Do Shoppers Like Electronic Coupons?

Journal of Retailing 84 (3, 2008) 297–307

Do Shoppers Like Electronic Coupons? A Panel Data Analysis Song-Zan Chiou-Wei a,∗ , J. Jeffrey Inman b a

Department of Managerial Economics and Institute of Economics, Nan-Hua University, Chiayi, Taiwan b Katz Graduate School of Business, University of Pittsburgh, Pittsburgh, PA, United States

Abstract Past research has yielded valuable insights into the drivers of traditional coupon redemption, but the applicability of these results to electronic coupons remains an open question. We investigate the determinants of electronic coupon redemption, employing a large panel dataset for five product categories (detergent, milk, cookies, shampoo, and orange juice) for the period 2003–2005. Our findings reveal that education and employment positively influence redemption rates and our analysis indicates that these findings are not due to unobserved individual effects. The focus is on comparing coupon-use discrepancies between national and private label brands when the characteristics of coupons are taken into account. A higher face value appears to be a critical element in electronic coupon format, and this gives rise to more purchases for nonperishables (shampoo and detergent). Results also show significant seasonal variations in milk and orange-juice coupon usage. Furthermore, the distance of consumers from the redemption location has a significantly negative effect, whereas the expiration date has no evident effect. The implications for electronic coupon research and practice are discussed. © 2008 New York University. Published by Elsevier Inc. All rights reserved. Keywords: Electronic coupons; Coupon usage; Count panel data; The generalized method of moments

Introduction In this study we focus on the determinants of coupon usage by one of the fastest-growing types of promotion in Taiwan – couponing at the store or at home electronically through shopping cards. In Taiwan, fierce competition has resulted in retailers exerting a tremendous amount of effort in providing incentives to consumers. The electronically linked coupon distribution network is a byproduct of this competition. According to a national survey conducted by Retailing Services (2005), the use of this technology has risen from 8 percent in 2003 to 17 percent in 2004 throughout supermarkets in Taiwan. What distinguishes electronic from traditional couponing is that electronic couponing provides a far more direct and convenient vehicle than the traditional coupon modes of free-standing inserts, on-pack coupons, and mail-in coupons. Shoemaker and Tibrewala (1985) and Swaminathan and Bawa (2005) argue that the time lag between the consumer’s

∗

Corresponding author. E-mail addresses: [email protected] (S.-Z. Chiou-Wei), [email protected] (J.J. Inman).

exposure to coupons at home and redemption in store might attenuate redemption rates. They also argue that phenomena such as misplacing and forgetting to use coupons, or stock-outs of the couponed brand often result in fewer coupons used. However, those barriers can be overcome with the aid of an electronic distribution network that gives customers convenient ways to retrieve coupons. Further, some of the relationships identified in prior research may not hold for this new coupon distribution technology. For example, easier use may offset the typical expiration date effect (e.g., Inman and McAlister 1994). Therefore, researchers need to revisit behaviors surrounding coupon redemption. In the context of electronic couponing, consumers browse the Web at home for any suitable coupons available on the store coupon pages and print them out. In the case of forgetting and/or misplacing, they can also get access to the coupon pages by running their shopping cards through terminals at the store and printing out the coupons. Electronic couponing also allows for the development of a detailed coupon redemption database, which managers can use to better understand the nuances of coupons’ effectiveness. Access to such a database frees us from the shortcomings of self-report coupon redemption measures used in previous studies.

0022-4359/$ – see front matter © 2008 New York University. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jretai.2008.07.003

298

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

Previous research has employed cross-sectional approaches, yet has generally been unable to identify consumers’ shopping patterns during the promotional period. To illustrate this point, suppose consumers do not favor a certain product brand, but that they choose to purchase more of the sponsored product earlier than do brand-loyal consumers. A positive relationship between coupon use and demographic characteristics would then appear in the cross-sectional data even if individual consumers did not increase their coupon use over time (sample attrition bias). Researchers such as Neslin and Clarke (1987), Mägi (2003), Patterson and Smith (2003), Bhatnagar and Ghose (2004), and Meyer-Waarden (2007) acknowledge the use of cross-sectional data as a limitation in that behavior intentions do not always equate to actual behavior. In the current study, we address several other interesting issues in couponing. Our target supermarket chain offered store brands in five product categories, allowing comparison between national and store brands. Experiencing increasing success over the past decades, private label brands are normally perceived as lower price and lower quality products than their national brand counterparts, but they have pressured manufacturers to compete more vigorously on price in order to win back market share lost to private label brands (e.g., Cotterill and Putsis 2000; Garretsona, Fisherb, and Burton 2002; Gedenk and Neslin 1999; Sayman and Raju 2004). Another overlooked issue we address is the effect of product category, which may play an important role in response to coupon promotions (Cronovich, Daneshvary, and Schwer 1997; Swaminathan and Bawa 2005). Finally, we offer an alternative model to the traditionally used Poisson model. Coupon use usually occurs in discrete, non-negative quantities, yet traditional models do not accurately account for such behavior. The Poisson model assumes that the mean equals the variance, so if the mean–variance equality does not hold (known as overdispersion (Winkelmann 1997)), then the model is mis-specified. Further, there are some other intrinsic restrictions embedded in the basic Poisson model when handling a panel dataset, such as the failure to take into account that individual effects may be correlated with the regressors, the assumption of the strict exogeneity of the regressors, and serial correlation not being allowed in the residuals. To address these issues, we employ the generalized method of moments (Hansen 1982) in our estimation. This article is organized as follows. After a discussion of the coupon usage literature, we lay out several testable hypotheses for electronic coupons. We then discuss data collection and estimation methods, and provide empirical results. The final section offers conclusions and some managerial implications. Theoretical foundations and hypotheses Two streams of literature are related to behaviors surrounding coupon redemption. The first focuses on determinants of household coupon usage and examines socioeconomic, demographic, and psychological drivers of coupon/deal redemption (e.g., Bagozzi, Baumgartner, and Yi 1992; Green 1995; Hernandez 1988; Schofield 1994; Tat and Bejou 1994). The second stream emphasizes coupon characteristics and other non-demographic

characteristics (product category and market share), as well as coupon attractiveness (monetary value and expiration date) (e.g., Bawa, Srinivansan, and Srivastava 1997; Inman and McAlister 1994; Kumar and Swaminathan 2005; Lichtenstein, Richard, and Burton 1990; Mittal 1994; Neslin and Clarke 1987; Raghubir, 1998, 2004; Reibstein and Traver 1982). Most of the above-mentioned studies apply cost–benefit theory, which argues that households use coupons when the monetary savings exceed the time and cost spent in coupon related activities. Variables like family income, employment status, and the presence of young children are usually classified as constraining factors. Under this theory, increases in family income should increase the household’s cost of coupon use and induce the household to reduce its level of coupon-related activity. However, Lee and Brown (1985) found that high-income households were significantly more likely to use coupons, while Narasimhan (1984), Bawa and Shoemaker (1987) and Goodwin (1992) found a positive but statistically insignificant income effect. Employment status is regarded as a time constraint, reducing time for couponing activities (e.g., Mittal 1994). There is, however, little empirical evidence to substantiate this assumption. Cronovich, Daneshvary, and Schwer (1997) have shown the positive effects from the presence of young children, which is influenced by the contra indicators of tighter budgets (arguably leading to greater coupon use) and less time (arguably leading to lower coupon use). Generally, in the case of children within the home, budget appears to trump time and we see increased coupon usage (Lee and Brown 1985; Narasimhan 1984). Although demographic variables have been well covered in previous work, interactions among variables have been given short shrift. For example, Strober and Weinberg (1980) report that working women in high-income groups are less likely to use coupons than non-working women in the same demographic group. Product type may also complicate matters because most distributed coupons are rendered for families to use on items like dairy products, crackers, and soft drinks. Bawa and Shoemaker (1987) examine coupon use for different types of products and find a positive relationship between income and usage. It is likely that other interactions will arise from interactions that have yet to be studied. The context of electronic couponing may further complicate matters, diminishing time effects and reinforcing the income effects. The electronic setting provides a more convenient context for consumers, via either a home computer or point of purchase terminal, lifting the traditional time constraint. Lowerincome consumers spend a higher percentage of their total earnings on living expenses, leaving them with a smaller budget and less time to acquire extra products and services (Becker 1965; Michael and Becker 1973), which could reduce in-store coupon acquisition. Similarly, Internet acquired coupons require technology ownership and connectivity, as well as basic Web usage skills. These factors could create an income threshold for electronic couponing. Relatedly, education has emerged as a positive driver of usage in many coupon studies. Following the efficiency hypothesis (e.g., Levedahl 1988), better-educated consumers seek more

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

variety, have lower substitution costs, and thus should use more coupons. Better-educated shoppers are, on average, more intense coupon users (Bawa and Shoemaker 1987; Narasimhan 1984), although Goodwin (1992) find no significant effect. Overall, since factors such as family income, employment status, and education are indicative of household income, we formulate the following general hypothesis: H1. The greater a shopper’s income, the greater his/her propensity to redeem coupons. Other factors related to a potential threshold effect include age and education, which are related to adoption of new technology (Eilers 1989; Furlong 1989; Loges and Jung 2001). Young people will be more readily exposed to new technologies at school and through other early adoptors, so they should be more comfortable with using electronic coupons than those who are older and have relatively less access to new technologies. Thus, we propose that new technology-based coupon use may decrease above a certain age, yielding a non-linear (concave down) relationship between age and electronic coupon use. H2. There is a non-linear (concave down) relationship between coupon use and age. There are numerous studies incorporating intervening variables to explain the behavior of coupon redemption. Among them, brand loyalty has proven to be significantly and negatively correlated with coupon redemption behavior (Bawa and Shoemaker 1987; Mittal 1994). Nevertheless, in Swaminathan and Bawa (2005) study, the hypothesis is supported in such categories as coffee and beauty salons while not supported in detergent and oil changes. Lichtenstein, Richard, and Burton (1990) and Swaminathan and Bawa (2005) found coupon proneness to be an important antecedent of coupon redemption, while Mittal (1994) and Bawa and Shoemaker (1987) observed that store loyalty is negatively correlated with coupon redemption. Arguably, a shopper’s frequency of browsing either the online or store coupon page should serve as a good indicator and his/her coupon proneness. The rationale is from Bhatnagar and Ghose (2004), who find that the more frequently consumers access the Internet for information, the more likely it is that their purchases are influenced by the information obtained on the Web. Likewise, brand loyal customers can be categorized by their brand switching frequency over a certain period. Store loyalty can also be assessed by behavior measures such as frequency of shopping (Olsen 2002). Though not empirically validated, those who frequent a store arguably value that store and should be more likely to respond to store promotions. Thus, we propose a set of three hypotheses, regarding the advantages of electronic metering. H3. The more frequently a shopper browses the coupon page, the greater is his or her propensity to redeem coupons across product categories. H4. Shoppers who switch brands more frequently are more likely to redeem coupons across product categories. H5. Frequent shoppers are more likely to redeem more coupons than less frequent shoppers across product categories.

299

Higher face value coupons are often associated with higher redemption rates (Bawa and Shoemaker 1987; Reibstein and Traver 1982; Ward and Davis 1978). However, this effect has not been well studied across product categories and brand types (national and store), making it difficult for managers to decipher a consumer’s response to a specific type of coupon, as suggested by Cronovich, Daneshvary, and Schwer (1997) and Swaminathan and Bawa (2005). As mentioned earlier, private label brands are often regarded as having lower price and lower quality. Based on attribution theory (e.g., Sawyer and Dickson 1984), a lower price for private label brands may be attributed to some problematic aspect of the product, which is then perceived as being inferior in quality. In comparison, national brands, frequently using a nationwide advertising coverage to reaffirm their positive attributes and quality, are less likely to be considered inferior in quality (Garretsona, Fisherb, and Burton 2002). That is, they should be less susceptible to negative quality attributions from coupons. Inman and McAlister (1994) use regret theory to explain the relationship between expiration date and coupon redemption behavior. They argue that consumers’ anticipation of feelings of regret at having missed an expired coupon’s savings causes the immediacy of anticipatory regret to increase as the coupon’s expiration date draws closer. However, using regret theory without considering the face value effect provides a incomplete picture of the expiration date effect. We, therefore, posit the following hypotheses. H6. High face value coupons are more likely to induce increased coupon redemptions than low face value coupons across product categories. H7. (a) National brand products, in combination with high face value coupons, are more likely to induce increased redemptions in different product categories, and (b) shoppers are more sensitive to this combination at the day of expiration. The distance from the redemption location is an important convenience factor in explaining redemption behavior, but little literature addresses this. Greater redemption location distance from the consumers’ residence should concomitantly increase the cost in time (Chiou-Wei 2004; Rhee and Bell 2002). This effect should be more pronounced for consumers with greater time constraints, such as consumers with lower income and more young children, as discussed earlier. Thus, we formulate the following hypothesis: H8. A greater distance from the store will result in lower coupon redemption for families with lower income and more young children at home. Data description and estimation method Our data consist of more than 160,000 coupon redemption records for a two-year period obtained from a prominent supermarket chain in Taiwan between January 2003 and January 2005 and for which complete demographic profiles of members were available. A total of 24,104 card members were used in the subsequent analysis, 64.3 percent of whom were employed, with an

300

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

average age of 27. Almost half (46.2 percent) of the members had more than one child under ten years of age. Nearly twothirds held a college degree (66.7 percent), and almost that same percentage (65.4 percent) earned between NT$30,000–50,000 per month (approximately US$910–1,515). In order to test H7, we separate the data into two categories – private label brands and national brands. To this end, five product categories are employed: shampoo (three national brands and two private label brands), detergent (two national brands and one private label brand), cookies (five national brands and four private label brands), milk (four national brands and two private label brands), and orange juice (three national brands and two private label brands). In general, the private label brands have lower prices and market shares compared with the national brands. The sponsoring retailer issued coupons each month in one of four configurations: (1) 5 percent off with three-day expiration; (2) 5 percent off with seven-day expiration; (3) 10 percent off with three-day expiration; (4) 10 percent off with seven-day expiration. Dates for coupon issue and expiration were determined by the retailer with no carry over into consecutive months. Within the shampoo and detergent category, 64 percent of the coupons had a seven-day expiration, with the remaining expiring in three days. Within the food category, 46 percent of the coupons had a seven-day expiration period, with the remaining expiring in three days. Coupons were scanned at check-out and each coupon was redeemable only once for one product. Table 1 describes the coupon face value distribution and the average number of coupons used for each product category during the two-year observation period. Within all product categories, redemption rates experienced an increase or stable growth in the mean coupon use over time (these results are not presented in Table 1), with a spike in January and February for food products (as seen in months 13 and 14) due to New Years and Chinese Lunar New Years celebrations. Second, the redemption rates in the milk and orange juice categories have always been above that of the other categories and seasonal spikes are evident during the summer months (indicated by months 7–9 and 19–21, respectively). Third, high face value coupons and national brands generally captured more coupon use across product categories. Table 2 provides an overview of the percentage of the sample not using a coupon for each of the 24 months. Two results stand out. First, in relation to adoption (incidence), coupon penetration increased from 26 percent in month 1 to 48 percent by month 24, indicating success in executing the electronic couponing program. Second, the high proportion of non-users supports our use of count models in the analysis, discussed subsequently. Table 3 gives the definitions of variables. Given that the dependent variable (monthly number of coupons redeemed) in our analysis is a non-negative integer, count data models are the natural choice for our regression analysis. The standard Poisson model is frequently used for such count data. However, this does not allow for individual effects. That is, there may be unobserved specific effects that impact coupon use decisions unique to a given supermarket or the location of products. The failure to include individual specific effects may lead to overdispersion (Winkelmann 1997). The individual effect prob-

lem (i.e., unobserved heterogeneity) can be partially solved by including the unobserved individual effect in the Poisson parameter (Hausman and Griliches, 1984). Specifically, Hausman and Griliches (1984) suggest incorporating an overdispersion parameter (δ) to account for the form of heteroskedasticity where the conditional variance exceeds the conditional mean. As δ approaches infinity, the negative binominal distribution converges to a Poisson function. There have been many efforts in econometric research to circumvent the problem of correlated individual effects in the basic Poisson model. Among them, both the conditional maximum likelihood (CML) approach (Blundell et al. 1995; Hausman and Griliches, 1984; Montalvo and Yafeh, 1994), and the pseudomaximum likelihood (PML) method (Gourieroux, Monfort, and Trognon 1984; Wooldridge 1999), have proven useful in obtaining consistent estimates. However, there is a common restriction embedded within these approaches. They both rely on the strong assumption that the original distribution is a Poisson distribution and the explanatory variables are all strictly exogenous, which is often difficult to justify. To address the problem of violating the strict exogeneity assumption, some economists have argued that the explanatory variables should be considered predetermined rather than strictly exogenous. Among them, Hansen’s (1982) generalized method of moments (GMM) has several advantages over other methods. First, it does not need to be restricted by the equi-dispersion condition. Second, it allows for heteroscedasticity and serial correlation in the error terms. Finally, it relaxes the strict exogeneity assumption of the regressors. We therefore employed GMM in testing our hypotheses.1 In the next section we discuss our estimation results derived from the GMM method. Results This section reports the estimation results with respect to the hypotheses in Section ‘Theoretical foundations and hypotheses’. We first estimate the count model – applying the negative binominal and the Poisson distributions. The under-dispersion test (Cameron and Trivedi 1990) consistently rejects the hypothesis of equi-dispersion for our data at the 1 percent significance level. Since the Poisson model is not appropriate for our data, we only report the negative binomial results using the GMM method. In addition, a Hausman test value higher than the 5 percent critical value suggests that the fixed-effects models fit the data better. Table 4 reports the estimation results of the determinants of coupon use from our count panel data econometric model. In view of hypothesis H1, we approximate the effect of a shopper’s labor market income, using control variable indicators: whether the shopper is employed, years of education, and family income, which presents a positive correlation between a shopper’s income and coupon usage. The hypothesis is partially confirmed – we find a statistically significant increase in coupon usage by 2.6 percent for every year of schooling, and those who are employed have significantly more coupon redemption than 1

Details are available from the first author.

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

301

Table 1 Coupon face value distribution and average number of coupons used by product category (N = 24104)a Product (item price)

Months with 10% discount

Mean

Month with 5% discount

Mean

Shampoo (in bottle) National (1) NT240 (2) NT235 (3) NT218

1, 3, 6, 9, 12, 17, 18, 21, 23 1, 4, 7, 10, 16, 19, 22 2, 4, 6, 7, 9, 11, 14, 17, 21, 24

0.317 0.274 0.342

2, 4, 8, 11, 14, 16, 20, 22 3, 8, 11, 12, 17, 18, 21, 24 1, 5, 8, 13, 16, 18, 20, 22

0.217 0.195 0.219

3, 6, 9, 12, 15, 18, 21, 24 2, 4, 7, 8, 11, 14, 16, 17, 19, 22

0.276 0.241

1, 5, 7, 11, 13, 17, 19 1, 6, 9, 10, 13, 15, 18, 21, 24

0.141 0.127

1, 4, 7, 10, 13, 16, 19, 21 2, 3, 7, 8, 11, 12, 15, 17, 19, 22 1, 3, 7, 8, 14, 16, 18, 21, 24

0.281 0.261 0.274

2, 6, 8, 9, 11, 14, 17, 20, 22 1, 6, 9, 13, 16, 18, 20, 24 2, 6, 9, 11, 12, 17, 22, 23

0.204 0.301 0.214

1, 5, 8, 10, 16, 18, 20, 21, 23 2, 4, 7, 9, 15, 17, 19, 20, 22

0.198 0.168

2, 7, 9, 13, 17, 18, 19, 22, 24 1, 5, 6, 8, 10, 13, 15, 18, 21

0.167 0.118

Cookies (in box) National (1) NT118 (2) NT104 (3) NT100 (4) NT99 (5) NT85

1, 3, 6, 8, 12, 14, 17, 20, 21 2, 5, 7, 12, 13, 16, 15, 18, 21, 23 3, 6, 14, 18, 20, 22 1, 4, 7, 13, 15, 16, 19, 23, 24 2, 3, 7, 8, 13, 14, 17, 19, 22

0.298 0.314 0.291 0.248 0.198

2, 5, 7, 9, 11, 16, 19, 22, 24 1, 4, 6, 8, 11, 15, 19, 20 1, 2, 4, 8, 16, 23, 24 3, 8, 9, 10, 17, 18, 19, 22 1, 6, 9, 16, 18, 20, 24

0.189 0.176 0.241 0.141 0.167

Private (1) NT110 (2) NT092 (3) NT85 (4) NT85

1, 2, 5, 7, 8, 11, 16, 18, 20, 22, 24 2, 5, 8, 11, 17, 18, 19, 24 2, 4, 6, 8, 10, 12, 14, 16, 18, 20 1, 5, 9, 11, 15, 19, 21, 24

0.231 0.214 0.173 0.192

3, 4, 6, 9, 14, 15, 17, 21 1, 4, 8, 9, 10, 14, 16, 18, 19, 21 3, 5, 7, 9, 11, 13, 15, 21, 23 4, 8, 12, 13, 16, 18, 22

0.116 0.131 0.096 0.101

1, 4, 7, 9, 13, 17, 19 2, 4, 5, 8, 10, 14, 17, 19, 21, 24 3, 8, 9, 11, 12, 14, 16, 18, 24

0.446 0.431 0.374

2, 8, 11, 14, 19, 21, 24 1, 7, 9, 11, 13, 16, 22 1, 2, 4, 6, 7, 10, 13, 17, 21, 22

0.368 0.288 0.314

1, 2, 4, 7, 9, 10, 14, 16 19, 21, 22 1, 4, 6, 7, 8, 10, 13, 16, 18, 19

0.289 0.312

3, 6, 11, 15, 20, 24 2, 5, 8, 12, 14, 15, 20, 21, 24

0.251 0.216

1, 3, 6, 9, 13, 14, 17, 19, 21 3, 8, 10, 13, 16, 19, 22, 24 1, 4, 7, 11, 15, 18, 21

0.371 0.432 0.471

2, 4, 6, 8, 10, 12, 15, 18, 22, 23 1, 4, 8, 12, 17, 18, 23, 24 2, 5, 9, 13, 16, 23, 24

0.302 0.376 0.401

2, 7, 8, 9 13, 17, 19, 21, 24 1, 5, 9, 10, 16, 18, 20, 21

0.361 0.301

1, 3, 5, 10, 11, 15, 16, 20, 22 2, 4, 6, 8, 12, 14, 18, 22, 24

0.248 0.234

Private (1) NT198 (2) NT189 Detergent (in bottle) National (1) NT238 (2) NT235 (3) NT227 Private (1) NT205 (2) NT195

Milk (in bottle) National (1) NT112 (2) NT107 (3) NT102 Private (1) NT99 (2) NT95 Orange juice (in bottle) National (1) NT120 (2) NT115 (3) NT112 Private (1) NT108 (2) NT90 a

Bold numbers indicate the individual month mean is significantly different from the sample mean at the 5% level.

those who are not. However, the indicators of family income are not statistically significant. The coefficients of the age variables, AGE and AGE2 , have opposite signs and are both significant. This suggests a curvilinear relationship between coupon use and age. Interestingly,

this quadratic relationship is common to many studies on the diffusion of technology. However, the relationship is usually concave downward, indicating that the adoption rate diminishes with age. We speculate that the time and cost needed to surf the coupon pages may preclude participation somewhat for younger

Table 2 Percentage of the sample not using a coupon per month Month

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

Percentage (%)

74

68

63

66

62

67

64

59

58

61

62

54

56

58

57

60

55

59

62

60

59

56

55

52

302

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

Table 3 Definitions of variables Variable

Definition

COUPON USE

Monthly coupon usage measured at the end of each month

FAMILY INCOME

0 if income ≤ NT30,000; 0.25 if income > NT30,000 and income ≤ NT40000; 0.75 if income > NT40,000, and income ≤ NT 50,000; 1 if income > NT5,000

EDUCATION YOUNG CHILDREN (age 1–10) EMPLOYMENT E-MAIL ADDRESS AGE AGE2 DISTANCE

BRAND TYPE MONTH

Years of schooling 0 if size ≤ 1; 0.5 if size >1 and size ≤ 3; 1 if size>3 1 if employed at time t; 0 otherwise 1 if the member provides an e-mail address; 0 otherwise Age at time of current observation (×10−1 ) AGE squared Number of kilometers from the member’s mailing address to the nearest store 1 if the member redeems a different brand from previous one in a product category; 0 otherwise Monthly frequency of browsing coupon pages on line and in the store Accumulative frequency of shopping at the time of observation 1 if high face value coupons are redeemed; 0 otherwise 1 if low face value coupons are redeemed; 0 otherwise Dates to redeem the coupons. For example, coupon with seven-day expiration redeemed on the first day is coded 1 and 2 on the second day. 1 if a national brand is purchased on a monthly basis; 0 otherwise Month dummy variables

PRODUCT CATEGORY SHAMPOO DETERGENT COOKIES MILK ORANGE JUICE

(On a monthly basis) 1 if a purchase is made in the shampoo category; 0, otherwise 1 if a purchase is made in the detergent category; 0, otherwise 1 if a purchase is made in the cookie category; 0, otherwise 1 if a purchase is made in the milk category; 0, otherwise 1 if a purchase is made in the orange juice category; 0, otherwise

BRAND SWITCH BROWSE TRIPS VALUE (H) VALUE (L) ENDING DAYS

Interaction effects FAMILY INCOME × ENDING DAYS FAMILY INCOME × DISTANCE YOUNG CHILDREN × DISTANCE VALUE (H) × VALUE (L) × PRODUCT CATEGORY VALUE (H) × PRODUCT CATEGORY VALUE (L) × PRODUCT CATEGORY BROWSE × PRODUCT CATEGORY BRAND SWITCH × PRODUCT CATEGORY TRIPS × PRODUCT CATEGORY BRAND TYPE × VALUE (H) BRAND TYPE × VALUE (L) BRAND TYPE × VALUE (H) × ENDING DAYS

members and most of the members were already in a qualified position to use the couponing system, given the sample’s higher education level and employment status, although this merits further investigation.

Family income interacted with the remaining day(s) to expiration The distance to the nearest store interacted with his or her income level The distance to the nearest store for a member in his or her child size level 1, if both high and low value coupons are used in the shampoo category. Likewise for four other product categories 1, if only high face value coupons are used in the shampoo category. Likewise for four other product categories 1, if only low face value coupons are used in the shampoo category. Likewise for four other product categories Monthly number of browsing coupon pages interacted with different product categories PRODUCT CATEGORY interacted with BRAND SWITCH dummy variables Accumulative frequency of shopping interacted with different product categories 1, if national brand coupons with high face value are redeemed; 0 otherwise 1, if national brand coupons with low face value are redeemed; 0 otherwise National brand coupons with high face value coupon interacted with the remaining days to expire

Hypothesis H3 postulates a positive relationship between browsing frequency and redemption rates in different product categories. As can be seen in Table 4, the variable capturing the monthly frequency of coupon page browsing (BROWSE)

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

303

Table 4 The determinants of coupon use: GMM parameter estimatesa Variable

Coefficient

Standard error

p value

Mean

FAMILY INCOME EDUCATION YOUNG CHILDREN EMPLOYMENT E-MAIL ADDRESS AGE AGE2 DISTANCE BRAND SWITCH BROWSE TRIPS VALUE (H) VALUE (L) ENDING DAYS

0.082 0.261 −0.078 0.241 0.391 −0.187 0.236 −0.126 0.152 0.213 0.079 0.418 0.088 0.076

0.146 0.072 0.064 0.029 0.054 0.040 0.059 0.042 0.136 0.082 0.112 0.046 0.079 0.106

.276 .034 .104 .000 .000 .024 .036 .022 .123 .047 .167 .000 .104 .187

0.587 1.302 0.367 0.643 0.612 2.713 7.360 4.873 0.371 0.366 12.14 0.482 0.312 3.878

0.072 0.074 0.102 0.088 0.132

0.087 0.052 0.142 0.093 0.097

.118 .098 .125 .138 .105

0.386 0.537 0.314 0.218 0.246

0.204 0.178 0.146 0.132 −0.169 −1.146 0.573 0.472 0.276 0.231 2.616 1.814 0.188 0.173 0.312 0.322 0.298 0.124 0.105 0.174 0.201 0.182 1.367 1.176 2.167 2.962 3.014 0.512 0.254 2.287 0.583 0.436 0.336 0.384 0.297 0.113 0.128 0.058 0.124 0.168

0.058 0.052 0.045 0.064 0.033 0.814 0.068 0.076 0.047 0.067 1.462 1.107 0.063 0.075 0.091 0.076 0.082 0.141 0.314 0.048 0.052 0.082 1.178 1.964 1.498 1.986 2.436 0.121 0.176 1.479 0.144 0.167 0.081 0.056 0.041 0.087 0.076 0.126 0.091 0.097

.018 .025 .032 .047 .000 .113 .000 .000 .014 .039 .089 .093 .001 .041 .026 .009 .027 .446 .487 .036 .007 .046 .110 .342 .121 .104 .118 .001 .101 .089 .002 .043 .014 .000 .000 .103 .084 .460 .137 .079

0.222 0.231 0.198 0.213 4.260 3.674 0.143 0.128 0.106 0.117 2.872 3.167 0.176 0.184 0.233 0.207 0.211 0.146 0.124 0.366 0.286 0.319 10.026 11.238 14.362 12.146 11.762 0.317 0.206 2.246 0.082 0.071 0.053 0.061 0.051 0.044 0.037 0.032 0.039 0.054

PRODUCT CATEGORY SHAMPOO DETERGENT COOKIES MILK ORANGE JUICE MONTHb MONTH × PRODUCT CATEGORYb JULY × MILK AUGUST × MILK JULY × ORANGE JUICE AUGUST × ORANGE JUICE FAMILY INCOME × DISTANCE YOUNG CHILDREN × DISTANCE VALUE (H) × SHAMPOOb VALUE (H) × DETERGENT VALUE (L) × SHAMPOOb VALUE (L) × DETERGENT VALUE (H) × ENDING DAYS VALUE (L) × ENDING DAYS BROWSE × SHAMPOO BROWSE × DETERGENT BROWSE × COOKIES BROWSE × MILK BROWSE × ORANGE JUICE BRAND SWITCH × SHAMPOO BRAND SWITCH × DETERGENT BRAND SWITCH × COOKIES BRAND SWITCH × MILK BRAND SWITCH × ORANGE JUICE TRIPS × SHAMPOO TRIPS × DETERGENT TRIPS × COOKIES TRIPS × MILK TRIPS × ORANGE JUICE BRAND TYPE × VAULE (H) BRAND TYPE × VAULE (L) BRAND TYPE × VALUE (H) × ENDING DAYS BRAND TYPE × VALUE (H) × SHAMPOO BRAND TYPE × VALUE (H) × DETERGENT BRAND TYPE × VALUE (H) × COOKIES BRAND TYPE × VALUE (H) × MILK BRAND TYPE × VALUE (H) × ORANGE JUICE BRAND TYPE × VALUE (L) × SHAMPOO BRAND TYPE × VALUE (L) × DETERGENT BRAND TYPE × VALUE (L) × COOKIES BRAND TYPE × VALUE (L) × MILK BRAND TYPE × VALUE (L) × ORANGE JUICE a b

The dependent variable is the monthly coupon counts. For brevity, only significant estimates are shown. Other results are available upon request.

304

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

is significant, and the coefficients from the interaction terms with the product category are all statistically significant with signs in the expected direction, implying that, ceteris paribus, shoppers who more frequently browse the coupon pages tend to have higher redemption rates. Thus, H3 is supported. However, we find that product heterogeneity is reflected in the estimated product-specific elasticities. In the food product category of cookies, milk, and orange juice, the elasticity of coupon usage with respect to browsing frequency is in the vicinity of 0.31, larger than the corresponding elasticity for shampoo and detergent which are 0.188 and 0.173, respectively. This may be because food products are consumed more quickly than shampoo and detergent products, which induces more browsing of food products coupons. Hypothesis H4 was examined using a brand switching dummy variable2 (BRAND SWITCH) to capture idiosyncrasies in the ways members redeem coupons. We predicted that shoppers who switched brands more frequently, implying a higher price consciousness and lower brand loyalty, would be more likely to redeem coupons. While the BRAND SWITCH coefficients are not statistically significant, the coefficients of the interacting variables, BRAND SWITCH × PRODUCT CATEGORY, are significant for the categories of cookies, milk, and orange juice, suggesting that consumers are more loyal to shampoo and detergent than to the other three products. Whereas our methods differ, our results are similar in spirit to Swaminathan et al.’s (2005) findings that consumers who are more price conscious are likely to exhibit a higher category-specific coupon proneness, and that higher brand loyalty in a category is associated with a lower redemption rate in that category. Therefore, our results extend those of Swaminathan et al. to electronic coupons. However, hypothesis H5 is not supported. Although the signs are consistent with prediction, neither the main effect variable (TRIPs) nor the interacting effect variables (TRIPS × PRODUCT CATEGORY) are significant. Hypothesis H6 addresses the effect of coupon face value on coupon redemptions in different product categories. To test this hypothesis, the sample was classified into two groups based on coupon use per month. The first group contains shoppers who redeem high face value coupons (VALUE (H)), while the second group contains those who redeem low face value coupons (VALUE (L)). Hypothesis H6 implies that the sign on the coefficient (VALUE (H)) and the two-way interaction terms involving product category will be significantly positive, indicating greater rates of redemption in a specific category when coupon face value is high. Our results on this topic are intriguing. The statistically significant positive coefficients of VALUE (H), VALUE

2 An anonymous reviewer suggested a more continuous switching index, given the longer purchase history per member, to capture switching behavior as a more stable household trait. Indeed, that is more appropriate than our approach. However, we confronted difficulty in establishing an index for each member, due to the fact that many members were buying in different product categories, such as hardware, music and books, which may not apply to the product categories used here. In addition, many members only had a limited number of purchase records, which were not sufficient to create a complete index for them.

(H) × SHAMPOO and VALUE (H) × DETERGENT support the predicted effect. Even though VALUE (L) is not significant, the results from the statistically significant VALUE (L) × SHAMPOO and VALUE (L) × DETERGENT, combined with the first results, reveal an important insight into the effect of shelf price – coupons for higher priced products are likely to be more attractive (across all products, detergent and shampoo are more expensive categories). Another possible explanation for the results may be attributed to the types of products analyzed. Given that shampoo and detergent are nonperishable, we speculate that coupon prone consumers have a preference to redeem such coupons for the purpose of stockpiling. Before addressing H7a, the initial question to consider is whether the theorized interactions between brand type, coupon face value, and product category account for a significant amount of variance in redemption behavior beyond that accounted for by other variables. While three-way interactions are complex, they are implied by H7a. The theory is not simply that brand type affects redemption, but that the relationship is contingent on both coupon face value and the chosen product. A Wald test for a two-way interaction explains a significant amount of variance in redemption rate (χ2 = 456.1, p < .00) as it does for the three-way interactions (χ2 = 64.37, p < .00). In other words, the three-way interactions explain a significant amount of variance in redemption, in addition to that explained by two-way interactions. In terms of direction of the effect, H7a implies that the two-way interaction term, BRAND TYPE × VALUE (H), and three-way interaction, BRAND TYPE × VALUE (H) × PRODUCT CATEGORY, should be statistically significant. Beyond statistical significance of specific interactions, however, H7a requires an examination of whether the differences between coefficients are significant. Particular interaction terms tell us something about the direction and robustness of differences between national brands and private brands, but they do not reveal whether the effect of a high coupon face value differs across product categories from that of a low coupon face value. We next examined these relationships with a Wald test of the differences between coefficients. Specifically, we tested whether the effects of high and low face values were equal for each product category, when shoppers choose the national brand products. The results indicate they are not (in the shampoo category, χ2 = 21.67, p < .00; in the detergent category, χ2 = 17.42, p < .00; in the cookies category, χ2 = 16.34, p < .00; in the milk category, χ2 = 14.36, p < .00 and in the orange juice category, χ2 = 32.43, p < .00). Therefore, there is a reinforcing effect on shoppers of a high face value, but not on a low face value. This provides strong support for H7a. Hypothesis H7b refers to the potential expiration date effect on the redemption rate of national products with a high face value coupon. The coefficients of interest are the two-way interaction variables, VALUE (H) × ENDING DAYS, VALUE (L) × ENDING DAYS and the three-way interaction variables, BRAND TYPE × VALUE (H) × ENDING DAYS. If H7b is valid, then the magnitude of the interaction terms will be positive, indicating higher rates of redemption at the day of expiration. The results from Table 4, however, fail to provide

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

305

Table 5 Summary of hypotheses test results Hypothesis

Variable of interest

Hypothesized relationship with coupon redemption

Result

H1

Labor market Income

Positive

H2 H3 H4 H5 H6

Age Browse × product category Brand switch × product category Trips × product category Value × product category

Nonlinear, concave down Positive Positive Positive Positive

Supported for education and employment status Non-linear, concave up Supported Supported Not supported Supported in shampoo and detergent categories

H7

(a) Brand type × value (H) × product category (b) Brand type × value (H) × ending days

Positive

(a) Supported (b) Not supported

H8

(1) Family income × distance (2) Young children × distance

Negative

(1) Supported (2) Not supported

support for H7b. To further delineate the decaying effect, we arranged the redeemed coupons in order, starting from the first redemption date to the last, into seven segments for coupons expiring in seven days and three segments for coupons expiring in three days. We expected a monotonically increasing pattern of coefficients; however, we found that those coefficients are not significant at all levels and mixed in magnitude.3 The results together indicate that the ease of electronic shopping might offset the expiration date effect found for conventional coupons (Inman and McAlister 1994). Because the main purpose of H8 is to explore the effects of constraining factors on coupon usage, we are most interested in the coefficients of the interacting effects, FAMILY INCOME × DISTANCE and YOUNG CHILDREN × DISTANCE. If H8 is valid, the magnitude of the two-way interaction terms, indicated by the coefficients on the FAMILY INCOME × DISTANCE and YOUNG CHILDREN × DISTANCE, should exhibit a higher redemption rate for shoppers with higher income and fewer young children at home. The constraining distance effect holds true for family income status, indicating that a longer distance hinders lower income shoppers from using coupons while the effect is insignificant with the presence of children. Thus, H8 is partially supported. The results indicate that the time and effort required to purchase lower priced products may dampen higher income consumers’ intention to redeem coupons compared with paying higher prices. In addition, people may find it easier to buy a variety of daily products in Taiwan, as there is a high density of convenience stores. The effect is illustrated by the semi-elasticities of the estimated coefficients. In the case of the explanatory variables, an estimated coefficient β corresponds to an (exp(β) − 1) × 100 percent change in the dependent variable. Thus, a 1-km increase in distance is associated with a decrease in coupon usage of 12.6 percent. A seasonal dummy variable (MONTH) was also used to account for variations in coupon-use behavior. Insignif-

3 To save space, the results from different segments are not reported in Table 4, but are available upon request.

icant results are not consistent with the spikes in mean coupon usage exhibited in Table 1, which may result from the shorter observation data period used in this study. However, a significant summer effect is found in the milk and orange juice categories, which is consistent with higher beverage consumption during summer. Also noted, although not included in the main hypotheses, shoppers who provide an email address have a higher propensity to use more coupons (p < .00). While further investigation is required, we speculate that providing an e-mail address represents the shopper’s greater desire to receive electronic coupons or other promotional notifications.4 Conclusions and managerial implications A short answer to the question, “Do shoppers like electronic coupons?” is “Yes, they do!” In particular, those who are well educated and employed are the most receptive target market segment. Table 5 provides a summary of the results for each hypothesis. Several theoretical and managerial implications can be drawn from our results. Theoretically, this study lends support to our argument that measuring coupon proneness on the basis of observed coupon use is incomplete without considering the characteristics of the coupons. A higher face value is a more important element of coupon format design in the electronic marketing context, serving to stimulate more purchases among products such as shampoo and detergent. As expected, distance has a significant negative effect on the level of coupon use, but the expiration date and shopping frequency are not important determinants. In the electronic context, it appears that more information on consumer’s access to computers and connectivity is required to create a precise targeting strategy. In general, information like coupon page browsing frequency are influential, a finding which is in spirit similar to Menon and Kahn (2002). They found that the characteristics of products and websites that are encountered early in online browsing can significantly influence the level of arousal and pleasure that con4

We thank an anonymous reviewer for making this point.

306

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307

sumers experience, and thereby can influence their subsequent shopping behavior. The current results have several important implications for marketers. First, the concave upward non-linear relationship between coupon usage and age indicates that younger shoppers’ redemptions lag behind those of older shoppers and this difference increases with the shopper’s age. Our results also suggest that a clear presentation of the coupon web page and the inclusion of a greater variety of coupon choices in different product categories would induce more browsing (and greater redemptions). Moreover, costs permitting, a specific location and instructions for point of purchase terminal use could be beneficial. Second, grocery stores that plan to use electronic couponing should expect better results in densely populated metropolitan areas where travel distances are shorter. Third, brand managers for national brands can improve redemption rates by using high face value coupons with less concern regarding expiration date. Given such a strategy, it is important to remember that low priced brands can be at a competitive disadvantage with respect to high value couponing. Fourth, before conducting coupon activity, managers should understand the different nature of product categories. For example, shoppers are likely to redeem coupons on less loyal goods, like cookies and orange juice, which often have more substitutes. For beverage product managers, summer is a better time to allocate coupon dollars. Finally, managers can send coupon updates by e-mail to members so as to solicit more responses. While the above recommendations follow directly from our empirical results, caution should be exercised in interpretation. Coupon effects can interact with other promotional activities, such as advertising and annual sales. Further, this study is based on a two-year observation sample of Taiwanese shoppers and has a relatively limited range of product categories. Generalization will be increased as our study’s findings are replicated with samples from other populations and other product categories. We hope that our research will encourage further efforts along these lines. Acknowledgements The authors gratefully acknowledge the contributions of the editors and two anonymous reviewers. References Bagozzi, Richard P., Hans Baumgartner and Joujae Yi (1992), “State vs. Action Orientation and the Theory Of Reasoned Action: An Application To Coupon Usage,” Journal of Consumer Research, 18 (March), 505–1. Bawa, Kapil and Robert W. Shoemaker (1987), “The Coupon-Prone Consumer: Some Findings Based on Purchase Behavior Across Product Classes,” Journal of Marketing, 51 (October), 99–110. Bawa, Kapil, Srini S. Srinivasan and Rajendra Srivastava (1997), “Coupon Attractiveness and Coupon Proneness: A Framework for Modeling Coupon Redemption,” Journal of Marketing Research, 34, 517–25. Becker, Gary S. (1965), “A Theory of the Allocation of Time,” The Economic Journal, 75, 493–517. Bhatnagar, Amit and Sanjoy Ghose (2004), “Online Information Search Termination Patterns Across Product Categories and Consumer Demographics,” Journal of Retailing, 80, 221–8.

Blundell, Richard, Rachel Griffith and John Van Reenen (1995), “Dynamic Count Data Models of Technological Innovation,” Economic Journal, 105, 333–44. Chiou-Wei, Song-Zan (2004), “The Determinants of Direct Mail Coupon Usage Revisited: Evidence From Count Panel Data Models,” Agribusiness, 20, 189–200. Cameron, Colin A. and Pravin Trivedi (1990), “Regression-Based Tests for Overdispersion in the Poisson Model,” Journal of Econometrics, 46 (3), 347–64. Cotterill, Ronald W. and William Putsis (2000), “Market Share and Price Setting Behavior for Private Labels and National Brands,” Review of Industrial Organization, 17, 17–39. Cronovich, Ron, R. Daneshvary and Keith Schwer (1997), “The Determinants of Coupon Usage,” Applied Economics, 29, 1631–4. Eilers, Merry L. (1989), “Older Adults and Computer Education: Not to Have the World a Closed Door,” International Journal of Technology and Aging, 2 (2), 56–7. Furlong, Mary S. (1989), “An Electronic Community for Older Adults: The SeniroNet Network,” Journal of Communication, 39 (3), 145– 53. Garretsona, Judith A., Dan Fisherb and Scot Burton (2002), “Antecedents of Private Label Attitude and National Brand Promotion Attitude: Similarities and Differences,” Journal of Retailing, 78, 91–9. Gedenk, K. and S.A. Neslin (1999), “The Role of Retail Promotion in Determining Future Brand Loyalty: Its Effect on Purchase Event Feedback,” Journal of Retailing, 75 (4), 433–59. Goodwin, Barry K. (1992), “An Analysis of Factors Associated with Consumers’ use of Grocery Coupons,” Journal of Agricultural and Resource Economics, 17 (July), 110–2. Gourieroux, Christian, A. Monfort and A. Trognon (1984), “Pseudo Maximum Likelihood Methods: Applications to Poisson Models,” Econometrica, 44, 701–20. Green, Corliss L. (1995), “Media Exposure’s Impact on Perceived Availability and Redemption of Coupon by Ethnic Consumers,” Journal of Advertising Research, 35 (2), 56–64. Hansen, Lars P. (1982), “Large Sample Properties of Generalized Method of Moments Estimators,” Econometrica, 50 (4), 1029–54. Hausman, Jerry A. and Zvi Griliches (1984), “Econometric Models for Count Data With an Application to the Patent-RandD Relationship,” Econometrica, 52 (4), 909–38. Hernandez, Sigfredo A. (1988), “An Exploratory Study of Coupon Use in Puerto Rico: Cultural vs. Institutional Barriers to Coupon use,” Journal of Advertising Research, 28, 40–66. Inman, Jeffrey J. and L. McAlister (1994), “Do Coupon Expiration Dates Affect Consumer Behavior?,” Journal of Marketing Research, 31, 423–8. Kumar, V. and Srinivasan Swaminathan (2005), “The Different Faces of Coupon Elasticity,” Journal of Retailing, 81 (1), 1–13. Lee, Jonq-Ying and Mark G. Brown (1985), “Coupon Redemption and the Demand for Frozen Concentrated Orange Juice: A Switching Regression Analysis,” American Journal of Agricultural Economics, 67, 647–53. Levedahl, William J. (1988), “Coupon Redeemers: Are They Better Shoppers?,” Journal of Consumer Affairs, 22 (December), 264–82. Lichtenstein, Donald R., Richard G. Netemeyer and Scot Burton (1990), “Distinguishing Coupon Proneness From Value Consciousness: An Acquisition Transaction Utility Theory Perspective,” Journal of Marketing, 54 (July), 54–67. Loges, William E. and Joo-Young Jung (2001), “Exploring the Digital Divide: Internet Connectedness and Age,” Communication Research, 28 (4), 536–62. Mägi, Anne W. (2003), “Share of Wallet in Retailing: the Effects of Customer Satisfaction, Loyalty Cards and Shopper Characteristics,” Journal of Retailing, 79, 97–106. Michael, Robert T. and Gary S. Becker (1973), “On the New Theory of Consumer Behavior,” The Swedish Journal of Economics, 75 (December), 378–96. Menon, Satya and Barbara Kahn (2002), “Cross-Category Effects of Induced Arousal and Pleasure on the Internet Shopping Experience,” Journal of Retailing, 78, 31–40.

S.-Z. Chiou-Wei, J.J. Inman / Journal of Retailing 84 (3, 2008) 297–307 Meyer-Waarden, Lars (2007), “The Effects of Loyalty Programs on Customer Lifetime Duration and Share of Wallet,” Journal of Retailing, 83 (2), 223–36. Mittal, Banwari (1994), “An Integrated Framework for Relating Diverse Consumer Characteristics to Supermarket Coupon Redemption,” Journal of Marketing Research, 31 (November), 533–44. Montalvo, Jose G. and Yishay Yafeh (1994), “A Microeconometric Analysis of Technology Transfer: The Case of Licensing Agreements of Japanese Firms,” International Journal of Industrial Organization, 12, 227–44. Narasimhan, Chakravavarthi (1984), “A Price Discrimination Theory of Coupons,” Marketing Science, 3, 128–47. Neslin, Scot A. and Darral G. Clarke (1987), “Relating the Brand Use Profile of Coupon Redeemers to Brand and Coupon Characteristics,” Journal of Advertising Research, 27 (Feburary–March), 23–32. Olsen, Svein Ottar (2002), “Comparative Evaluation and the Relationship Between quality, satisfaction, and repurchase loyalty,” Journal of Academic Marketing Science, 30 (3), 240–9. Patterson, Paul G. and Tasman Smith (2003), “A Cross-Cultural Study of Switching Barriers and Propensity to Stay with Service Providers,” Journal of Retailing, 79, 107–20. Raghubir, Priya (1998), “Coupon Value: A Signal for Price?,” Journal of Marketing Research, 35 (August), 316–24. Raghubir, Priya (2004), “Coupons in Context: Discounting Prices or Decreasing Profits?,” Journal of Retailing, 80, 1–12. Reibstein, David J. and Phillis A. Traver (1982), “Factors Affecting Coupon Redemption Rates,” Journal of Marketing, 46, 102–13. Retailing Services (2005), The E-trend, Vol. XXI. Taiwan: Retailing Services. Rhee, Hongjai and David R. Bell (2002), “The Inter-Store Mobility of Supermarket Shoppers,” Journal of Retailing, 78, 225–37.

307

Sawyer, A.G. and P.R. Dickson (1984), “Psychological Perspectives on Consumer Response to Sales Promotion,” in Research on Sales Promotions: Collected Papers, Josz K. E. ed. Cambridge, MA: Marketing Science Institute, 1–21. Sayman, S.S. and J.S. Raju (2004), “How Category Characteristics Affect the Number Of Store Brands Offered by the Retailer: A Model and Empirical Analysis,” Journal of Retailing, 80, 279–87. Schofield, Albert (1994), “Alternative Reply Vehicles in Direct-Response Advertising,” Journal of Advertising Research, 34 (September–October), 28–34. Shoemaker, Robert W. and Vikas Tibrewala (1985), “Relating Coupon Redemption Rates to Past Purchasing of the Brand,” Journal of Advertising Research, 25 (October–November), 40–7. Strober, Myrah H. and Charles B. Weinberg (1980), “Strategies Used by Working and Nonworking Wives to Reduce Time Pressures,” Journal of Consumer Research, 6 (March), 338–4. Swaminathan, Srinivasan and Kapil Bawa (2005), “Category-Specific Coupon Proneness: The Impact of Individual Characteristics and Category-Specific Variables,” Journal of Retailing, 81 (3), 205–14. Tat, Peter K. and David Bejou (1994), “Examining Black Consumer Motives for Coupon Usage,” Journal of Advertising Research, 34 (March–April), 29–35. Ward, Ronald W. and James E. Davis (1978), “A Pooled Cross-Section Time Series Model of Coupon Promotions,” American Journal of Agricultural Economics, 60 (August), 393–401. Winkelmann, Rainer (1997), “Econometric Analysis of Count Data,” Heidelberg, New York: Springer. Wooldridge, Jeffery M. (1999), “Distribution Free Estimators of Some Nonlinear Panel Data Models,” Journal of Econometrics, 90, 77–9.