آیا هزینه های اطلاعات بر انتخاب مصرف کننده تاثیر می گذارد ؟ برچسب های تغذیه ای در آزمایش یک سوپر مارکت

This paper investigates whether information costs under currently regulated nutritional labeling prevent consumers from making healthier food choices. We implement five nutritional shelf label treatments in a market-level experiment. These labels reduce information costs by highlighting and summarizing information available on the Nutritional Facts Panel. Following a difference-in-differences and synthetic control method approach, we analyze weekly store-level scanner data for microwave popcorn purchases from treatment and control stores. Our results suggest that consumer purchases are affected by information costs. Implemented low calorie and no trans fat labels increase sales. In contrast, implemented low fat labels decrease sales, suggesting that consumer response is also influenced by consumers' taste perceptions. A combination of these claims into one label treatment increases information costs and does not affect sales significantly.

Existing research documents consumers' general understanding of the link between food consumption and health, and widespread interest in the provision of nutritional information on food labels (e.g. Williams, 2005 and Grunert and Wills, 2007). However, consumers cannot verify this information at any point from purchase to consumption.1 Instead, they base their product choice on beliefs arrived at by way of a labyrinth of information printed on food packages. In such markets, firms might not have an incentive to fully reveal their product quality (Bonroy and Constantos, 2008), might try to highlight certain attributes in their advertising claims while shrouding others (Gabaix and Laibson, 2006), or provide information in a less salient fashion (Chetty et al., 2007). The Nutrition, Labeling, and Education Act (NLEA) of 1990 gave the Food and Drug Administration (FDA) the authority to require nutritional labeling for most food products. In 1994, the Nutrition Facts Panel (NFP) was implemented in order to improve consumer access to nutritional information and to promote healthy food choices. This paper uses a supermarket-level experiment to address the relationship between information costs and healthy food choices under these labeling regulations. About 50% of consumers claim to use the NFP when making food purchasing decisions (Blitstein and Evans, 2006). Consumers trying to lose weight are more likely to read the NFP (Mandal, 2008), and NFP use can result in weight loss and a decrease in obesity (Variyam and Cawley, 2006). However, self reported consumer use of nutrition labels declined from 1995 to 2006, with the largest decline for younger age groups (20–29 years) and less educated consumers (Todd and Variyam, 2006). This decline could be a result of consumers' inability to perform quantitative tasks (Levy and Fein, 1998), and preferences for short health claims and short front label claims instead of NFP's lengthy back label explanations (e.g. Levy and Fein, 1998, Williams, 2005, Wansink et al., 2004 and Grunert and Wills, 2007).2 Yet, simple claims, such as low fat labels could potentially mislead consumers and increase their caloric food intake through perceptions of an increased acceptable serving size and a reduction in consumption guilt (Wansink and Chandon, 2006), especially when combined with a positive image and suggestive health references (Geyskens et al., 2007). Conversely, perceived tradeoffs between nutritional considerations and taste preferences could prevent consumers from choosing reduced-fat alternatives if they are labeled as such ( Yeomans et al., 2001, Stubenitsky et al., 2000 and French et al., 1999). The limited number of market-level empirical studies exhibits mixed results regarding consumer use of nutritional information. Displaying lists of information on vitamins and minerals as well as sugar content in supermarkets resulted in increased nutritional information use (Russo et al., 1986), and voluntary labels had significant effects on consumer choices prior to the NLEA (Ippolito and Mathios, 1990). Still, Mojduszka and Caswell (2000) argue that information provided by firms voluntarily prior to the NLEA was incomplete and not reliable. Mathios (2000) finds that mandatory guidelines resulted in a significant decline in sales of high fat products, despite prior voluntary disclosure of low-fat products, and Teisl et al. (2001) find that consumer behavior was significantly altered by the NLEA, but purchases of “healthy” products increased only in some of the product categories. Less attention has been paid to interdependencies of regulation and alternative information sources in these studies. This is important because experimental research (Cain, et al., 2005) suggests that people do not sufficiently take motives of the information source into account when evaluating information, even after disclosure of conflicts of interest. In this context, Ippolito and Pappalardo (2002) suggest that regulatory rules and enforcement policy induced firms to move away from reinforcing nutritional claims. Critical news coverage of regulatory challenges (Nestle, 2000), and the “Food News Blues” in general (Kantrowitz and Kalb, 2006) could have also contributed to decreased labeling use over time. Our experimental design adds to this literature by focusing on information costs under current NFP labeling. Conducting our experiment in a real market setting eliminates possible bias generated in hypothetical experiments and survey responses, and controls for potential confounding factors such as marketing claims and media coverage. We implemented nutritional shelf labels for one product category (microwave popcorn) in cooperation with a major supermarket chain in five treatment stores over a period of four weeks. The supermarket chain also provided store-level scanner data for a total of 32 stores, covering a time period before and after our labeling implementation. Our collected NFP information indicated substantial variation in nutrient content and suggested serving size across products included in the data. Consumers trying to compare products based on their nutritional characteristics might therefore face significant information costs. We reduce information costs by either repeating or summarizing NFP information and providing it a new format. Using low calorie, low fat, and no trans fat claims, we address the following questions: (i) Are consumer purchases affected by nutritional shelf labels? (ii) Do effects differ depending on nutrients displayed (e.g. calories versus fat content)? (iii) Do effects depend on disclosure of information source (FDA)? (iv) Do effects differ depending on display of a single versus multiple nutrients on a label? and (v) Do we find evidence consistent with consumers making inferences about the nutritional content of unlabeled products? Following a difference-in-differences and synthetic control method approach, we find results consistent with information costs mattering and conclude that nutritional information is not provided effectively under current labeling guidelines. In particular, we find that a shelf label of no trans fat significantly increases sales of treated products, even though this information is already provided in a less uniform format. Low calorie labels also significantly increase sales of treated products. Low fat labels, on the other hand, significantly reduce quantity sales of targeted products, especially when adding an FDA approval to our labeling treatments. We attribute this effect to consumers having less favorable taste perceptions of low fat foods than of low calorie foods. When combining claims in a single label, we do not detect significant purchase responses because this treatment increases information costs for the consumer. Finally, we find no consistent evidence that consumers make inferences about unlabeled products and their relatively inferior nutritional quality. The synthetic control method further detects the largest labeling effect immediately following our initial implementation. Labeling effects dissipate quickly after our treatment period for the low calorie and low fat treatment, but persist for the no trans fat label. No trans fat products are highlighted in manufacturer claims and are easier to identify by consumers under the current NFP labeling. In the next section, we describe our experimental design and the main features of our data. We introduce our empirical specification, report estimation results, and test the robustness of our findings in Section 3. In Section 4, we conclude by discussing our results and their relevance for regulatory changes.

In this paper, we analyzed whether information costs prevent consumers from fully optimizing their purchase decisions with regards to currently available nutritional information. Focusing on consumer information costs under the NFP, we use a market-level experiment to estimate the effect of making nutritional information more salient and easier to process. Our implemented nutritional shelf labels allow consumers to make direct comparisons regarding nutrient content by either repeating available information in a more uniform format (no trans fat label), or translating quantitative information into categorical statements (low calorie and low fat label). Our interventions focus on one nutrient or a combination of nutrients in a single label. Combining multiple nutrient claims in one label increases the label's information content, but information costs also increase. We also analyze whether unlabeled products were affected, as we implicitly provide information about the superior nutritional content of these product alternatives. Our empirical design further allows us to incorporate and test previous findings in the literature on consumer response to labeling information. We were able to address potential differences in consumer purchase response based on the nutrient displayed. We tested low calorie labels because calorie content has been determined as the most relevant nutrient in relation to weight gain and obesity prevention (CDC, 2008).17 The World Health Organization (WHO), in contrast, endorses the promotion of low fat products as one strategy to reduce obesity rates (WHO, 2004). Yet, simple low fat claims might increase overall food intake due to reduced consumption guilt (e.g. Wansink and Chandon, 2006), or trigger negative taste perceptions (e.g. Yeomans et al., 2001). We further wanted to compare possible information effects for these nutrients to no trans fat labeling information. Health concerns related to trans fats received a lot of media attention, and no trans fat advertisements were readily adopted by food manufacturers. Consumers might therefore be well informed about this nutrient and more able to readily incorporate information into their purchase decision. Finally, as consumers might view these labels as in-store nutritional advertisement, we added an FDA approval to one of our label treatments to investigate whether it increased the credibility of the provided information. We implemented five labeling treatments for one product category (microwave popcorn) in five stores over a time period of four weeks in the fall of 2007. The supermarket chain provided weekly store-level scanner data for these treatment stores and 27 control stores within the same price division over a period of 14 weeks. By adding the information provided on the NFP for each product included in this data, we find substantial variation in nutrient content and suggested serving size in our product category. Consumers trying to compare products based on their nutritional characteristics might therefore face significant information costs. Estimations of average treatment effects of our labeling intervention are based on difference-in-differences and triple-difference approaches identified by a cross-sectional and time-series control structure. In addition, we draw inference about the effect of our labeling treatments on product sales with a synthetic control method approach. Our analysis suggests that consumer purchases are affected by our labeling treatments. Information costs prevent some consumers from incorporating nutritional information in their purchasing decisions under currently implemented labeling regulations. Our findings are not driven by consumers simply paying more attention to labeled products, since we find no statistically significant effects of pooling all labeling treatments. However, a labeling treatment focusing on calorie content significantly increases sales, while focusing on fat content decreases sales. Displaying no trans fat labels also has positive and significant effects on sales, even though this information is already provided on the NFP and is highlighted in manufacturer claims. This effect dissipates, however, when combining the no trans fat claim with additional nutritional claims. Throughout the specifications, we find that a combination of claims into a single label— an improvement in information content—does not result in a significant effect on sales. This may happen because multiple claims also increase information costs for consumers. Finally, our analysis suggests that the most sizable impact is observed right after the label implementation, with effect dissipating after the treatment period for low calorie and low fat labels. For the no trans fat labeling treatment, the effect persists even after the treatment period, possibly due to the fact that consumers can more easily recall their product choice in this regard using the NFP or manufacturer claims. The observed divergent effect of low fat versus low calorie labels highlights an important challenge with regards to promoting healthier food choices. While our results confirm perceived tradeoffs between taste and nutritional content reported in the literature (e.g. French et al., 1999, Yeomans et al., 2001 and Stubenitsky et al., 2000) for the low fat label, we do not observe a similar negative response to the low calorie label. This seems especially relevant since these two treatments exhibit a fairly large overlap of products. In general, treated products were significantly lower in sales as compared to the unlabeled products, potentially indicating taste preferences for high fat (high calorie) product alternatives in our product category. However, consumers seem to associate more favorable taste perceptions with the low calorie label than the low fat label. We also observe that overall category sales decreased as a result of our labeling interventions, suggesting that this substitution to healthier product alternatives was not offset by an overall increase in consumption. Labeling regulations under the Nutritional Labeling and Education Act have been implemented for over a decade, yet obesity rates keep rising. The FDA is currently considering a change to the format and content of food nutrition labels to promote increased label use. While our study looks at only one product category, it adds a market-based approach on how nutritional shelf labeling affects purchase decisions to the existing literature. Our reduced-form approach precludes us from providing welfare estimations, but results suggest that consumers may benefit from simple shelf or front package labels that focus on calorie content. A provision of these categorical statements instead or in addition to detailed quantitative statements could enable consumers to better incorporate nutritional information into their purchasing choices. A focus on calories also seems in alignment with policy objectives since calorie intake has been identified as the main contributor to weight gain and obesity. Focusing instead on fat content, as suggested by the World Health Organization, might trigger negative taste perceptions in some consumers and prevent them from making healthier food choices.