Product recalls and the moderating role of brand commitment

Product recalls and the moderating role of brand commitment

Frank Germann & Rajdeep Grewal & William T. Ross Jr. & Rajendra K. Srivastava

Published online: 6 July 2013 # Springer Science+Business Media New York 2013

Abstract We assess attenuating and augmenting effects of brand commitment on consumer responses when product recalls occur. Consistent with our theorization, results from a laboratory experiment and an event study show that high levels of brand commitment attenuate negative consumer responses in low-severity product recalls but augment them in high-severity product recalls. Thus, while brand com- mitment seems to provide a reservoir of goodwill in the former case, it acts as a liability in the latter. These findings add to the extant brand and product recall literature by demonstrating that brand commitment has a complex effect on consumer responses when product recalls occur. Because product recalls are widespread, these findings also have managerial relevance.

Keywords Brand commitment . Product recall . Corporate crises . Experiment .

Event study

Mark Lett (2014) 25:179–191 DOI 10.1007/s11002-013-9250-5

F. Germann (*) 395 Mendoza College of Business, University of Notre Dame, Notre Dame, IN 46556, USA e-mail: fgermann@nd.edu

R. Grewal Smeal College of Business, Pennsylvania State University, 407 Business Building, University Park, PA 16802, USA e-mail: rgrewal@psu.edu

W. T. Ross Jr. School of Business, University of Connecticut, 2100 Hillside Road Unit 1041, Storrs, CT 06269, USA e-mail: bill.ross@business.uconn.edu

R. K. Srivastava Lee Kong Chian School of Business, Singapore Management University, 50 Stamford Road, Singapore 178899, Singapore e-mail: rajs@smu.edu.sg

1 Introduction

Product recalls are ubiquitous and familiar, including toys with toxic paint, seafood contaminated with dangerous antibiotics, and cars that accelerate seem- ingly on their own (e.g., Cleeren et al. 2013). In 2012 alone, the US Consumer Products Safety Commission announced more than 250 product recalls, and the Food and Drug Administration reported on over 300 such events. These numbers are sobering and suggest that no manufacturing firm is immune to product recalls.

The prevalence and potential harmfulness of product recalls has prompted several streams of research in Marketing, including investigations of their performance consequences (e.g., van Heerde et al. 2007), comparisons of proactive, firm- initiated recall strategies with reactive, passive approaches (Chen et al. 2009), and examinations of how consumer-level differences affect recall perceptions and re- sponses (e.g., Cleeren et al. 2008).

Most research indicates that consumers respond negatively to product recalls (e.g., Lei et al. 2012), and one research stream focuses specifically on how industry-, firm-, and/or consumer-related factors influence consumer responses. For example, in studying product recalls in the U.S. automobile industry, Rhee and Haunschild (2006) examine the moderating role of firm quality reputation on future product sales, Klein and Dawar (2004) note how corporate social responsibility affects consumers’ attributions in product harm crises, Lei et al. (2012) examine how the frequency of such crises in an industry influences consumers’ attributions for failure, and Ahluwalia et al. (2000) examine brand commitment’s role when brands receive negative publicity and find that brand commitment attenuates consumers’ responses to negative information.

In this research, we examine how brand commitment moderates negative infor- mation effects of product recall announcements. Specifically, building on Ahluwalia et al. (2000), we investigate whether brand commitment provides a “reservoir of goodwill” (Jones et al. 2000) for the recalling firm by attenuating the generally negative consumer responses to product recalls.

In addition, we reason that brand commitment might also augment the negative information effects of a product recall. For example, highly committed consumers may experience a feeling of betrayal and hence be especially disappointed when the products they feel close to get recalled. Thus, departing from Ahluwalia et al. (2000) and adding to the extant literature, we also examine whether brand commitment might in fact serve as a liability for the recalling firm.

To unravel these seemingly contradictory effects of brand commitment (i.e., reservoir of goodwill or liability), we recognize that product recalls are not homoge- neous events and suggest that product recall severity—defined as the consumer-based tribulations (or potential thereof) caused by the recalled product—is an important distinguishing factor of product recalls. To understand the role of brand commitment, we examine how product recall severity interacts with brand commitment during a product recall event. We posit that brand commitment can have both attenuating and augmenting effects on negative consumer responses to recalls, and we predict that, while brand commitment should provide a reservoir of goodwill in low severity recalls, it likely becomes a liability in high severity recalls.

180 Mark Lett (2014) 25:179–191

We use a laboratory experiment and an event study to test our ideas. Confirming the theorized effect, our results from the experiment indicate that high levels of brand commitment attenuate negative consumer responses in low severity recalls but augment them in high severity recalls. Moreover, the results from the event study suggest that the stock market responds in a manner consistent with the expected consumer responses to product recalls.

2 Conceptual background and hypotheses

In this section, we describe the conceptual background of the research. We begin by describing our two focal constructs, brand commitment, and product recall severity and then present our hypotheses regarding brand commitment’s dual role in product recalls.

2.1 Brand commitment

Consumers can become attached to brands, form close relationships with them (e.g., Fournier 1998), and have a general desire to maintain this close relationship (e.g., Beatty et al. 1988). In line with extant research (e.g., Ahluwalia et al. 2000), we define a consumer as committed to a brand if s/he displays these characteristics.

2.2 Product recall severity

Product recalls are not homogeneous but rather vary in their severity, among other factors. Consistent with Cheah et al. (2007), we define recall severity as the consumer-based tribulations (or potential thereof) caused by the recalled product. For example, considering past product recalls, some products had caused serious health problems (e.g., Toyota’s recall at the end of 2009 after reports that vehicles experienced unintended acceleration; the error has been linked to over 20 deaths and many severe injuries); others were responsible for only minor injuries or caused no harm at all (e.g., Chrysler’s recall in 2004 because of a wiring issue; no injuries or deaths were linked to the error). The Toyota recall can be classified as a high- and the Chrysler recall as a low-severity recall.

An important dimension of recall severity is the recall’s perceived ambiguity. In particular, the level of recall severity correlates with the perceived ambiguity of the recall event (e.g., Cheah et al. 2007), such that a non-severe recall tends to appear ambiguous whereas a severe one does not. Consumers may not be able to determine the extent to which the firm has committed a transgression when recall severity is low. In fact, the recall even may suggest that the recalling firm is acting responsibly and putting customers’ needs first. In contrast, high-severity recalls do not seem ambig- uous, because, at least, some consumers experience significant tribulations in these cases, including injuries or even death.

2.3 Brand commitment’s role when product recalls occur

Most research indicates that consumers respond negatively to product recalls (e.g., Lei et al. 2012). However, Ahluwalia et al. (2000) show that committed consumers

Mark Lett (2014) 25:179–191 181

exhibit greater resistance to negative information about well-liked brands. They also show that committed consumers engage in biased processing of negative information by counterarguing the negative information, which, in turn, atten- uates negative consumer responses (i.e., negative attitude change) following the learning of the negative information. In addition, Ahluwalia et al. (2000) show that when commitment to a brand is lower, consumers process the negative information more objectively. Consequently, counterarguing is less and negative attitude change is more prevalent among less committed consumers. This finding is particularly relevant for our study because it suggests that brand commitment attenuates the generally negative consumer responses to product recall announcements.

Yet, research has also identified mechanisms by which high brand commitment may augment the negative consumer responses to a product recall announcement. Specifically, building on the well-established expectancy–disconfirmation effect (Oliver 1993), committed consumers may come to expect more from the brand they like and thus feel especially disappointed when the brand gets recalled. Indeed, committed consumers might view a product recall as a “breach of contract,” and hence might exhibit more negative responses following a recall announcement than their less committed counterparts.

The preceding discussion outlines two conceivable logics for predicting how brand commitment affects consumer responses to product recalls. Given the conflicting nature of these logics, the question arises whether a moderator might help disentangle the contradictory perspectives. We posit such a moderator in recall severity, and we outline our reasoning below.

As noted above, perceived ambiguity is an important dimension of recall severity such that a low-severity recall tends to appear ambiguous whereas a high-severity recall does not. Building on the ambiguity dimension, we posit that highly committed consumers should tend to counterargue the negative information of a low-severity recall and thus discount the recall as an aberrant one-off event unlikely to reoccur. In contrast, lacking the strong positive associations, less committed consumers should handle the negative information more objectively and counterargue it less. Accordingly, and consistent with Ahluwalia et al. (2000), we expect that highly committed consumers should experience less negative attitude change than less committed consumers in low- severity recalls.

However, for a high-severity recall, we posit that the brand’s unambiguous negative performance impedes counterarguing and instead refutes the generally high expectations of the committed consumers which, in turn, should provoke disconfirmation effects. This disconfirmation effect might feel personal for committed consumers, like a feeling of betrayal, and we thus expect them to express thoughts suggesting that the recall is not at all consistent with their expectations (referred to as incongruity thoughts in the following). In contrast, without strong positive associations, less committed consumers should lack expectations about the brand’s performance, so the disconfirmation effect should be weaker for them. Accordingly, we also expect less committed consumers to express significantly fewer incongruity thoughts. As a result, we expect that highly committed consumers should experience more negative

182 Mark Lett (2014) 25:179–191

attitude change than less committed consumers in high-severity recalls. Thus, we propose:

H1: Brand commitment attenuates negative consumer responses in low severity product recalls but augments negative consumer responses in high severity product recalls.

H2: The effect of brand commitment on negative consumer responses is medi- ated by counterarguments in low severity product recalls and by incongruity thoughts in high severity product recalls.

In what follows, we first test our hypotheses in a laboratory setting (i.e., test H1 and H2) and then using an event study (i.e., (re-)test H1).

3 Laboratory experiment

A total of 133 students from a U.S. university participated for extra credit. The participants were randomly assigned to one of two conditions (recall severity: high versus low).

3.1 Procedure

When they arrived for the experiment, participants were told that they would be participating in a media study, in which they would evaluate “breaking news articles” published online on the morning of the experiment by a well-known news provider (CNN.com). We also indicated that the news articles contained information about brands and that we would ask them to evaluate these brands. We administered the experiment using Qualtrics, and the experiment included three contiguous sections. In the first section, participants were asked to evaluate the brands prior to learning more about them in the news articles. We also measured participants’ commitment to the brands in the news articles in this section using the three-item brand commitment measure developed by Beatty et al. (1988).1

In the second section, we provided three news articles for participants to read. To control for position effects, the focal, fictional article about a product recall always appeared second. The remaining two articles that served as filler were based on real articles and served to reduce the likelihood of ceiling effects due to excessive attention focused on the target message (Ahluwalia et al. 2000). After reading each of the three articles, the participants were instructed to take 2 min to list all their thoughts while reading the article. As we describe in more detail, we used these thought protocols for our process test.

Finally, in the third section, participants were asked again to evaluate the brands mentioned in the previous news articles. We debriefed participants after they finished

1 The three items are (1) “If brand X were not available at a store, it would make little difference to me if I had to choose another brand”; (2) “I consider myself to be highly loyal to brand X”; and (3) “When another brand is on sale, I would purchase it rather than brand X.” (coefficient alpha=0.87). Ahluwalia et al. (2000) used the same scale to measure brand commitment.

Mark Lett (2014) 25:179–191 183

by stating that the target article was fictional and that they should therefore ignore the information it presented.

3.2 Stimuli and variables

We selected smartphones as the target product category, because the student respon- dents were familiar with this product category. We fabricated target messages on the basis of a series of pretests. The high-severity message focused on a recent scientific report that indicated that iPhone smartphone users were 207 times more likely to suffer from life-threatening brain hemorrhages. The message also indicated that a recall was unavoidable. The low-severity report stated that about 10,000 iPhone5s were being recalled over a battery issue. In a pretest, 62 participants read either the high- or low-severity message, in relation to an unknown brand of smartphones, and rated event severity on a seven-point scale (1–7). They considered the messages significantly different in severity (meanhemorrhages=6.53; meanbattery=3.36; t=9.72, p<0.001). The target messages are available from the authors.

We measured consumer responses to the recall (i.e., our dependent variable) by assessing the degree to which the participants’ attitude toward the iPhone changed from before they read the message to after. Specifically, for each participant, we subtracted the postmessage mean attitude score from the premessage mean attitude score; these scores came from four seven-point Likert scales (anchored by “good/bad,” “beneficial/harmful,” “desirable/undesirable,” “favorable/unfavorable”) (coefficient alpha=0.96), adapted from Ahluwalia et al. (2000).2

3.3 Results

Our prediction that brand commitment attenuates negative consumer responses (i.e., attitude change in the experiment) for low-severity recalls but augments them for high-severity recalls implies an interaction between recall severity and brand com- mitment. To test our prediction, we performed an ordinary least squares regression on attitude change with independent variables (1) brand commitment, (2) a dummy variable for high (=1) and low (=0) recall severity, and (3) their interaction. Overall, the model was significant (F (3, 129)=43.21, p<0.001), and the results showed a significant interaction (bBrand Commitment×Recall Severity=0.26, t=2.44, p<0.05). To further explore the interaction, we next examined the slopes of brand commitment at each level of severity. As we predicted, the slope of brand commit- ment was (marginally) significant and negative (bBrand Commitment=−0.076, t=−1.89, p<0.10) in the low-severity condition and (marginally) significant and positive (bBrand Commitment=0.185, t=1.78, p<0.10) in the high-severity condition. We also conducted a spotlight analysis at one standard deviation above and below the mean of brand commitment. We present the results from the spotlight analysis in Fig. 1 which

2 Measuring attitude change as a difference raises the issue of whether the difference scores are reliable. Extant research (e.g., Collins 1996) has shown that difference scores are unreliable only when the pretest (x) and posttest (y) standard deviations are equal (i.e., 1=σx/σy=1) and when the correlation between the two scores is high (ρxy≈1). Considering our data, 1=0.57 and ρxy=0.47, in support of the reliability of our measure. The temporal proximity of the pre- and posttest scores remains a limitation.

184 Mark Lett (2014) 25:179–191

illustrates that the highly committed consumers experienced less (more) attitude change in the low (high) severity condition than the less committed consumers.

3.4 Process tests

As outlined above, we expect that committed consumers express significantly more counterarguments in the low-severity recall condition than their less committed counterparts. Furthermore, we reason that these counterarguments account for (i.e., mediate) the observed difference in attitude change between high- and low- commitment consumers. We also expect that committed consumers express signifi- cantly more incongruity thoughts in the high-severity recall condition than less- committed consumers, and we again anticipate that these incongruity thoughts account for (i.e., mediate) the observed difference in attitude change between the high- and low-commitment consumers.

Two judges coded participants’ listed thoughts related to the focal article using two categories: counterarguments and incongruity thoughts. We followed Ahluwalia et al.’s (2000) approach for the coding of the counterarguments. The judges achieved high intercoder reliability (agreement>83 %) and resolved disagreements through discussion. We used these thought protocols in our process analysis.

Considering the low-severity recall scenario, counterarguments were more preva- lent among high- rather than low-commitment consumers (meanlow BC=0.71, meanhigh BC=1.34; t=2.44, p<0.05). A different pattern emerged in the high-severity condition: Again as predicted, incongruity thoughts were much more prevalent among high- than low-commitment consumers (meanlow BC=1.06, meanhigh BC=1.87; t=2.76, p<0.01).

We conducted the mediation analysis separately for the low- and the high-severity conditions. To test if counterarguments mediate the identified attenuating effect of brand commitment in low-severity recalls, we included the number of counterargu- ments as mediators of the effect of brand commitment on attitude change. Following Zhao et al. (2010), we assessed mediation using the bias corrected bootstrap test of

0.3267

2.9364

0.4628

2.6045

0.599

2.2727

0

0.5

1

1.5

2

2.5

3

3.5

Low Severity High Severity

A tt

it ud

e ch

an ge

( at

ti tu

de b

ef or

e- at

ti tu

de a

ft er

)

Brand Commitment (+1 SD) Brand Commitment (Mean) Brand Commitment (-1 SD)

Fig. 1 Spotlight analysis: illustrating the attenuating and augmenting effects of brand commitment

Mark Lett (2014) 25:179–191 185

the indirect effect. As expected, counterarguments emerged as a significant mediator of brand commitment’s effect on attitude change. Using 5,000 bootstrap samples, the bias corrected 95 % confidence interval for the indirect effect of the path through counterarguments was [−0.051; –0.001] with a point estimate of −0.019. We note that, since the 95 % confidence interval does not include zero, we can conclude that the estimate of the indirect path from brand commitment to attitude change through the number of counterarguments is significant at p<0.05. Thus, counterarguments mediate the difference in attitude change between high- and low-commitment con- sumers in low-severity recalls.

We then repeated the mediation analysis for the high severity condition and included the number of incongruity thoughts as mediators of the effect of brand commitment on attitude change. As predicted, incongruity thoughts emerged as a significant mediator of brand commitment’s effect on attitude change. Again using 5,000 bootstrap samples, the bias-corrected 95 % confidence interval for the indirect effect of the path through incongruity thoughts was [0.061; 0.262] with a point estimate of 0.147. The confidence interval again did not include zero. Thus, incon- gruity thoughts mediate the difference in attitude change between high- and low- commitment consumers in high severity recalls.

In summary, the experiment provides empirical support for both H1 and H2.

4 Event study

To increase the external validity of our study, we also conducted an event study consid- ering product recalls in the automobile industry. To ensure that stockmarket returns offer a good measure of consumer responses to product recalls, we turn to Wiles et al.’s (2010) study of deceptive advertising. In their survey of stock analysts, they find that the respondents attended assiduously to the effects of the transgression on consumers’ perceptions of the firm, in the belief that those perceptions would affect sales and thus financial performance. This reasoning is evenmore applicable for product recalls because, in this case, the products themselves fail, not just the communication. Thus, we assert that a product recall may be even more susceptible to negative reactions by consumers.

4.1 Sampling procedures

Similar to other event studies (e.g., Chu et al. 2005), we used the Wall Street Journal (WSJ) index to identify car manufacturer product recalls that the WSJ reported between 2001 and 2009. Our initial sample consisted of 66 recalls. We then excluded any duplicate announcements due to repeated recalls and conducted a Factiva database search (e.g., McWilliams and Siegal 1997) to remove firms with confounding events. Our final sample consists of 55 recalls of seven publicly traded car manufacturers. The seven firms are General Motors, Ford, Nissan, Daimler, Honda, Chrysler, and Toyota.

4.2 Variables and analysis

Our dependent variable is the firm’s cumulative average abnormal return (CAAR) resulting from a recall event. We used the Fama-French four-factor model (Fama and

186 Mark Lett (2014) 25:179–191

French 1993; Srinivasan and Bharadwaj 2004) to generate the expected return for security i on day t.

Brand commitment We used Interbrand’s “Best Global Brands” ranking and its brand values as a measure of the focal firms’ brand commitment. According to Interbrand, two key aspects of their brand value measure are (1) the brand’s ability to create loyalty and (2) the portion of purchase decisions that can be attributed to the brand (Interbrand 2013). Thus, while certainly not a perfect measure, we reason that Interbrand’s brand value measure is an acceptable surrogate measure of brand commitment.

Interbrand makes the ranking and values of the top 100 brands available on its website going back to 2001 (Interbrand 2013). Five of the seven car manufacturers that form our sample appear in the top 100 ranking during the focal years, and we used the respective yearly brand values listed as an estimate of the recalling firm’s brand commitment. We note that the brand values of our sample firms varied over time. Furthermore, we used a brand value of zero for the two firms (General Motors and Chrysler) that did not appear in any of the yearly rankings. We note that Daimler does not appear in the ranking either; however, its main car brand, Mercedes, does. Hence, we use the respective Mercedes brand value as the brand value for Daimler. We view this as unproblematic as all Daimler recalls involved the Mercedes brand. We also note that Chrysler and Daimler were one legal entity during parts of our observation period. However, given the different brand values attached to the two companies’ cars, we treated the two as separate entities in our analysis. As we mention later, we control for firm specific heterogeneity, which should parse out firm specific idiosyncrasies.

Recall severity Objective and/or third-party severity scores for our sample recalls were not available. We hence relied on three expert coders who rated our sample recalls as either high or low in recall severity based on the information provided in the WSJ articles. The inter-coder reliability was 86 %, and all disagreements were resolved through discussion. Of the 55 recalls in our sample, 28 (51 %) were coded as high- severity recalls and 27 (49 %) as low-severity recalls. We used an indicator variable for recall severity (high-severity recall=1; low-severity recall=0) in our analysis.

Modeling approach The typical approach in event studies is to regress abnormal returns on a set of explanatory variables (MacKinlay 1997). We follow this approach here. Furthermore, each of our sample firms issued at least three recalls during our observation period on which the WSJ reported (Ford, 15, General Motors, 12, Toyota, 11, Chrysler, 7, Daimler, 4, Nissan, 3, and Honda, 3). Thus, we have repeated observations per sample firm, and we hence estimated a random-effects regression model as specified below to test our hypothesis.3 Using a random-effects regression model greatly reduces the possibly pernicious effect of an omitted variable bias.

3 We also estimated a fixed-effects model and then conducted the Hausman test to determine whether the random effects model is appropriate. The Hausman test yielded a statistically non-significant χ2 (χ2=0.296) suggesting that the random effects model is appropriate.

Mark Lett (2014) 25:179–191 187

CAARit ¼ β0 þ β1Brand Commitmentit þ β2Recall Severityit þβ3Brand Commitmentitx Recall Severityit þ αi þ εit ð1Þ

where CAARit is the cumulative average abnormal return for firm i at time t, β’s are coefficients to be estimated, αi is the random intercept for each firm (i.e., the between- firm error), and εit is the within-firm error. We present descriptive statistics in Table 1.

4.3 Event study results

Consistent with prior event studies (e.g., Wiles et al. 2010), we employed the CAAR from the [−1,0] event window in our analysis. The test statistics revealed a (marginally) significant, negative CAAR for the [−1,0] event window (−0.37 %; generalized sign test z=−1.913; p<0.10). We show our regression results in Table 2.

Overall, the model is significant (Wald χ2 (3)=15.90, p<0.01). Moreover, we again find empirical support for H1. First, the brand commitment main-effect was positive and significant (bBrand Commitment=0.000068, z=2.03, p<0.05) suggesting that high levels of brand commitment attenuate negative returns in low-severity recalls. Note that, due to our coding structure (i.e., high severity recalls are coded as 1), the brand commitment main effect captures brand commitment’s impact on abnormal returns in low-severity recalls. Second, the interaction term between brand commitment and recall severity was negative and significant (bBrand Commitment×Recall Severity=−0.00016, z=−3.67, p<0.01). We also investigated the nature of the slope of abnormal returns in high-severity recall cases considering brand commitment by adding the bBrand Commitment and the bBrand Commitment x Recall Severity coefficients and calculating the standard error for the expression. As expected, the combined coefficient was negative and significant (bCombined=−0.000094, z=−3.24, p<0.01), suggesting that brand com- mitment augments negative returns in high-severity recalls. We also reversed the recall severity coding structure (i.e., we coded low severity recalls as 1) and re-ran the model. In this case, the brand commitment main effect captures brand commitment’s impact on abnormal returns in high severity recalls. The results, of course, were the same (i.e., bBrand Commitment=−0.000094, z=−3.24, p<0.01).

Table 1 Correlations and summary statistics

Correlations

1 2 3

Variables

1. Cumulative average abnormal return [−1, 0] 1.000 2. Brand commitment −0.105 1.000 3. Recall severity −0.137 −0.198 1.000

Summary statistics

Mean −0.370 12045 0.509 Standard deviation 1.902 10850 0.505

None of the correlations are significant at p<0.05

188 Mark Lett (2014) 25:179–191

5 Conclusions

Our study contributes to marketing theory in two ways. First, we extend the product recall literature by revealing the importance of brand commitment in product recall incidents. Product recalls are increasingly rampant in the marketplace, and they have provoked a significant amount of research attention (e.g., Cleeren et al. 2008; Chen et al. 2009; van Heerde et al. 2007). In this study, we systematically explore how brand commitment, in combination with recall severity, affects consumer responses to product recalls. We find that, while brand commitment attenuates negative consumer responses in low-severity recalls, it augments them in high-severity recalls. Thus, while brand commitment seems to provide a reservoir of goodwill in the former case, it acts as a liability in the later. Second, our study contributes to the brand commit- ment literature. To the best of our knowledge, this investigation is one of the first to pinpoint circumstances under which brand commitment constitutes a liability.

We believe that our findings also offer useful implications for marketing practice. We show that brand commitment can produce negative outcomes, so marketing managers must take our findings into consideration. Noting the value of brand commitment, common beliefs seem to imply that a brand with many committed consumers enjoys a reservoir of goodwill, regardless of negative events. Specifically, we surveyed 35 U.S. executives about whether brand commitment should help or hurt when product recalls occur; most (81 %) asserted that brand commitment would be advantageous for a recalling firm. Our study contests this widespread conventional wisdom.

While we believe that we have broken some new ground with this work, there are clear limitations, several of which provide avenues for further research. First, we consider consumer responses immediately following the recall announcement; we thus cannot examine how the firm’s handling of the recall might affect consumer responses. A well-managed, high brand commitment/high-severity recall may offset the augmenting effects of brand commitment, whereas a poorly managed, high brand commitment/high-severity recall could evoke even more negative consumer responses. Also, perhaps a well-managed, high brand commitment/low-severity recall could lead to positive consumer responses. Additional research should examine how brand commit- ment, recall severity, and recall management jointly affect consumer responses.

Table 2 Random-effects regression results with cumulative average abnormal return as the dependent variable

Variable β SE z

Brand commitment 0.000068 0.000033 2.03

Recall severity 1.381041 0.7203483 1.92

Brand commitment×recall severity −0.000162 0.000044 −3.67 Constant −1.066828 0.5752478 −1.85 n 55

Wald χ2 15.9

df 3

Pvalue <0.01

Mark Lett (2014) 25:179–191 189

Second, product recalls are multi-faceted, heterogeneous events, and they vary across several factors, with recall severity being one of them. For example, some recalls involve more than 100,000 units whereas others only involve 10,000 units. Also, some recalls involve convenience products (e.g., toothpaste), and others in- volve shopping products (e.g., cameras). It is conceivable that consumers might respond differently depending on, e.g., the amount and type of product involved, and future research might examine additional dimensions of recall events.

Third, our results suggest that product recalls must stem from a serious, potentially life-threatening offense for the augmenting effect of brand commitment to play a role. This raises questions of whether the augmenting effect also manifests in less egre- gious recall events and, more generally, where the “tipping point” is, beyond which brand commitment acts as a liability.

Finally, we believe that it would be interesting to examine whether brand com- mitment’s dual role also occurs in other types of negative firm events besides product recalls. For example, what happens when the firm is accused of having polluted the environment? Will brand commitment attenuate or augment potentially negative consumer responses? Future research should test whether brand commitment’s atten- uating and augmenting effects also manifest in other types of firm events.

References

Ahluwalia, R., Burnkrant, R. E., & Unnava, H. R. (2000). Consumer response to negative publicity: The moderating role of commitment. Journal of Marketing Research, 37(2), 203–214.

Beatty, S. E., Kahle, L. R., & Homer, P. (1988). The involvement-commitment model: Theory and implications. Journal of Business Research, 16(2), 149–167.

Cheah, E. T., Chan, W. L., & Chieng, L. L. (2007). The corporate social responsibility of pharmaceutical product recalls: An empirical examination of U.S. and U.K. markets. Journal of Business Ethics, 76(4), 427–449.

Chen, Y., Ganesan, S., & Liu, Y. (2009). Does a firm’s product recall strategy affect its financial value? Journal of Marketing, 73(6), 214–226.

Chu, T.-H., Lin, C.-C., & Prather, L. J. (2005). An extension of security price reactions around product recall announcements. Quarterly Journal of Business and Economics, 44(3/4), 33–48.

Cleeren, K., Dekimpe, M. G., & Helsen, K. (2008). Weathering product-harm crisis. Journal of the Academy of Marketing Science, 36(2), 262–270.

Cleeren, K., van Heerde, H. J., & Dekimpe, M. G. (2013). Rising from the ashes: How brands and categories can overcome product-harm crises. Journal of Marketing, 77(2), 58–77.

Collins, L. M. (1996). Is reliability obsolete? A commentary on ‘are simple gain scores obsolete. Applied Psychological Measurement, 20(3), 289–292.

Fama, E. F., & French, K. R. (1993). Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, 33(1), 3–56.

Fournier, S. (1998). Consumers and their brands: Developing relationship theory in consumer research. Journal of Consumer Research, 24(4), 343–373.

Interbrand (2013). Best global brands. http://www.interbrand.com. Accessed March 15, 2013. Jones, G. H., Jones, B. H., & Little, P. (2000). Reputation as reservoir: Buffering against loss in times of

economic crisis. Corporate Reputation Review, 3(1), 21–29. Klein, J., & Dawar, N. (2004). Corporate social responsibility and consumers’ attributions and brand

evaluations in a product-harm crisis. International Journal of Research in Marketing, 21(3), 203–217. Lei, J., Dawar, N., & Gurhan-Canli, Z. (2012). Base-rate information in consumer attributions of product-

harm crises. Journal of Marketing Research, 49(3), 336–348. MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of Economic Literature, 35(1),

13–39.

190 Mark Lett (2014) 25:179–191

McWilliams, A., & Siegal, D. (1997). Event studies in management research: Theoretical and empirical issues. Academy of Management Journal, 40(2), 626–657.

Oliver, R. L. (1993). Cognitive, affective, and attribute bases of the satisfaction response. Journal of Consumer Research, 20(3), 418–430.

Rhee, M., & Haunschild, P. R. (2006). The liability of good reputation: A study of product recalls in the U.S. automobile industry. Organization Science, 17(1), 101–117.

Srinivasan, R. & Bharadwaj, S. (2004). Event studies in marketing strategy research. In: Moorman C. and Lehmann D. (eds). Assessing marketing strategy performance. MSI

Van Heerde, H., Helsen, K., & Dekimpe, M. G. (2007). The impact of a product-harm crisis on marketing effectiveness. Marketing Science, 26(2), 230–245.

Wiles, M., Jain, S. P., Mishra, S., & Lindsey, C. (2010). Stock market response to regulatory reports of deceptive advertising. Marketing Science, 29(5), 828–845.

Zhao, X., Lynch, J. G., Jr., & Chen, Q. (2010). Reconsidering Baron and Kenny: myths and truths about mediation analysis. Journal of Consumer Research, 37(3), 197–206.

Mark Lett (2014) 25:179–191 191

Copyright of Marketing Letters is the property of Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder’s express written permission. However, users may print, download, or email articles for individual use.

  • Product recalls and the moderating role of brand commitment
    • Abstract
    • Introduction
    • Conceptual background and hypotheses
      • Brand commitment
      • Product recall severity
      • Brand commitment’s role when product recalls occur
    • Laboratory experiment
      • Procedure
      • Stimuli and variables
      • Results
      • Process tests
    • Event study
      • Sampling procedures
      • Variables and analysis
      • Event study results
    • Conclusions
    • References

Comments are closed.