeem coupons in order to formulate effective promotional strategies. However, previous research appears largely to have ignored 4

cross-category variations in coupon proneness. In this paper we investigate how categoryspecific factors such as purchase frequency, perceived coupon availability and brand loyalty (Webster 1965; Bawa and Shoemaker 1987a) vary with category-specific coupon proneness.

A MODEL OF COUPON REDEMPTION Model Specification Following Bawa, Srinivasan and Srivastava (1997), we use the two-parameter model derived from Item Response Theory (Birnbaum 1968; Hambleton and Swaminathan 1985) to model the individual's intention to redeem a coupon. Redemption intention is modeled as a function of the individual’s category-specific propensity to redeem coupons and the attractiveness of the coupon, the latter varying with such factors as face value1 and effort required to redeem the coupon. In the model, category-specific propensity to redeem coupons and coupon attractiveness are treated as latent constructs and are estimated from redemption intentions. This is achieved by measuring the intentions to redeem different types of coupons for a sample of consumers. The resulting distributions of response probabilities for each type of coupon are then used to simultaneously estimate the attractiveness of the coupons and the individuals' category-specific propensities to redeem coupons. Formally, let Z j be the exponential of the ‘category specific propensity to redeem coupons’ of individual ‘j’ in a product category. Let Ei be the ‘attractiveness for coupon of type ‘i’. Let

G i be the inverse of E i . A customer’s intention to redeem coupon in a particular category is a function of both the individual’s coupon proneness in the category and the attractiveness of the coupon under consideration. Using the item response theoretic approach: Log (odds of intention to redeem a coupon) = Z j * E i = Z j .

Gi

1 As noted by Raghubir (1998) and Raghubir, Inman and Grande (2004), higher coupon face value also might signal higher product price under some conditions. The attractiveness of the coupon of a higher face value is the net effect of the positive evaluation due to increase in coupon face value and the negative evaluation due to perceived increase in price. We thank an anonymous reviewer for pointing this out.

5

Let θ j = ln( Z j ) and b i = ln( G i ). Therefore, intentions to redeem coupons =

odds 1 + odds

θ

(θ

−

eb = e b . (θ −b ) 1 + eθ eb 1+ e

= e

j

i

j

i

j

i

j

)

i

The parameter b i is inversely related to the coupon attractiveness and the parameter θ j is function of the coupon proneness of individual j, in that particular category. This model can be expanded to include an item discrimination parameter and the two parameter logistic model is given by: exp[a (θ − b )] i j i Pi (θ j ) = 1+ exp[a (θ − b )] ≡Pij i

j

i

(1)

Where a i and b i are model parameters relating to the unobserved attractiveness of coupon i. The model places individuals and coupons in the same, unobserved, continuum. Individuals are represented by θ j which can be interpreted as the propensity to redeem coupons. The larger the θ j, the higher the likelihood of redemption for given ai and bi. The coupon "position parameter" bi is inversely related to the unobserved coupon attractiveness of coupon i because a lower bi yields a higher redemption probability for a given θ j. Similarly, an individual with a larger value of θ j will have a higher redemption probability for a given level of coupon attractiveness bi. This ‘coupon characteristic curve’ is illustrated in Figure 1, which shows plots of the function in equation (1) for each of three coupon stimuli. Note that the characteristic curve for the $1.00 coupon is to the left of (and therefore represents higher attractiveness than) the characteristic curves for the 75 and 40 cent coupons. Hence, more attractive coupons--i.e., those with smaller values of bi--are likely to be redeemed even by individuals with low levels of coupon proneness. But less attractive coupons with high bi's will be redeemed only by highly coupon prone individuals (i.e., individuals with large latent θj 's). [Insert Figure 1 about here]

6

Note that the model only requires data on the dependent variable (redemption intentions or behavior) in order to estimate coupon proneness and coupon attractiveness. In other words, as long as a manager has household-level data on whether or not a set of coupons was redeemed, he or she can estimate coupon attractiveness and coupon proneness. This represents a distinct benefit compared to traditional survey-based methods of measuring coupon proneness based on multi-item scales (e.g. Lichtenstein et al. 1990). This Item Response Theoretic model has been used by marketing researchers in other contexts. Kamakura, Ramaswami and Srivastava (1991) have used it in evaluating households’ financial expertise and Balasubramaniam and Kamakura (1989) have used it to measure consumers’ attitude toward the market place. Role of redemption intentions In the past, researchers have investigated the impact of coupon characteristics on both redemption behavior (Reibstein & Traver, 1982) and redemption intentions (Shoemaker & Tibrewala, 1985). Investigating the impact of coupon characteristics on redemption rates is appealing as managers can use the findings to directly predict redemption rates and the associated costs of the couponing program. However, there are some compelling reasons to study the impact of coupon characteristics on redemption intentions. In the case of store causal variables like temporary price reductions, consumers are exposed to the deal inside the store and need to act immediately to take advantage of the deal offer. But in the case of coupons such as free standing inserts or direct mail, the customer is usually exposed to the coupons at home and cannot redeem the coupons immediately. Instead, the consumer first forms the intention to redeem the coupon (or not) and this intention drives subsequent redemption behavior (Bagozzi, Baumgartner and Yi 1992). In the absence of the intention, there is no coupon redemption. It is therefore important to study the impact of coupons on redemption intentions as a first step to understanding consumer response to coupons. Hypotheses 7

A number of hypotheses about category-specific propensity to redeem coupons are suggested by the literature on couponing. In formulating these hypotheses we draw on the cost-benefit model of coupon usage proposed by Bawa and Shoemaker (1987a). The model posits that the presence of fixed costs associated with coupon usage implies that if coupons are used at all, it is in the household’s interest to use them in several product classes where the benefits exceed the variable costs. This suggests that an individual with a high propensity to redeem coupons in one category will also have a high propensity to redeem coupons in other categories. We therefore hypothesize that: H1:

Category level propensities to redeem coupons are positively correlated.

A number of authors have observed that brand loyalty is negatively related to coupon redemption behavior (Mittal 1994, Bawa and Shoemaker 1987a, 1987b). Bawa and Shoemaker (1987b) suggest this is because consumers who are more brand loyal incur higher substitution costs (i.e., the costs of buying a nonpreferred brand) in order to use coupons. This implies that the substitution costs incurred by consumers will make coupons less attractive in categories where the consumers are more brand loyal. We expect, therefore, that propensity to redeem coupons in a category will be negatively correlated with brand loyalty in that category, leading to the hypothesis: H2:

The higher consumers’ brand loyalty in a category, the lower their propensity to redeem coupons in that category.

Irons, Little and Klein (1983) observe that the more opportunities consumers have to purchase in a category, the greater the potential for a large coupon effect. This suggests that heavier users of the category are likely to have a higher propensity to redeem coupons in the category because they have more opportunities for coupon usage and also stand to benefit more (in terms of total savings) from using coupons. Accordingly, we posit that:

8

H3:

The higher the consumers’ usage rate of the category, the higher their propensity to redeem coupons in that category.

Manufacturers in different product categories do not use couponing to the same extent, leading to wide variation in the availability of coupons from one category to the next (NCH 1996b) and consequently to differences in coupon redemption behavior (Bawa and Shoemaker 1987a). Past research indicates that consumers make inferences about the pricing and promotional patterns of manufacturers based upon their experience in the market place (Krishna 1991). Consumers' propensity to redeem coupons in a category, therefore, will be shaped by their perception of the availability of coupons in that category. For example, in categories where fewer coupons are distributed, consumers are less likely to actively look for coupons as opposed to categories where coupons are widely distributed. We therefore hypothesize that: H4:

The higher the availability of coupons in a category perceived by consumers, the greater their propensity to redeem coupons in that category.

Consumers who are highly coupon prone in general (as defined by Lichtenstein et al. 1990) are likely to be so in each category as well. This is because these consumers respond favorably to the coupon form per se regardless of value (Lichtenstein et al. 1990). Such behavior might occur, for example, if the consumer had low costs of coupon usage or obtained large psychological benefits (Bawa and Shoemaker 1987a). The implication is that for such consumers propensity to redeem coupons in different categories is likely to be positively related to their general coupon proneness. This suggests the following hypothesis: H5:

The higher the consumer's general coupon proneness, the greater his or her propensity to redeem coupons in different categories.

Several authors have argued that value consciousness is related to coupon proneness or usage. Lichtenstein, Netemeyer and Burton (1990) define value consciousness as a concern for paying lower prices subject to quality constraints and argue that value consciousness is driven by 9

acquisition utility (i.e., the utility of the purchased good less the purchase price) while coupon proneness is determined by transaction utility (defined as the internal reference price less the purchase price). Because consumers seek to maximize their total utility, value consciousness is positively related to coupon proneness. Zeithaml (1988) notes that segments of consumers define value in terms of “sales”, “specials” or “ability to use coupons” and many of these consumers use coupons as extrinsic signals of good value.

Assuming that value conscious consumers seek to maximize value or utility regardless of the product category, we hypothesize that: H6:

The higher the consumers’ value consciousness, the higher their propensity to redeem coupons in different product categories.

Unlike temporary price cuts which offer lower prices to all the consumers, coupons offer a reduction in purchase price only to those consumers who redeem coupons. Narasimhan (1984) argues that coupons are used to price discriminate between consumers and shows that consumers who redeem coupons are relatively more price sensitive than consumers who do not redeem coupons. Accordingly, we hypothesize that: H7:

The higher the consumers’ price consciousness, the higher their propensity to redeem coupons in different product categories.

METHODOLOGY Research Instrument An instrument was developed to measure each of the variables described in the preceding section. General coupon proneness and value consciousness were measured using the coupon proneness scale developed by Lichteinstein, Netemeyer, and Burton (1990). A number of other 10

constructs were measured including (a) the seven item value consciousness scale (Lichteinstein, Netemeyer, and Burton 1990), (b) the four item price consciousness scale (Wells and Tigert 1971), (c) the seven item brand switcher scale (Raju 1980), (d) category level brand loyalty and (e) category level perception of coupon availability. All the scales used exhibit high internal consistency: Cronbach’s α for the scale items ranged from 0.74 to 0.93. Finally, redemption intentions were measured for coupons in four categories using a dichotomous scale (will redeem/will not redeem): two grocery products, coffee and detergent, and two services, beauty salon and oil change for the car. Data on both products and services were collected to investigate the generalizability of our results.2 For each of the four categories analyzed, a number of coupon profiles were constructed by systematically varying the type of coupon, the coupon face value and preference for the couponed brand. The individual's intention to redeem each of these coupons was measured. The coupon profiles were based on three types of coupons (FSI, on-pack and mail-in), three face values (40 cents, 75 cents and $1 in the case of FSI and on-pack coupons, and 75 cents, $1 and $ 1.50 in the case of mail-in coupons)3, and two brands (most frequently purchased brand and occasionally purchased brand)4. Pretesting and Survey Administration A preliminary version of the instrument was administered to 12 doctoral students in business at a Southwestern university and the questionnaire was revised based on their input. It was subsequently administered to another group of 56 graduate and undergraduate students at the same university and the questionnaire was again modified based on the feedback obtained from

2 This study mainly uses existing scale items which are already reported in other published papers. References where these scale items are reported are given in the methodology section. Readers interested in getting a copy of the scale items used in the study can obtain the same by contacting the first author. 3 Pretests indicated that a 40 cent face value mail-in coupon was not realistic and was unlikely to generate any response given mailing costs. A $1.50 face value was therefore selected for testing instead. 4 For the two service categories, the profiles were constructed differently since the couponing environment is different for services. Pretests indicated that mail-in and on-pack coupons were inappropriate for service categories. Coupon profiles were thus based on the type of service provider (provider normally patronized and provider occasionally patronized) and face value of the coupon. For the beauty salon service, coupon face value was described in the form of a percentage off the regular price.

11

this group. Next, the questionnaire was pretested with 37 shoppers drawn at random from three grocery stores in a Southwestern city and further refined based upon their inputs. Five hundred questionnaires of this final version were distributed in two stores of a major grocery chain in the same city. Customers waiting in the checkout line were randomly approached and were requested to participate in the survey. Customers were given a copy of the survey instrument, a prepaid return envelope and a one dollar bill as a token of appreciation. Three hundred and fortyfive completed questionnaires were received, representing a response rate of 69%.

Model Estimation In this study, the model was estimated using the maximum likelihood method implemented in MULTILOG (Thissen 1991). The log likelihood of observing the responses (intention to redeem) is maximized with respect to the parameters a, b and θ. In order to eliminate indeterminacy in the model, the θ j parameters within each category are standardized with zero mean and unit standard deviation across the sample. The θ j parameter therefore represents the individual's coupon proneness in the category relative to other individuals in that category.5 For details of the estimation procedure please see Bawa, Srinivasan and Srivastava (1997).

RESULTS AND DISCUSSION Estimates of the propensity to redeem coupons ( θ j ) were obtained for each individual in each of four categories (coffee, detergent, beauty salon and oil change). We then examined correlations between the θj values across categories and between the θj's and general coupon

5 We chose to use a relative measure of coupon propensity (by standardizing θ across individuals) because the extent of couponing activity tends j to vary substantially across product categories, and by using a relative measure we are able to compare individuals' coupon propensity across categories. In addition, the use of a relative measure follows previous researchers (e.g. Bawa and Shoemaker 1987a).

12

proneness (as measured on the scale developed by Lichtenstein et al.1990) to test H1 and H5. These analyses are presented in Table 1. [Insert Table 1 about here]

The results in the table provide several interesting insights into the manner in which consumers’ propensity to redeem coupons in different categories are related to each other and to general coupon proneness. First, all the cross-category correlations between the θj values are positive (varying from a low of 0.167 to a high of 0.572) and significant. This provides support for H1 and implies that consumers who are relatively coupon prone in one category are likely to be so in other categories as well. Also all the correlations are significantly different from both zero and one. This indicates that category specific propensities to redeem coupons across different categories are neither orthogonal nor perfectly correlated. Second, cross-category correlations between the θj values vary substantially depending upon the nature of the categories. The correlation between propensities to redeem coupons for the two grocery products is relatively high (0.572), as is the correlation between the two services (0.439), whereas the correlations are much lower between grocery products and services. For example, the correlation between the coffee and beauty salon propensities is only 0.167 and this is significantly lower than both 0.439 (p