|
Article Excerpt
Product-harm crises are among a firm's worst nightmares. A firm may experience (i) a loss in baseline sales, (ii) a reduced own effectiveness for its marketing instruments, (iii) an increased cross sensitivity to rival firms' marketing-mix activities, and (iv) a decreased cross impact of its marketing-mix instruments on the sales of competing, unaffected brands. We find that this quadruple jeopardy materialized in a case study of an Australian product-harm crisis faced by Kraft peanut butter. We arrive at this conclusion by using a time-varying error-correction model that quantifies the consequences of this crisis on base sales, and on own- and cross-brand short- and long-term effectiveness. The proposed modeling approach allows managers to make more informed decisions on how to regain the brands' pre-crisis performance levels.
Key words: brand management; product recalls; brand equity; marketing and public policy; error-correction models; time-varying parameters; time-series models; missing-data problems; Gibbs sampling methods
1. Introduction
Most market-oriented firms allocate huge resources to build their brands. A brand's equity, however, can be very fragile. Among its biggest threats are product-harm crises, which can be defined as well-publicized events wherein products are found to be defective or even dangerous (Dawar and Pillutla 2000). (1) Product-harm crises can distort long-standing favorable quality perceptions, tarnish a company's reputation, cause major revenue and market-share losses, lead to costly product recalls, and devastate a carefully nurtured brand equity. Usually, the crisis relates to a particular brand. In 2000, Bridgestone/Firestone recalled 6.5 million tires after news broke that more than 100 people had died in accidents involving defective tires (Advertising Age 2000). In 1999, Coca-Cola was forced to withdraw 30 million cans and bottles in northern Europe following a scare in Belgium (Guardian 1999). Other notorious cases include Intel's flawed Pentium chip, Johnson & Johnson's cyanide-laced Tylenol, and the benzene contamination of Perrier. Occasionally, the crisis involves an entire product category such as poultry (bird flu), silicon breast implants, and beef (mad-cow disease).
Because of the increasing complexity of products and closer scrutiny by manufacturers and policy makers as well as higher demands by consumers, product-harm crises are expected to occur ever more frequently (Dawar and Pillutla 2000), while heightened media attention will also make them more visible to the general public (Ahluwalia et al. 2000). As Kahn pointedly put it, "The good news about brands is that people know who you are. The bad news is that if something goes wrong, everyone knows" (Knowledge@Wharton 2005). However, in spite of the devastating impact of product-harm crises, little systematic research exists to assess its marketing consequences. Academic studies in the area have either experimentally investigated consumer reactions to hypothetical product crises (Ahluwalia et al. 2000, Dawar and Pillutla 2000) or used aggregate, event study-based financial measures (Davidson and Worrell 1992, Marcus et al. 1987). Very limited attention has been devoted to adequately quantify the impact of actual product crises on relevant marketing metrics such as sales or market share. This quantification might not only be relevant to the affected brands' management but also to policymakers or judges who must assess how much a company needs to be compensated for a product-harm crisis brought on by a third party (for a discussion of the use of marketing models in the public-policy domain, see Hanssens et al. 2005).
In this paper, we argue that the implications of a brand-specific product-harm crisis often go beyond the "obvious" short-run sales or market-share loss for a variety of reasons. First, the brand's own marketing-mix effectiveness might be reduced. For instance, because customers' trust might have been breached, advertising might now give less "bang for the buck" than before the crisis. Moreover, the brand might now have less potential to attract potential switchers or has become more sensitive to competitive activities. The latter phenomenon is especially relevant because competitors might try to exploit the marketing opportunities that arise because of the brand's misfortune by reducing their own price or increasing their advertising expenditures. Michelin North America, for instance, hiked its advertising budget to run a print campaign touting tire safety and quality in the wake of Bridgestone/Firestone's tire recall (Advertising Age 2000). Because of this changed own and cross-effectiveness, relying on the before-crisis estimates can seriously underestimate the extent of corrective action needed. Insights into postcrisis marketing effectiveness are crucial to managers who want to make informed decisions on how to restore brand performance to its precrisis level.
To investigate whether the implications of a product-harm crisis reach beyond the obvious losses in sales, we present in this paper a case study of a devastating product-harm crisis that affected Kraft Food Australia in the summer of 1996. More than 100 cases of salmonella poisoning potentially linked to Kraft-made peanut butter made management recall its two key brands for multiple weeks. Using weekly advertising and store scanning data for a precrisis period of more than a year and a postcrisis period of more than three years, we calibrate a time-varying error-correction model that quantifies the consequences of this crisis on both brands' base sales and on own- and cross-brand short- and long-term effectiveness, allowing management to make more informed decisions on how to regain the brands' precrisis performance levels.
The rest of the paper is structured as follows. First, we discuss the literature on product-crisis effects. Next, we present our model, discuss the data and results, and conclude with managerial implications and suggestions for further research.
2. Product-Crisis Effects
Even though product-crisis incidents are increasingly prevalent, fairly little systematic research has been conducted on the topic (Klein and Dawar 2004). Existing research can be broadly classified into three streams. The first stream consists of descriptive, often case-based studies suggesting which strategies work or do not work in the marketplace. Checklists are typically provided detailing the appropriate managerial actions to avoid product crises and how to respond when they occur (e.g., Mitroff 2004, Mitroff and Kilmann 1984, Rupp and Taylor 2002, Smith et al. 1996, Weinberger et al. 1993). These studies, while offering sound advice, provide little direction for understanding the underlying mechanisms through which product crises harm the company or brand (Ahluwalia et al. 2000), nor do they quantify the extent of the damage incurred (or averted).
Such an understanding of the underlying mechanisms is explicitly sought in a second research stream where lab experiments are used to assess the impact of hypothetical crises and moderating variables on brand evaluations, such as consumer expectations (Dawar and Pillutla 2000), commitment to the brand (Ahluwalia et al. 2000), brand loyalty (Stockmeyer 1996), the perceived locus of the problem (Griffin et al. 1991), and prior corporate social responsibility (Klein and Dawar 2004). Lab experiments have also been used to determine whether gender differences matter in blame attributions with a product-harm crisis (Laufer and Gillespie 2004). While these studies are well grounded in various psychological theories, their use of experimentally manipulated hypothetical product crises is likely to limit the external validity of the insights. Moreover, these studies typically do not attempt to quantify the financial implications of the crises.
The third stream of research focuses on gauging the effects of actual product-harm crises on a variety of performance measures including security prices (e.g., Chu et al. 2005, Govindaraj et al. 2004, Davidson and Worrell 1992, Marcus et al. 1987) and category consumption (e.g., Burton and Young 1996, Marsh et al. 2004). However, both aforementioned performance metrics are aggregate indicators and may not be as informative as more disaggregate analyses. Primary-demand measures, for example, do not account for the fact that not all incumbents can be affected to the same extent by a crisis. Indeed, the locus of the problem may be internal to some but external to others (Klein and Dawar 2004), while they may also have reacted differently to the crisis (Griffin et al. 1991). Stock-price reactions, while firm specific, do not identify the underlying mechanisms through which the resulting value loss emerged: Is it entirely due to a loss in baseline sales, or do investors also penalize the company for a potential loss in marketing-mix effectiveness because of the crisis? Do they fear that the brand has lost so much equity that it will become more vulnerable to future competitive actions? Moreover, if the company has umbrella branded its products, what part of the combined stock-market reaction can be attributed to, respectively, the product affected directly by the crisis and negative spillovers to other products sold under the same label (Sullivan 1990)?
Our paper contributes to the third research stream in that we explicitly quantify the performance implications of the crisis. However, unlike previous studies in this tradition, we present a much more disaggregate picture of the postcrisis situation in that (i) we explicitly distinguish between the different incumbent firms, recognizing that some players may actually benefit from the misfortune of their competitor(s), (ii) we allow for differential performance implications for different brands owned by the same company, and most importantly (iii) we identify various ways through which the brand may be affected, both in the short and in the long run: a loss in baseline sales, a reduced own marketing-mix effectiveness, an increased cross-sensitivity to competitive actions, and a reduced cross-brand impact for the own actions.
2.1. Baseline Sales
The most obvious effect of a product-harm crisis is the immediate loss in own-brand sales or market share. For example, sales at Wendy's restaurants in the San Francisco Bay area dropped 30% after a woman claimed to have found a finger in her chili (Financial Times 2005). Similarly, following a food-poisoning scandal in June 2000, sales of Snow Brand milk in Japan dropped 88% compared to a year earlier, while the brand's market share tumbled from 40% to less than 10% (Finkelstein 2005). To revive the brand's sales (and, in some cases, the entire category), managers might feel inclined to reduce the product's price or to substantially increase its advertising support. For example, after years of disastrous quality problems and product recalls, General Motors ran a major "road to redemption" campaign claiming that the company was "building the best cars and trucks in our history" (New York Times 2004). Advertising and promotional efforts could be increased to create awareness about the comeback and regain trust from risk-averse consumers (Byzalov and Shachar 2004).
2.2. Own Effects
Perhaps less obvious is that the crisis might have affected the effectiveness of marketing instruments. The crisis could have damaged the brand's equity (Dawar and Pillutla 2000), the firm's credibility (MacKinsey and Lutz 1989), and thereby the customer-brand relationship (Aaker et al. 2004). These consequences might negatively affect the effectiveness of subsequent advertising investments (Aaker 1991, Goldberg and Hartwick 1990). Similarly, when customers are exposed to negative information about the product, its perceived differentiation might be reduced (Ahluwalia et al. 2000), which could in turn increase the magnitude of its price elasticity (Boulding et al. 1994). Moreover, one should take into account that marketing-mix effects can reach well into the future (Dekimpe and Hanssens 1999, Jedidi et al. 1999), making it important to consider the long-run effects of the crisis as well. Product-harm crises can imperil long-standing favorable impressions and have performance implications that linger into the future. Indeed, negative information is known to be more informative and persistent than positive...
|
