Page 1

International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

www.iji-cs.org

Examining Consumer Motives for Coupon Usage Huei-Chen Hsu1, Ya-Li Chung2, Hui-Chun Chan3 Department of Marketing Management, Department of Business Administration, Department of Business Administration, TransWorld University Yunlin, Taiwan maggie@twu.edu.tw; 2yali@twu.edu.tw; 3chwen@twu.edu.t

*1

Abstract The most general managerial implication of this paper is that coupon redemption varies according to different coupon plans. The study also compares the difference in coupon usage between two forms of coupon books. This paper approached the question on the effectiveness of DM coupon programs using longitudinal data to empirically investigate the effectiveness of direct mail coupon programs. The regression results render an opportunity to compare the study with other studies employing cross-sectional analysis. Evidence shows that the results also confirm the implications based on the regret theory that a good coupon face value offer leads to higher coupon usage. Keywords Regret Theory; Direct Mail; Coupon Usage

Introduction Coupons, always one of the more popular shopping aids, have become even more important in the second half of 2009 (http://www.shopperculture.com/shopper_culture/201 0/02/), indicating the economic effect of the recession on shopping behavior. Use of mobile phones, while still relatively limited, has nonetheless increased over the course of the year potentially due to the rapid development of smart-phone shopping and information-seeking applications. This is especially true with younger shoppers (18-24), 15% of whom has reported to use mobile phones as shopping aids versus 5% for those aged 35 and over. According to Nielsen (2009), a category of consumers which calls “coupon enthusiasts” accounts for the lion’s share of couponrelated purchases in America. There are two groups of consumers. The first are essentially price shoppers who compare prices; the other buyers may be less aggressive and less price sensitive but more brand’s loyal. The focus of this

article is on a common marketing tactic called “coupons”, which continues to be an important vehicle for manufacturers and retailers. Coupon redemption rates has reached 5.8 billion in 2009. A robust review on the literature has revealed the focus more limited to the discussion of coupon incentives which positively influence consumers’ attitude and behaviors toward consumer goods. Research has tended to be limited on inadequate attention to coupon effects on brand and choice model. Criticism of the mixed results has focus on insufficient explanatory variables in the proposed models and different sample size treatment. Mittal argued that consumers’ behavior to use coupon could not be explained only by demographics. The relative nondemographic effects, such as psychological characteristics and coupon attractiveness (for example, coupon with higher face values and expiration dates) are also important parts. Some consumer researchers use economic paradigm to understand the psychological aspects of price and price changes, especially as it concerns consumer reactions to a specific price for a particular brand. However, as Blattberg, Briesch, and Edward noted that, neither discipline nor frequent changing prices is very informative regarding consumer implications in more complex pricing contexts. Although many previous researches have made to identify the consumers’ perception on coupon use, although their studies are primary based on cross-sectional data, and rarely has research been directed specially at a longitudinal view of consumers’ involvement in the frequency of coupon use. The problem is that crosssectional data cannot identify a consumer’s “true” shopping behavior during the promotional period. Suppose consumers do not favor a certain product (not brand loyal), but purchase more of

41


www.iji-cs.org

International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

manufacturers-sponsored coupon products on a trial basis. The positive result appeared in the crosssectional data even if the consumer did not increase their coupon use over time. Any conclusion fails to be based on longitudinal data could therefore be misleading, and the analysis would be at least incomplete. The initial objective of this study is to ascertain the effects of coupons manifested in the across-time model. To capture the interaction between coupon use and promotions, the direct mail (DM) coupons which are defined as a package of price-discount tickets, sent through the mail is chosen. In this study two coupon booklets have been designed for this survey, in which the first coupon booklet has included all the brands (indicated as an “Random-Selected” form), and the second booklet has the brands that consumer requested sending each month(indicated as a “SelfRequest” form). The coupon booklet comparison, which has received less attention in recent coupon program studies, will help re-examine the existing major hypotheses related to coupon effects and provide managers with an effective instrument to predict how a given consumer will respond to a specific coupon. The goal is to show whether different coupon promotion can exercise discrepancy effect by means of the examination on the coupon use discrepancies between redeemers who request their own brands in the coupon book and those accept all of the brands offered and to reinvestigate the effects of coupons by using a longitudinal model to capture the interaction between coupon use and coupon promotions. The potential of coupon marketing is unlikely to be fully realized without being examined from a customer’s cross-time viewpoint. The approach taken in this paper is distinguished from existing studies in two directions. Firstly, most research has typically focused on socioeconomic and demographic determinants or consumer shopping patterns, making it difficult for a marketer to draw conclusions on the effectiveness of a coupon program (for example, Narasimhan; Lee and Brown; Bawa and Shoemaker; Kaufman and Hernandez; Goodwin; Cronovich et al.,). In light of marketing practice, it does not make much sense to talk about the “determinants” of a promotional activity. Rather, it would be more interesting to examine the responsiveness of the target consumers when they participate in the promotional activity. Secondly, it is important to note that a coupon program’s effectiveness predicts a longitudinal (as opposed to 42

cross-sectional) relationship between coupon use and the consumers’ involvement in the program. Literature Review Regret Theory Regret theory means that people anticipate regret if they make a wrong choice, and take this anticipation into consideration when making decisions. Fear of regret can play an important role in dissuading or motivating someone to do something. Regret theory is a model of choice under uncertainty. Lommes and Sugden generalized the mini-max regret approach. Choice is modeled as the minimum of a function of the regret vector, defined as the difference between the outcome yielded by a given choice and the best outcome that could have been achieved in that state of nature. Regret theory has been employed with some success to explain the results of experiments which yield violations of expected utility theory, and has also shown some predictive power in experiments designed to test for the existence of regret effects. From regret theory, it is proposed that a good coupon face value offer leads to higher coupon usage. Econometric Formulation This section has designed an empirical test of the participation effect on a direct mail coupon program, utilizing a fixed effect model in conjunction with unbalanced panel data to serve this purpose. A twostage estimation procedure was employed to identify the participation effect from the participant’s prior experience. The following section lays out a design for the empirical test of the coupons’ redemption effect. (1) Yit=α+βδi+γθt+λLn(PARTit)+ eit Yit: the monthly coupon usage in time period t for ith individual PARTit: the total number of times the ith individual participated in a coupon program as measured at the end of period t eit: the individual effect ( It is independently and identically as well as normally distributed disturbance term with mean zero and variance σe2) θt: a time vector of variables that have the same value during a given month, but changing over time δi: defined as a vector of individual-specific variables that remains unchanged during the course of the


International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

individual’s stay in the program δi might include variables such as gender, income, age, presence of young children, education, employment and marital status; and the dummy variables capture some of the individual’s ideological beliefs about the appropriate level of coupon use. Term θt included variables that might capture idiosyncrasies in the way that coupons are designed; variables such as coupon expiration dates and a coupon’s face value. The estimation of the equation(1)would violate the basis OLS assumption. If the disturbance term eit is composed of a fixed individual-specific component, such as: (The authors employed a Song Zan’s specification test for random versus fixed model. The hypothesis that the individual effects are uncorrected with repressors is rejected(χ2=9.176, P<0.01).) (2)eit= ui+εit To solve the above problem, the following fixed-effects model is proposed, (3)Yit=α+

T −1 ∑ WsDs s =1

N −1 ∑ θ kDk k =1

+λLn(PARTit)+εit

where T and N refer to the total number of time periods and individuals, respectively. The first set of variables contains time-specific dummy variables, defined as Ds=1 if s=t, and 0 otherwise. The second set of variables hosts individual-specific dummy variables, defined as Dk=1 if k=1, and 0 otherwise. One problem with equation(3)is its inclusion of a large number of individuals-specific dummy variables that would likely cause a severe multi-colinearity problem. According to Judge, Song Zan (2004), the following equation is transformed: ( 4 ) Yit - Y i =

Li 1 ∑ (1− )WsDs Li − Fi +1 s = Fi

+ λ(Ln(PARTit) -

Ln(PARTi))+(εit -εi) where Y i , Ln(PARTt), and εi are the sample averages of Yit, Ln(PARTit), and εit over the period of time the ith individual is observed in the program, while Fi and Li are defined as the first and last months of this period, respectively. It still was concerned about whether a prior coupon redemption experience has a long-lasting effect on current coupon usage. A consumer’s prior experience may serve as one type of informational input to the decision to act in addition to attitudes and subjective norm, and may act like a methodology, as a covariate

www.iji-cs.org

controlling for the effects of omitted variables. Ideally, we would test this hypothesis by using a variable measuring the prior experience with the participation variable(PART)in equation(4). It must be noted on the specification in ( 4 ) , the problem of unobserved heterogeneity that could arise in estimating models such as equation ( 1 ) can be avoided. It is reasonable to assume that those who stay longer in the program are observed to exhibit greater coupon use than those who leave the program earlier. Transforming the dependent variable into the deviation between a participant’s current coupon usage and the mean usage level controlling the deviations in coupon usage across participants. For interpretative purposes, equation ( 4 ) can be regarded as the change of regression in coupon usage on the change in the participation time(s). (The transformation of equation (4) does introduce heteroscedasticity into the error term because of correlation betweenεit andεi. Thus, we use White’s procedure to estimate the standard errors in order to correct problems due to heteroscedasticity.) The preceding discussion on prior experience also alerts us to the fact that the specification of equation (4) is unable to identify a participation effect from prior experience, since a heavy coupon user would likely continue his/her past coupon-use pattern regardless of what time the promotional program is introduced. That is, the participation effect would be mixed. To address this problem, we adopted the following two-stage estimation procedure described in Hsiao ( 1986 ) to solve the individual-specific use effects, Yi, where (5)Yi= Y i -

T 1 ∑ ( )WsDs i =1 Li − Fi +1

-λLn(PARTi)

Parameters Yi can be thought of as the individualspecific use effects that would have resulted from a direct estimation of equation(3). These individualspecific parameters- as predicted use variables are then used as the dependent variable in the following regression, ( 6 ) Yi = β0 + β1EXPIRATION DATESi + β2FACE VALUEi +β3MARITAL STATUSi +β4EMPLOYMENTi + β5INCOMEi + β6PRESENCE of YOUNG CHILDRENi + β7GENDERi + β8EDUCATIONi + β9Ln (AGE i)+β10Ln(PRIORUSE i) Where Expiration Dates and Face Value are binary dummy variables defined as the consumer’s overall satisfaction with the expiration dates and face value of

43


www.iji-cs.org

International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

the coupon book available in a specific month, respectively. Marital Status, Employment, Income, Presence of Young Children, Gender, and Education are individual-specific dummy variables, while Ln (AGE i)is the log of the participant’s age at the beginning of the program, and Ln(PRIORUSE i)is the log of the frequency of redemption during a pretest period. Taking the derivative of equation(6)with respect to PRIORUSE yields ∂Yi

(7) ( ∂PRIORUSE ) EQUATION (6)

=

β10 PRIORUSEi

Equation(7)estimates the effect of prior experience independent of the participation effect by using the two-stage procedure. 8

( ∂Yit (

∂PART

Pure

) EQUATION ( 4) − (

participation

effect

∂Yi ∂PRIORUSEi

) EQUATION (6)

The second variable moderating the relationship between participation in the coupon program and coupon use is the distance from the redemption location. It is supposed that the longer the redemption location distance is from the consumer’s residence, the more likely he/she would find it more costly and timeconsuming to redeem a coupon. Evidence, such as Tat and Bejou ( 1994 ) , has demonstrated that the perception of time and effort is negatively related to coupon use. However, for the same reason given above, the inclusion of a time-invariant variable in equation ( 4 ) is impossible. We thus divide the sample into five sub samples stratified by the distance from redemption location, which in turn are: (1) 0 ≤ DISTANCE ≤ 1 , (2) 2 ≤ DISTANCE ≤ 5 (3) 6 ≤ DISTANCE ≤ 10 , (4) 11 ≤ DISTANCE ≤ 15 , and (5) 16 ≤ DISTANCE . As the coupon redemption was only available in the designated downtown shopping center, we calculated the distance using the given current mailing addresses. The mediating effects of using pattern(PRIORUSE) and distance from the redemption location ( DISTANCE ) were also taken into account by separating the overall sample into stratified sub samples and comparing the estimated participation effects across sub samples. Methodology Table 1 reports the descriptions of summary statistics for the observation samples, with monthly data

44

collected through a self-administered survey on several major grocery chain stores in middle Taiwan between April 2009 and April 2011. In the first two data collection phrases, a total of 192 subjects were randomly selected in the checkout lines, who were requested to complete a preliminary survey and mail back all grocery shopping receipts for two months(all was offered was a small monetary incentive for their assistance). Of 250 subjects, 187 mailed their receipts back in pre-paid envelopes. In the second phase, the subjects were asked to participate in the manufacturers-sponsored promotion program, and of these, 166 ( N=166 ) agreed to join the two-year program(and hence, T=24). With the cooperation of manufacturers, there were three kinds of expiration dates offered for the coupon-one week, two weeks, and three weeks. The coupon book carries the same expiration date each month and each product also carries the same face value in a specific month. A letter of appreciation and introduction as well as a list of forty sponsored coupons such as food products, and household goods aids were sent to each participant. To test the hypothesis that customers who requested their own brands would be most likely to have a higher interest in coupons than those who accepted all the brands offered. The study follows design of Song Zan’s with two coupon books for the participants to choose. The first coupon book included the random selected brands ( indicated as an “Random-Selected” form), and the second book had the brands that the customer requested each month(indicated as a “SelfRequest” form). The coupon book was mailed at the beginning of each month and only useful at the designated grocery stores. Table 1 shows summary statistics for the data sets. A telephone survey was also conducted each month to solicit general opinions from participants about the coupon books and for information use. Redemption was measured by asking each participant to go through the coupon book, indicating how many coupons were used. The participants were free to withdraw from the program anytime over the phone. The final sample consisted of 64% subjects who are women with an average of 31 years, an average of 1.2 young children, and a mean income of NT$26,750 per month. Of the 166 participants, seventy-six(N=76) participants entered the All-Brand program, while ninety(N=90)chosed the Self-Request form.


International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

www.iji-cs.org

TABLE 1 SUMMARY STATISTICS FOR THE DATA SETS Variable COUPONUSE PART INCOME

EDUCATION

Definition Monthly coupon usage measured at the end of each month The total amount of time(s) that an ith individual participates in the promotion program Dummy variables taking the value 1 if (1)income≦NT$20000, 0 otherwise; (2)income>NT$20001 and income≦NT$30000, 0 otherwise; (3)income> NT$30001 Dummy variables taking the value 1 if (1)education≦high school, 0 otherwise; (2 )education≦college, 0 otherwise; (3 )education≦graduate, 0 otherwise (4)education>graduate, 0 otherwise

MARTIAL STATUS

Dummy variable taking the value 1 if married at time t and 0 if single

PRESENCE OF YOUNG CHILDREN

Dummy variables taking the value 1 if (1)size≦1, 0 otherwise; (2)1<size≦3, 0 otherwise; (1)size>3, 0 otherwise

EMPLOYMENT

Dummy variable taking the value 1 if employed at time t and 0 otherwise

GENDER

Dummy variable taking the value 1 if female at time t and 0 otherwise

FACE VALUE

Dummy variable taking the value 1 if the subject claims that the face value serves his/her needs and 0 otherwise Dummy variable taking the value 1 if the subject claims that expiration dates serve his/her needs and 0 otherwise Number of coupon usages prior to the actual promotion activities Number of Kilometers from the participant’s residence to the designate shopping grocery Dummy variable taking the value 1 if the participant requests a different combination of brands from previous solicitation and 0 otherwise Age at time of current observation

EXPIRATION DATES PRIORUSE DISTANCE BRAND-SWITCHING AGE

Statistical Results The important variable in equation ( 6 ) is Ln ( PRIORUSE ) , which measures the log of the individual’s coupon use behavior to the promotion program. The coefficient values estimate the effect of prior use on differences in coupon usage across participants. It was found here that the coefficients for both forms were significant, but opposite of that for the Ln(PART) coefficient. This means that the participation effects measured in the first-stage estimation there lacks an appropriate word ro connect the two noun phrases he “pure” effect of participation on coupon usage. If the longest-staying participants in a coupon program were more likely to use more coupons, then one would hypothesize and test a positive relationship at the aggregate level between coupon use and length of participant in the program. Table 2 reports the main results from our empirical analysis. The first-stage estimates consist of coefficient estimated for the fixed-effects model represented by equation ( 4 ) . The Self-Request form equation includes a Brand-Switching dummy variable, which attempted to identify any variation in coupon-use behavior due to changes in brands. Our major concern in fact was with the coefficient sign of the Ln

Sample Means (N=166) 7.93 17.95 (1)26.8% (2)34.8% (3)38.4% (1)0.11% (2)70.7% (3)22.1% (4)6.1% (1)42.4% (0)57.6% (1)43.4% (2)33.2% (3)23.4% (1)63.5% (0)36.5% (1)64.1% (0)35.9% (1)53.2% (0)46.8% (1)42.3% (0)57.7% 1.68 3.75 (1)38.2% (0)61.8% 31

( PART ) variable. The hypothesis predicted a positive coefficient for the participation variable and the first-stage estimates from Table 2 led support to this hypothesis. It is found that the coefficients for the Ln(PART) variable are indeed significant. In other words, an increased length of participation is associated with higher coupon usage. To be more specific, a 100 percent increase in participation length is predicted to increase coupon usage approximately 4.8 times for Self-Request form and 3.0 times for the RandomSelected form. These results were similar to the aggregate results indicated earlier. The insignificant Brand-Switching parameter in the Self-Request form suggests that brand-switching behavior has insignificant effects on coupon use. Even these first-stage estimates are able to identify the longitudinal (as opposed to cross-sectional) relationship between length of participation and coupon usage. However, the first-stage estimates cannot identify participation effects from prior use effects. While there is no empirical evidence indicating the influence of prior use on coupon usage, the importance of control over its influence is indeed recognized. The second-stage estimates reported in 45


International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

Table 2 comprise our basis for separating the effect of prior use from the effect of length of participation in the program. These results have been obtained using the calculated values for the individual-specific participation effects derived from the first-stage equations as individual observations of the dependent variable in a regression equation having a general specification for equation ( 6 ) . These individualspecific participation effects can be interpreted as the respective individual’s mean value for coupon use, minus the contribution of any included time-varying explanatory variables. Thus, the second-stage estimates provide a measure of the contribution of specific, timeinvariant variables on observed differences in the means coupon usage of individual participants.

A. Random-Selected form

All-Brand form 16 14 12 10

PART Effect

www.iji-cs.org

8 6 4 2 0

1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Time

6

TABLE 2 ESTIMATION OF THE EFFECT OF PARTICIPATION ON COUPON USE

First-Stage Estimates: Ln(PART) Brand-Switching Adjusted R2 Second-Stage Estimates: EXPIRATION DATES FACE VALUE MARITAL STATUS EMPLOYMENT PRESENCE OF YOUNG CHILDREN size≦1 1<size≦3 size>3 EDUCATION education≦high school education≦college education≦graduate education>graduate

RandomSelected

SelfRequest 16

0.30(2.01)* -----------0.20

0.48(3.65)* 0.12(1.07) 0.33

2.12(2.23)* 1.45(1.42) 1.41(0.84) 0.65(1.27)

2.52(2.62)* 2.42(3.69)* 0.85(1.01) 0.47(1.62)

0.37(1.26) 0.41(1.31) 0.58(1.47)

0.93(1.23) 1.12(1.33) 1.34(1.52)

14 12 10 8 6 4 2 0 1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Time

0.02(0.98) 1.98(2.32)* 1.75(2.11) 0.75(1.14)

0.01(0.71) 2.01(2.64) * 1.87(2.19) 0.84(1.25)

INCOME NT$0-20000 NT$20001-30000 NT$30001-40000 NT$40001 or higher GENDER Ln(AGE) Ln(PRIORUSE)

1.22(2.38)* -0.18(1.65) -0.38(1.82) -0.25(1.27) 0.58(1.58) 2.02(1.47) -1.13(3.17)*

Adjusted R2

0.572

1.52(2.97)* -0.27(1.56) -1.01(1.12) -0.58(1.37) 1.11(1.89) 1.85(1.97) 1.09(3.04)* 0.587

Observations

76

90

Note: Asymptotic t-statistics are reported in parentheses and calculated from White’s standard errors. Asterisks identify significance at 5 percent level (two-tailed test). “First-Stage Estimates” are selected coefficient estimates derived from the OLS estimation of equation(4)in the text. “Second-Stage Estimates” are derived from equation(6)in the text. Note that there is no constant term in the equation, so the usual interpretation of R2 is not valid.

46

Self-Request form

Part Effect

Variables

FIG. 1 ESTIMATED PART EFFECTS FOR THE RANDOMSELECTED FROM AND SELF-REQUEST FROM

Figure 1 graphically represents the estimated effect of participation using the coefficients from Table2. The “Gross” participation effect identifies the summation of the estimated participation and prior use effects on an individual’s coupon use over time. The “Net” effect is partial out the estimated prior use effect so as to arrive at a “pure” participation effect, as indicated in equation(8). The respective participation effects are calculated by assuming that a participant has the means for prior use value and coupon use. As indicated in Figure 1, there is a very significant difference between the Gross and Net effects, implying that our concern with separation of the effects of prior use and coupon use is warranted. Figure 1 also shows that participant time(s) has a practical impact on coupon usage by the participants.


International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

www.iji-cs.org

Focusing on the “pure” participation effect, it is found that 20 times of participation is associated with an approximated 6.2 times increase in Random-Selected coupon usage and a 8.1 times increase in Self-Request coupon usage. As a frame of reference, the sample’s standard deviation of the estimated individual-specific use effect---the Yi derived from the untransformed equation(3)is 1.43. Thus, our estimates of the effect of participation on coupon use suggest that participation length is a relatively important determinant of an individual’s coupon use behavior when compared to the factors generating differences in coupon use between participants.

as for greater DISTANCE values. To test this hypothesis, the samples were stratified into five different sub samplesor each of which equation(4) was estimated separately and the participation coefficients were then compared across sub samples. From Figure 2, the evidence is found to support the prior-use hypotheses, meaning that prior use monotonically affects the relationship between coupon use and participation, while distance shows no monotonically relationship. Figure 3 also is also applied to the same results which are in fact consistent with the earlier argument that a prior use effect is warranted.

A. Random-Selected form

The coefficients on the presence-of-children dummies are statistically insignificant, suggesting that there is no systematic relationship between the presence of children and coupon use. This finding is generally contradictory to Lee and Brown(1985)and Bawa and Shoemaker, but consistent with Cronovich, Daneshvary and Schwer. As argued by Cronovich, Daneshvary and Schwer, the positive effect from the presence of young children(operating on a budget constraint)could dominate the negative effect, which is exercised by the time constraint. Even with the insignificant estimates, it is actually found hat coupon usage increases with a greater presence of young children. The income effect also shows that there could be a budget constraint existing for lower-income participants.

FIG. 2 THE EFFECT OF PRIORUSE AND DISTANCE ON THE ESTIMATED PART B. Self-Request form

Fig. 3 The effect of PRIORUSE and DISTANCE on the estimated PART

The effects of prior use patters ( PRIORUSE ) and distance ( DISTANCE ) on the participation coefficient. As hypothesized above, the PART coefficient will be smaller for greater prior use, as well

Education attainment was shown to be a significant factor in determining coupon usage in the works of Narasimhan and Bawa and Shoemaker, whereas insignificance was found in Goodwin’s and Cronovich, Daneshvary and Schwer’s results. Our findings are similar to Bawa and Shoemaker’s, based on the panel data analysis. The possible explanation is found on Bawa, Srinivasan, and Srivastava’s assumption that better-educated consumers seek more variety, and have lower substitution costs, and thus making greater use of coupons. The possibility that less-educated participants left the program sooner, on average, compared to better-educated participants who were raised when the sample attrition bias was discussed earlier. In this case, better-educated participants would, on average, be more profligate coupon users than less-educated participants, which would generate a positive relationship between coupon use and education. Table

3

confirms

the

above

possibility

as

a

specification identical to equation ( 1 ) has been

47


www.iji-cs.org

International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

estimated. In equation ( 1 ) , there is no allowance made for the problems associated with the individualspecific fixed effects. Note here that most estimates are now significant for both forms of coupon programs. This shows strong evidence that previous studies using cross-sectional data captured something other than the longitudinal data. TABLE 3 ESTIMATION OF THE EFFECT OF PARTICIPATION ON COUPON USE WITHOUT CONTROLLING FOR INDIVIDUALSPECIFIC FIXED EFFECT Variables

Random-

Self-

Selected

Request

Constant

2.14(3.32)*

3.25(3.21)*

Ln(PART)

1.38(2.66)*

0.98(3.08)*

Brand-Switching

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

1.03(3.52)*

EXPIRATION DATES

3.58(3.24)*

3.69(2.85)*

FACE VALUE

3.44(4.67)*

2.83(4.61)*

MARITAL STATUS

1.90(2.13) *

1.91(2.96)*

0.58(1.85) *

0.23(2.17)*

YOUNG CHILDREN size≦1

1.21(1.89)

1.12(1.87)

1.42(2.65)

1.72(3.25)*

1<size≦3

1.89(2.78) *

1.92(3.55)*

1.12(2.01)

1.32(2.18)

EMPLOYMENT PRESENCE

OF

size>3 EDUCATION education≦high school education≦college

1.74(2.56)*

1.87(2.77)*

2.39(3.24)*

2.45(3.58)*

education≦graduate

2.77(3.85)*

2.87(3.95)*

NT$0-20000

0.12(2.36)*

0.38(2.99)*

NT$20001-30000

0.25(2.78)*

0.27(2.79)*

NT$30001 -40000

-1.26(1.32)

-1.03(1.01)

NT$40001 or higher

-1.15(1.25)

-1.24(1.32)

GENDER

0.78(1.94)

1.55(2.75)*

Ln(AGE)

3.87(3.15)*

3.24(3.02)*

Adjusted R2

0.478

0.431

Observations

76

90

education>graduate INCOME

Note: Coefficient estimates are derived from the OLS estimation of a model having the general specification of equation (1) in the text. Asymptotic t-statistics are reported in parentheses and calculated from White’s standard errors. Asterisks identify significance at 5 percent level (two-tailed test)

Conclusions and Suggestions The most general managerial implication of this paper is that coupon redemption varies according to different coupon plans. The study also compared the difference in coupon usage between two forms of coupon books. The regression results render an opportunity to compare the study with other studies employing cross-sectional analysis. The second-stage estimates in Table 2 contain the coupon use regression 48

results for coupon characteristics and socioeconomic as well as demographic variables. It was considered that the expiration dates and face values were positively correlated with coupon use. Mittal and others found a significant empirical relationship among these variables. However, the result found that a correlation between coupon use and expiration dates was significant only in the Self-Request form, while face value is significant for both programs. A possible explanation for the insignificant results in the Random-Selected form is that consumers who request their own brands are more likely to be bothered by allowing a coupon to expire unused than those who will accept any brand. The results also confirm the implications of the regret theory that a good coupon face value offer leads to higher coupon usage. At the aggregate level, the mean usage for the Random-Selected form was 6.83 times per month, while the usage was 9.03 times per month for the SelfRequest form users. The results generally imply that the Self-Request form is a better choice for managers, but caution should be paid to the interpretation of the empirical results due to the limited sample size. This paper approached the question on the effectiveness of DM coupon programs using longitudinal data. In general, the answer to this question is “yes” in the designated area. Two forms of coupon programs were activated simultaneously, showing the Self-Request form yield a better participation rate and greater coupon use than the Random-Selected coupon form. The roles of expiration dates and face value, even when manipulated in attitudinal terms rendered important insights into the design of a coupon promotion program. Consistent with most theoretical and empirical literature, both higher face values and longer expiration dates were significant determinants that maximized consumer response. Prior use, which has received less attention in recent studies, was found to be a critical indicator for the number of coupons used. An important implication of this finding from company standpoint is that managers can obtain information about a consumer’s prior coupon use beforehand in order to ensure that they send coupons to the appropriate target consumers. Despite substantial research along this line, little has been directed towards this issue related to the accumulated participation effect. Although this study shows that there is an increased significant participation effect on coupon use, previous literature, such as Song Zan, has suggested that coupon effectiveness research should be conducted in several


International Journal of Information and Computer Science Volume 2 Issue 4, May 2013

cities, because effectiveness differs from place to place. Future research, both longitudinal and cross-sectional, is needed to expand this issue across cities or compare the cultural discrepancies in coupon-use behavior among different countries. In addition, the tendency that antecedents and relationships of motives evolve as consumers progress from being aware of the e-coupon should also be included in further studies.It must be realized that technological advances currently enable personalized applications across a variety of interfaces.

Journal

of

Marketing

Research

31(1994) :533-44. Monroe, K.B. Pricing: Making Profitable Decisions, New York: McGraw-Hill,1990. Narasimhan,

C.

”A

Price

Discrimination

Theory

of

Coupons.” Marketing Science 3 (1984): 128-47. Nielsen, A., and Clarke S. “Relating the Brand Use Profile of

Coupon

Redeemers

to

Brand

and

Coupon

Characteristics.” Journal of Advertising Research 27 (1987): 23-32.

REFERENCES

Bawa, K. and Shoemaker R. “The Coupon-prone Consumer: Some Findings Based on Purchase Behavior Across Product Classes. “ Journal of Marketing 51 (1987): 99-110. Bawa, K. “Influences on Consumer Response to Direct Mail Coupons: An Integrative

Review.”

Psychology

Marketing 11(1) (1996): 77-95. Bawa, K., Srinivasan, S., and

Srivastava R.K. ”Coupon

Attractiveness and Coupon Proneness: A Framework for Modeling Coupon Redemption.” Journal of Marketing Research 26 (1997) : 517-525. Blattberg, R.C., Briesch B., and Edward J.F., “How Promotions Work.” Marketing Science 14(Summer) (1995) : 122-132. Cronovich, R., Daneshvary R., and Schwer K. ”The Determinants of Coupon Usage.” Applied Economics 29 (1997): 161-41. Goodwin, K.

Redemption.”

www.iji-cs.org

“An Analysis of Factors Associated with

Consumers’ Use of Grocery Coupons.” Journal of Agricultural and Resource Economics 17(1992): 110-120. Hsiao, C. Analysis of Panel Data, Cambridge: Cambridge University Press,1986. Kaufman, J., and Hernandez A. ”Barrier to Coupon Use: A View from Bodega.” Journal of Advertising Research 30 (1990): 18-25. Lee, L., and Brown G. “Coupon Redemption and the Demand for Frozen Concentrated Orange Juice: A Switching Regression Analysis.” American Journal of

Nielsen

Clearing

House

DG.

Worldwide

Coupon

Distribution and Redemption Trends

2010,

Lincolnshire (IL): NCH Marketing Services, 2010. Nielsen

(2009)

2010/09/20

http://gesterling.wordpress.com/2009/09/11/nielsen-19drive-81-of-coupon-usage/ Song Zan, C.W. “The Effectiveness of Coupon Booklets: When

for

Self-requested

Brands.”

Sun

Yat-Sen

Management Review 12(5) (2004): 19-44. Tat, K., and Bejou D. “Examining Black Consumer Motives for Coupon Usage.” Journal of Advertising Research 34 (1994): 29-35. http://www.shopperculture.com/shopper_culture/2010/02/ coupon-usage-continues-to-increase.html Huei-Chen Hsu received PhD degree in marketing management from National Yunlin University, Yunlin, Taiwan, in 2008. Since 2009, she has been an associate professor in the Department of Marketing Management and as a director of Center for Computer & Information Service in 2010 at TransWorld University where she instructs communication networks. Her research activity focuses on consumer behavior and network communication. Ya-Li Chung is a lecturer in the Department of Business Administration of TransWorld University, Taiwan. Her research interest is in the area of Business Administration. Hui-Chun Chan is a lecturer in the Department of Business Administration of TransWorld University, Taiwan. Her research interest is in the area of Information Management in computer.

Agricultural Economics 67 (1985): 647-653. Loomes, G., Sugden R. “‘Regret Theory: An Alternative Theory of Rational Choice under Uncertainty.” Economic Journal 92(4) (1982): 805–24. Mittal, B. ”An Integrated Framework for Relating Diverse Consumer

Characteristics to Supermarket Coupon

49

Examining Consumer Motives for Coupon Usage  

http://www.iji-cs.org The most general managerial implication of this paper is that coupon redemption varies according to different coupon p...

Read more
Read more
Similar to
Popular now
Just for you