Bundling the Way to Bankruptcy: Economic Theory Should Inform the Design of Sugary-Drink Menus Used in Research

Profit-Maximizing Menus

Economics research has long been used to design menus that maximize profits. One strategy that businesses use for this purpose is nonlinear pricing (Wilson, 1993), a classic price-discrimination tool (Varian, 1989) from economics. In this strategy, price per ounce changes as a function of quantity purchased. John et al. (2017) claimed that they “priced drinks inexpensively (being careful to keep the pricing ratio between sizes roughly comparable to that used in U.S. fast-food restaurants)” (p. 622), but an examination of their menus does not seem to bear this out. Figure 1 shows the traditional and bundle menus used in our study (Wilson et al., 2013) and in the study by John et al. Only our menus used nonlinear pricing.

OPEN IN VIEWER

Fig. 1.

The traditional and bundle menus used in the studies by Wilson, Stolarz-Fantino, and Fantino (2013) and John, Donnelly, and Roberto (2017). Both menus in Wilson et al. incorporated the traditional nonlinear-pricing strategy used by restaurants in order to maximize profits. In contrast, the menus in John et al. had the same (extremely low) price per ounce for all size options. The bundle menu in Wilson et al. also emphasized the value of the larger item by allowing a direct comparison between one 16-oz soda for $1.59 (~$0.10 per ounce) and two 16-oz sodas for $1.99 (~$0.06 per ounce). No such comparison existed in the bundle menu John et al. used.

John et al. (2017) charged the same price per ounce for all drink sizes (i.e., $0.20/16 oz = $0.30/24 oz = $0.0125/oz), which did not add value to the larger-size option. Although this alone creates a problem for comparing their results with ours (Wilson et al., 2013), it creates a larger problem for generalizing their results to real-world scenarios. Menus constructed in this manner are unlikely to be profit maximizing because they ignore the law of diminishing marginal utility (Marshall, 1890). Because additional ounces added by larger sizes are valued less than the first ounces in the small size, people are unwilling to pay as much for them. This is one reason why nonlinear pricing is important for maximizing profits from drink sales. Consider a hypothetical example. Ignoring overhead and other expenses, suppose that soda costs a restaurant $0.01 per ounce. The restaurant sells 16-oz servings for $1.00, making $0.84 profit per drink. If the restaurant also offers 32-oz servings for $1.50, customers will perceive the lower price per ounce as a bargain and be tempted to buy the large drink. The restaurant nets $1.18 per large drink, which makes it more profit than the smaller portion. With nonlinear pricing, the restaurant creates the perception of a bargain and increases customers’ willingness to pay for the larger size. Although with linear pricing the restaurant would make more on each large-size drink sold, it would sell fewer large-size drinks and have lower profits.

Menu Appeal and Variety

In addition to not considering business profit in their menu design, John et al. (2017) created a bundle menu that was confusing and unappealing. The bundle option was presented as “Large: 2 × 12oz drinks = 24 ounces total. Price: $0.15 each (i.e., $0.30 total). Note: this offer is available as a bundle only (i.e., you cannot order only one 12oz drink)” (p. 8 in the Supplemental Material for John et al.). Even with the extra cognitive demand of deciphering the bundle option, this menu was ineffective at reducing caloric consumption. Thus, bundle menus cannot be ruled out as a factor that could undermine a ban on large-size sugary drinks. Moreover, although minimized by John et al., features that can inherently make bundle menus more appealing would likely be highlighted by businesses.

Bundle menus can allow people to have two different types of drink. Considerable research has demonstrated that people consume more when given different flavors than when given only one (Remick, Polivy, & Pliner, 2009). One reason consumption increases is sensory-specific satiety, wherein a flavor becomes less pleasant, relative to other flavors, after being consumed (Brondel et al., 2009). Bundles can allow people to switch to another drink to which they have not been satiated.

Drink variety is popular and widely available, as evidenced by the installation of Coca-Cola Freestyle machines throughout the world (Moye, 2016). These machines have a touch screen that allows access to more than 150 drink flavors. Pepsi has a similar machine touted as being able to create more than 1,000 beverage variations (PepsiCo, 2014). Businesses embrace variety to maximize profits; they do not prohibit “mix-and-matching” (p. 8 in the Supplemental Material for John et al., 2017), as did John et al.’s bundle menu. Oddly, John et al.’s free-refills instructions did not have the same explicit prohibition on getting a different refill flavor, and they concluded that “participants found this smaller drink with free refills appealing” (p. 626). Free-refills conditions ² also provided the only opportunity for participants to get a lower price per ounce for the “large” drink. A lower price per ounce and the prospect of variety do sound appealing.

Game Theory and Policy Considerations

For effective public-health regulations to be developed, researchers and policymakers must consider both consumption and business profit. A theoretical model unifying both can strengthen research and inform policy. One model that guided our conceptualization of sugary-drink regulation in Wilson et al. (2013) is a widely-used two-stage sequential game (e.g., Frank, 2006), which we apply to this policy issue in Figure 2. Policymakers move first and want to minimize calories consumed; businesses move second and want to maximize profit. Businesses can respond to regulation by creating menus that satisfy the letter of the regulation but not necessarily the intent or spirit.

OPEN IN VIEWER

Fig. 2.

Example of a simple two-stage sequential game tree in which policymakers move first and businesses move second. Hypothetical Calories consumed and business profits (i.e., revenue – cost) are shown for some of the possible menus that businesses can offer when policymakers regulate and when they do not regulate. For simplicity, Calories consumed and profit are presented as averages per customer rather than as totals. For example, when policymakers regulate, businesses would offer Regulated Menu 2 in preference to Regulated Menu 1, in order to obtain higher profit. Because businesses move second, a menu that does not maximize profit (e.g., Regulated Menu 1) is unlikely to be offered. When policymakers do not regulate, businesses can offer unregulated menus until forced to respond to regulation.

As we argued in Wilson et al. (2013), if businesses eliminated larger-sized drinks from their menus and offered only 16-oz drinks, consumption would likely decrease. For example, in Figure 2, Regulated Menu 1 results in fewer Calories consumed than does the unregulated menu (100 vs. 150 Cal). Regulated Menu 1, however, results in lower profit than Regulated Menu 2. Businesses would almost certainly create a menu that maximizes profits given the new constraints of the regulation, and this menu (e.g., Regulated Menu 2) might defeat the intent of the regulation. Our Supplemental Material available online shows how a game tree could be applied to John et al.’s (2017) experiments.

In designing experiments that test the potential effects of regulating menu offerings, researchers must be cognizant of the range of menus businesses can offer in response to a regulation. Even if a particular menu that complies with a regulation reduces purchasing, researchers cannot automatically conclude that the regulation would be effective without first establishing that this menu is likely to be profit maximizing given the constraints of the regulation. If this is not first established, then it is difficult to draw conclusions about the regulation’s effectiveness from the observation that the menu reduced purchasing. Did the menu reflect what businesses would do to make it appealing? For example, did it emphasize variety or highlight that bundles can allow customers to have a drink at lunch and take a smaller cup for later rather than having to carry a large cup with their remaining drink? Researchers need to address such questions in order for experiments to have real-world relevance. Making generalizations to the real world is especially challenging because businesses are creative in their efforts to maximize profit. Therefore, substantial forethought is needed in research design to create menus that real businesses would actually use.

Businesses will not offer a bundle menu like the one John et al. (2017) offered because it is unlikely to maximize profit. However, even if such a menu were offered, it would likely be ineffective at reducing caloric consumption, as the experimental data in John et al. suggest. In order to draw meaningful conclusions, researchers must keep in mind that regulation and business response is a two-stage game. Businesses go second, and they adjust to regulation in order to maximize their profit. Regulators may revise policies, but if adverse consequences are created, it may be hard to gain public support for subsequent corrections. Curbing the obesity epidemic is important; we suggest that researchers and policymakers fully think through this two-stage sequential game before the first move is made.

Supplemental Material