Demonstrations and Price Competition in New Product Release

Raphael Boleslavsky Department of Economics, University of Miami, Coral Gables, Florida 33146, [email protected] Christopher S. Cotton Department of Economics, Queen’s University, Kingston Ontario K7L 3N6, Canada [email protected] Haresh Gurnani School of Business, Center for Retail Innovation, Wake Forest University, Winston-Salem, North Carolina 27106, [email protected]

e incorporate product demonstrations into a game theoretic model of price competition. Demonstrations may Winclude product samples, trials, return policies, online review platforms, or any other means by which a firm allows consumers to learn about their value for a new product. In our model, demonstrations help individual consumers to learn whether they prefer an innovative product over an established alternative. The innovative firm controls demonstration informativeness. When the innovative firm commits to demonstration policies and there is flexibility in prices, the firm is best off offering fully informative demonstrations that divide the market and dampen price competition. In contrast, when a firm can adjust its demonstration strategy in response to prices, the firm prefers only partially informative demonstrations, designed to maximize its market share. Such a strategy can generate the monopoly profit for the innovative firm. We contrast the strategic role of demonstrations in our framework with the strategic role of capacity limits in models of judo economics, which also allow firms to divide a market and reduce competition. Keywords: judo economics; Bertrand competition; marketing strategy; product demonstrations; money-back guarantees; return policies; product reviews History: Received February 13, 2013; accepted January 6, 2016, by J. Miguel Villas-Boas, marketing. Published online in Articles in Advance July 14, 2016.

1. Introduction The informativeness of these opportunities, which When a new product is released, consumers face uncer- we call “demonstrations,” can vary: an in-store dis- tainty about how well the product will meet their play at Best Buy may merely display a video of game needs. Firms can offer free samples, in-store trials, play footage, or it may allow consumers to play their access to reviews and consumer reports, and other game of choice on the video game console, affecting opportunities for consumers to resolve some (or all) of the consumer’s ability to learn about the console’s their uncertainty before buying. Return policies and capabilities. Auto dealers typically choose the route money-back guarantees also enable consumers to learn and duration of test drives, which may limit a driver’s more about products before fully committing to their ability to learn about all aspects of the car’s perfor- purchases. mance. Trial software often offers only a limited set of Allowing consumers to learn about their values for features. In some other examples, producers of innova- a product is an important part of a ﬁrm’s marketing tive personal hygiene products, household cleaning strategy. Apple allows consumers hands-on interaction supplies, exercise equipment, and a variety of other with their products in the curated environment of their products provide money-back guarantees or extended

Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. stores. Other companies design displays and interactive trial periods, which resolve most or all of a consumer’s trials, either for their own stores or for retail chains. valuation uncertainty before the purchase decision is For example, Samsung and Microsoft sometimes staff final (Heiman et al. 2001). Thus, the degree of informa- their own “mini stores” inside of retailers such as Best tion conveyed to consumers before purchase depends Buy where consumers can try video game consoles, on the demonstration design, which is a choice variable phones, and computers. Similarly, wineries or other for a firm. food producers visit grocery stores to offer samples of Our analysis incorporates a firm’s strategic choice of their products. Automakers offer test drives. Software demonstration informativeness into a simple model of companies offer trial periods. price competition between a firm selling an innovative

product for which consumers have uncertain value, and adjust than prices. In this case, increases in demon- a firm selling an established alternative. Consumers stration informativeness decrease the market share know their values for the established product, but contested by the innovative firm (the market division they are uncertain whether the new product meets effect). This reduces the established firm’s incentive to their needs. If the innovation does meet their needs, defend its market share by setting a low price, thereby then consumers value the innovation more than the dampening subsequent price competition. Simulta- established product; if it does not meet their needs, then neously, the product differentiation effect increases the innovation is worthless. During a demonstration, the favorable consumers’ willingness to pay for the consumers privately observe signal realizations, which innovative product. Together, the dampening of price we refer to as their “impressions” of the innovative competition arising from the market division effect product. Consumers with “unfavorable” impressions and the product differentiation effect imply that the are confident that the new product does not meet their innovating firm’s profit is increasing in demonstration needs, and they will never purchase it. Those with informativeness. Consequently, the innovating firm favorable impressions believe that the new product prefers a maximally informative demonstration, under- is more likely to meet their needs, but they may not mining competitive pressure to the greatest possible be completely certain that this is the case. A more extent. informative demonstration offers greater opportunity When demonstrations are more flexible than prices for consumers to realize that the new product fails to (which we consider in Section4), firms first set prices, meet their needs: imagine a longer or less restrictive and then the innovating firm selects its demonstration product trial. design. This is consistent with settings where pricing Increases in demonstration informativeness affect strategies are centralized (made by upper management), demand for the innovative product in two ways. First, and store managers or sales staff have flexibility to with a more informative demonstration, fewer con- adjust demonstration policies (e.g., test drives, displays, sumers for whom the innovation is not appropriate or in-store interactions) at the point of sale. It is also draw favorable impressions. Therefore, a favorable consistent with situations in which contracts or repu- impression conveys “better news” about the new prod- tation concerns lead to sticky prices (e.g., Rotemberg uct: it reveals that the innovation is better adapted to 1982, Blinder 1994). Because prices have already been the consumer’s needs, increasing the favorable con- determined, the innovative firm chooses a demonstra- sumer’s willingness to pay and the differentiation tion to maximize its market share given these prices: between products. Second, demonstrations divide the the demonstration is designed to persuade consumers, market into two groups: the group with favorable rather than dampen price competition. In equilibrium, impressions is interested in the innovation, and the the innovative firm offers the least informative demon- group with unfavorable impressions is not interested. stration for which the favorable consumers prefer to A more informative demonstration increases the proba- buy its product. If the demonstration is less infor- bility that a consumer for whom the innovation is not mative, then a favorable impression does not convey appropriate draws an unfavorable realization, revealing enough good news to entice any consumers to purchase the mismatch between her needs and the product’s the innovation. If the demonstration is more informa- attributes. Consequently, with a more informative tive, then all favorable consumers buy the innovation, demonstration, a larger share of the market learns but too many potential customers draw unfavorable that the innovation is not for them. This increases the impressions, revealing that the product is not for them. established firm’s market share and decreases the share By reducing informativeness slightly, the innovating of the market contested by the innovator, which we firm reduces the mass of consumers with unfavorable refer to as the “market division effect.’’ The product impressions, increasing its market share. Therefore, if differentiation and market division effects interact with the firm selects its demonstration design in response to price competition to shape the incentives for strategic prices, equilibrium demonstrations convey some, but demonstration design. not all, relevant information to consumers. The strategic role of demonstrations depends on the The advantage of dividing the market ahead of price

Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. relative flexibility of prices and demonstration policies. competition has been explored by Gelman and Salop When prices are more flexible (which we consider (1983), introducing the notion of “judo economics.” in Section3), the firms adjust prices in response to These authors analyze sequential price competition the innovative firm’s demonstration policy. This is when a market entrant can commit to limit its produc- consistent with settings in which the innovative firms tion capacity. By doing so, the entrant ensures that commits to a satisfaction guarantee or return policy, some fraction of the market will only have access to or in which the development of in-store experiences, the incumbent’s product, dividing the market. Once the review platforms, or product trials requires significant entrant commits to its capacity and price, the incum- time, planning, or effort, making them more difficult to bent has two possible responses. The incumbent can Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS 3

either undercut the entrant, enticing all consumers to A significant literature considers strategic informa- buy its product, or it can accommodate the entrant by tion provision by a monopolist. Lewis and Sappington conceding the portion of the market that the entrant (1994), Schlee (1996), and Johnson and Myatt (2006) is contesting and extract the monopoly profit from consider a seller that allows buyers to acquire private the uncontested fraction. The smaller the entrant’s information about their value for an item prior to capacity, the higher the incumbent’s monopoly profit in purchase. Villas-Boas (2004) considers the interaction of the uncontested portion of the market. By limiting its informative advertising that communicates a product’s production capacity, the entrant makes accommodation existence with a monopolist’s choice of product line more attractive for the incumbent. offerings. Gill and Sgroi (2012) allow firms to conduct In our model, an unfavorable signal realization publicly observable product tests. Che (1996) considers reveals that the innovation does not match the con- the use of customer return policies by a monopolist sumer’s needs. As informativeness increases, con- seller when customers learn about their valuation after sumers who have low values for the innovation are purchase. Other papers analyze a monopolist’s incen- more likely to realize that they are not interested in tives to signal its private information about product purchasing it, reducing the innovator’s market share quality through observable actions, such as prices or (this is the market division effect). Thus, increasing product warranties. The signaling role of prices is demonstration informativeness affects firms’ market explored by Bagwell and Riordan (1991) and uninfor- shares in the same way that the capacity limits in mative advertising is explored by Milgrom and Roberts Gelman and Salop (1983). Therefore, like capacity lim- (1986) and Bagwell and Ramey (1988). Moorthy and its, commitment to a demonstration policy can be used Srinivasan (1995) and Grossman (1981) consider money- to dampen subsequent price competition. However, back guarantees and product warranties as signaling unlike capacity limits (which do not affect consumers’ instruments. Gardete (2013) considers a cheap talk valuations), increases in demonstration informativeness communication by a firm. In our analysis, valuation increase the expected valuations of favorable consumers uncertainty is only about consumer tastes or needs, through the product differentiation effect. Because and the firm does not have any private information of this additional benefit, when the demonstration about these attributes. Thus, in our model, information affects subsequent price competition, the innovating provision does not play a signaling role. firm selects a demonstration that is maximally informa- A number of papers consider the interaction of infor- tive. When demonstrations are selected after prices mation provision and other aspects of firm competition. are set, the differences between capacity limits and Moscarini and Ottaviani (2001) analyze price competi- demonstrations are even more pronounced. After prices tion between firms when buyers learn about their value are set, dividing the market by imposing capacity for a product prior to purchase. In contrast to our anal- limits is worthless. In contrast, because of the product ysis, the informativeness of product demonstrations is differentiation effect, designing a demonstration after exogenous, while in our model the informativeness of prices are set is a powerful tool for innovating firms to a product demonstration is strategically selected by expand their market share and increase profit, with the innovative firm. Iyer et al. (2005) consider a model significant consequences for the market’s equilibrium.1 of firm competition with targeted advertising (which informs consumers of product existence) and targeted 1 To further explore the interaction of capacity limits and demonstra- prices, showing that the ability to target advertising tion informativeness, the online appendix (available as supplemental to consumers is an important channel to soften price material at http://dx.doi.org/10.1287/mnsc.2016.2449) analyzes an competition. Meurer and Stahl (1994) analyze a related extension of the game in which the innovating firm can simul- model, in which firms send messages to consumers that taneously use both capacity limits (e.g., Gelman and Salop 1983) perfectly reveal which product the consumer prefers. and demonstrations as part of its strategy. Because an increase in demonstration informativeness generates both market division and Unlike our analysis, messages are always perfectly product differentiation effects, while limiting capacity generates informative, and therefore demonstration informative- only a market division effect, it may seem that increasing informa- ness is not a strategic instrument. Kuksov and Lin tiveness is an unambiguously more desirable strategic instrument. (2010) also consider information provision by two Indeed, when demonstrations are more flexible than prices, the

Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. competitive firms that differ in the quality of their prod- innovating firm never chooses to limit its capacity in equilibrium. When demonstrations are determined before prices, however, the ucts. In their framework, the high-quality firm has an innovating firm may choose to limit capacity in addition to providing incentive to provide information resolving uncertainty demonstrations. This happens because even a fully informative about product quality, and the low-quality firm may demonstration may not reduce the size of the contested market have an incentive to provide information resolving enough to avoid an aggressive price response from the established consumer uncertainty about their preferences over firm. In this case (when the innovation is widely appealing but provides low added value) limited capacity and demonstration quality. The distinguishing feature of our framework is informativeness are complementary instruments for dampening that we allow the innovative firm to not only choose price competition. whether to provide demonstrations, but to also choose Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition 4 Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS

how informative to make their demonstrations.2 We in-store demonstrations and samples, satisfaction guar- also consider how the timing of demonstration design antees, return periods, review platforms, and consumer (whether it is chosen before or after prices) affects the reports). role of demonstrations in new product release. To our Formally, a demonstration is modeled as a binary knowledge, this question is novel to the literature. random variable from which consumers draw either a “favorable” or “unfavorable” realization, correspond- ing to their “impression” of product . A consumer 2. Preliminaries who has a high value for product always draws a favorable impression. A consumer with a low value 2.1. Model for product , draws an unfavorable impression with We model market competition between two firms: firm probability d ∈ 601 17 and draws a favorable realization offering an established product and firm offering with probability 1 − d. Variable d therefore represents an alternative, innovative product for which consumers the demonstration’s informativeness. Given d, the dis- are uncertain about their valuations. A continuum of tribution of a consumer’s posterior expected valuation consumers exists, normalized to a mass of one. Each generated by the demonstration is given by â: consumer shares a common value v = 1 for firm ’s  established product. Consumers and firms are uncertain 0 with probability 41 − 5d1  about each consumer’s value for firm ’s product. This â = value can be either high with v = > 1, or low with  with probability 1 − 41 − 5d0  1 − 41 − 5d v = 0. It is common knowledge that an individual consumer independently draws a high value with A consumer with an unfavorable impression is certain probability ∈ 401 15. Thus represents the fraction that he has a low value for the innovative product. A of consumers for which product is a good match. consumer with a favorable impression, however, is Parameters and capture different aspects of the generally left with some uncertainty about whether innovation’s demand: is a measure of the product’s she has a high or low value. horizontal quality (or taste), reflecting the size of the This class of demonstration is most appropriate market that finds it appealing (e.g., d’Aspremont et al. for innovative products with a number of possible 1979), and is a measure of vertical quality or added “deal-breaking” attributes or features. A low-valuation value, reflecting the intensity of preference among the consumer does not like one of the “deal breakers” and consumers who find the innovation appealing (e.g., is unwilling to purchase the innovation if this critical 3 Shaked and Sutton 1982). attribute of the product is encountered. Meanwhile, Price competition takes a simple form: the firms a high-valuation consumer likes the attributes of the product and could never encounter a deal-breaking simultaneously post prices p and p for their respective products. Consumers take firms’ prices as given when product attribute. The more consumers interact with the product, and the fewer restrictions placed on their deciding whether to purchase product , product , interaction, the more likely a low-valuation consumer or neither product. When a consumer purchases a encounters a deal-breaking attribute. Hence, if a con- product of value V at price p, her payoff is u = V − p. sumer experiences a demonstration with significant If the consumer does not make a purchase, then her freedom and does not encounter a deal-breaking fea- payoff is 0. Each consumer has unit demand, and it is ture, the consumer rationally infers that he or she is only feasible for a consumer to purchase one of the more likely to have a high valuation for the innovation. products. Thus, high values of d in the demonstration design 2.1.1. Product Demonstrations and Their Effects. represent prepurchase interactions with significant The innovating firm provides consumers with an information content: long return periods, exhaustive opportunity to learn about their values for the innova- money-back guarantees, or extensive in-store or at- tive product before finalizing their purchase decision. home trials. Conversely, low values of d represent These “demonstrations,” encompass a variety of prac- prepurchase interactions with less information; an tices that facilitate learning (e.g., prepurchase trials, in-store video of game play footage is less informative Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. about a video game than an in-store trial, which in turn is less informative than an at-home trial over 2 In this way, our analysis is also related to the emerging literature an extended period (Heiman et al. 2001, Heiman and that considers the strategic design of an informative signal by a Muller 1996, Davis et al. 1995). “sender” who wishes to influence the actions of a “receiver” who observes the signal’s realization. Kamenica and Gentzkow (2011) and Rayo and Segal (2010) consider strategic signal design by a 3 Deal-breaking attributes are often encountered in new product single sender and receiver. Boleslavsky and Cotton (2014, 2015) releases. When the iPhone was released, for example, some Blackberry model signal design in an environment in which two senders try to users refused to switch to the iPhone merely because they did not influence a single receiver. like the experience of its virtual keyboard. Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS 5

An increase in demonstration informativeness after both firms commit to prices. This is the case when changes the distribution of consumer valuations in two firms are reluctant (for reasons that we do not model) ways. Let 4d5 denote the portion of consumers with to adjust prices too often (e.g., Rotemberg 1982, Blinder favorable impressions: 1994), or when prices are set by manufacturers and demonstrations are chosen at the point of sale. 4d5 ≡ + 41 − 541 − d5 = 1 − 41 − 5d0 Timing II. When the firm has flexibility to choose a demonstration policy after prices are established, 0 Note that 4d5 is decreasing in d (i.e., 4d5 = −41 − 5 the game takes place as follows: (1) Price competition. < 0). This is the “market division effect’’: as informa- (2) Choice of demonstration policy. (3) Demonstration tiveness increases, a consumer with a low value for experience. (4) Purchase. the innovation is more likely to draw an unfavorable It is important to recognize that the only difference impression during the demonstration, ensuring that she between models is the sequencing of price competi- will not buy it. Consequently, when the demonstration tion and demonstration design. In both models, the divides the market into favorable and unfavorable demonstration design and pricing stages are identical. groups, an increase in informativeness leaves fewer We solve for the perfect Bayesian equilibrium of the consumers in the favorable group, shrinking the por- game by backward induction, starting with the con- tion of the market that the innovating firm contests. sumers’ purchase decision (which is the same across Let 4d5 denote the expected valuation of a consumer both timing regimes). with a favorable impression: 2.2. Consumer Purchase Decision 4d5 ≡ = 0 In the final stage of the game, each consumer i ∈ 1 − 41 − 5d 4d5 601 17 makes a purchase decision. Before doing so, Note that 4d5 is strictly increasing in demonstration she observes the demonstration design d and either a informativeness (i.e., 04d5 = 41−5/4d52 > 0). This is favorable or unfavorable impression of the innovative the “product differentiation effect’’: as informativeness product. Let i denote consumer i’s expected value increases, consumers with favorable impressions are for product after experiencing a demonstration: more convinced that the product will satisfy their i is consumer i’s realization of â (equal to 0 if the needs. These two effects shape the incentives for price impression was unfavorable and 4d5 if the impression competition and demonstration informativeness, but was favorable). the role each plays depends on the model timing. Consumer i’s expected payoff from purchasing good is ui45 = i − p, and the consumer’s payoff 2.1.2. Timing. We analyze the game with two from purchasing good is ui45 = 1 − p. If the consequences of moves. First, we consider the possibil- sumer purchases neither product, the payoff is 0. It is ity that firm selects demonstration informativeness sequentially rational for the consumer to purchase the before prices are established. This corresponds to an product that offers the higher expected payoff, pro- environment in which prices are more flexible than vided that this expected payoff is positive. Consumer i demonstrations. This is the case when firms can quickly therefore purchases product if and easily change their prices (with online pricing, e.g., Gorodnichenko et al. 2014), or when the innovating i − p ≥ 1 − p and i − p ≥ 0 firm issues a blanket commitment to a return period or money-back guarantee. and purchases product if Timing I. When price competition follows demon-

strations, the game takes place as follows: i − p < 1 − p and 1 − p ≥ 0. (1) Choice of demonstration policy2 Firm chooses a demonstration policy d ∈ 601 17. By setting p > 1, ﬁrm is guaranteed never to make a (2) Price competition2 The two ﬁrms simultaneously sale. These prices are therefore (weakly) dominated by

set prices p and p. p = 1. We focus on equilibria in which ﬁrm does not (3) Demonstration experience2 Consumers interact with choose a weakly dominated strategy: in equilibrium Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. the product receiving a favorable or unfavorable impres- p ≤ 1. This immediately implies that we can ignore the sion. They update their beliefs about their valuations case in which the consumer purchases neither product, according to Bayes’ rule, accounting for both demon- as purchasing is better than purchasing nothing. stration informativeness and their realized impression. Hence, consumer i purchases product whenever

(4) Purchase2 Each consumer decides whether to u ≥ u and otherwise purchases product . Therefore, purchase product , product , or neither product. a consumer’s purchase decision is determined by a Second, we consider the possibility that the inno- single threshold for her posterior belief: she purchases vating firm retains flexibility over its choice of d until firm ’s product whenever she is sufficiently convinced Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition 6 Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS

that her valuation for the innovation is likely to be the established ﬁrm sets the highest price for which it high. Let captures the entire market. This type of equilibrium

4p¯ 1 p5 ≡ 1 − p + p can arise only when d is sufficiently small: 4d5 is increasing in d, while 4d5 is decreasing; moreover, denote the critical threshold in the posterior belief. 415 = > 1 > 415 = . Intuitively, because the innova- Therefore, in equilibrium, a consumer with a favorable tive product’s value exceeds the established product’s, impression of product purchases it if and only if with a sufficiently informative demonstration and low 4d5 ≥ 4p¯ 1 p 50 price, the innovating firm can always capture some of the market in equilibrium. Otherwise the consumer purchases product . We have When 4d5 > 4d5, there is no pure strategy equi- assumed that if the consumer is indifferent between librium, as one firm would always want to adjust its products, then the consumer purchases product . This price in response to the price set by the other firm.5 In assumption is without loss of generality, regardless of the mixed strategy equilibrium, the established firm the model timing. randomizes between its two types of best responses, sometimes setting price p = 1 to extract the maximum profit from those consumers that have an unfavorable 3. Demonstration Design Before impression of the innovation, and sometimes discount- Price Competition ing its price in an attempt to undercut the other firm

In this section, we solve the model for the case where and capture the entire market, drawing p from a the innovating firm chooses a demonstration strategy continuous distribution supported on 61 − 4d51 17. before the firms announce prices. This is consistent with The innovator randomizes over a range of prices that a firm’s long-standing commitment to a satisfaction prevents the established firm from always undercutting

guarantee or return policy, and it is appropriate for and capturing the market, choosing p from a con- settings in which firms can adjust prices more easily tinuous distribution supported on 64d5 − 4d51 4d57 than demonstrations. For example, if developing an with no mass points. Both firms’ mixing densities are in-store experience or product trial requires significant explicitly characterized in the appendix. time, planning, and effort, then changing a price is simple by comparison. Proposition 1. Suppose that the demonstration is determined before prices. 3.1. Price Competition • When 4d5 < 4d5, the equilibrium of the pricing In the appendix, we derive the equilibrium strategies subgame is p = 1 − 4d5 and p = 0. Firm sells to the in the pricing subgame for any choice of d. Here, we entire market. Profits are = p = 1 − 4d5 and = 0. describe the intuition and the results. • When 4d5 > 4d5, the equilibrium of the pricing To understand the strategic forces underlying the subgame is in mixed strategies. Firm ’s price is drawn from pricing stage, suppose that demonstration informa- a continuous random variable supported on 61 − 4d51 17 tiveness, d, has been set, and consider firm ’s best and a mass point on 1. Firm ’s price is drawn from a continuous random variable on 64d5 − 4d51 4d57 with response to p. Two types of strategies can be best responses. (1) Firm can either target only the share of no mass points. Expected profits are = 1 − 4d5 and 2 the market with unfavorable impressions of the inno- = 4d544d5 − 4d55 = − 44d55 . vation by setting p = 1 and generating profit 1 − 4d5, Although some of the effects in our pricing stage are or (2) it can offer consumers a slightly higher payoff reminiscent of Gelman and Salop (1983), the simulta- than firm , capturing the entire market, resulting in neous price setting in our model introduces crucial − + 4 profit (arbitrarily close to) 1 4d5 p. differences. When the innovator prices first (as in To make positive profit, firm must avoid being Gelman and Salop 1983), it anticipates the established priced out of the market. However, this is not always firm’s response, and it can always select a price for possible: sometimes even if firm prices as aggres- which p = 1 is a best response, generating a pure sively as possible (setting p = 0), firm still prefers to go for the entire market (setting p = 1 − 4d5). Com- Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. 5 p < 4d5 − 4d5 paring ’s profits reveals that this equilibrium exists If the innovator prices aggressively ( ), then the established firm prefers to focus on the portion of the market that whenever 4d5 < 4d5. Therefore, whenever demonstra- dislikes the innovation, setting a price of p = 1. If it does so, then tion informativeness or added value are sufficiently the established product offers 0 payoff to consumers, and there low, the equilibrium is similar to asymmetric Bertrand is therefore no reason for the innovator to price aggressively (it = competition: the innovating firm sets a price of 0, and would set price p 4d5 instead). Meanwhile, if the innovator sets a relatively high price (p > 4d5 − 4d5), then the established firm would have an incentive to marginally undercut, capturing the entire 4 To capture the entire market, firm must set a price for which market. However, then the innovator too would respond with a

1 − p > 4d5 − p or, equivalently, p < 1 − 4d5 + p. marginal price cut. Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS 7

strategy equilibrium. If the established firm cannot firm prefers d = 1, generating profit 1 − . However, observe the innovator’s price, then both firms must act when 4d5 < 4d5, firm ’s profit, 1−4d5, is decreasing unpredictably to avoid exploitation. in demonstration informativeness. Hence, for values of d such that 4d5 < 4d5, the established firm prefers 3.2. Demonstration Informativeness d = 0, generating profit 1−. Because > 1, comparing An increase in demonstration informativeness gener- these profits reveals the following corollary. ates product differentiation and decreases the size of the contested market share, both of which dampen Corollary 3. When demonstrations are established subsequent price competition. Indeed, as demonstration before prices, firm ’s profit is highest when firm chooses informativeness increases, in equilibrium both firms a fully informative demonstration. are more likely to set higher prices. Because the market division effect dampens competi- Corollary 1. When demonstrations are established tion to the greatest possible extent and increases the before prices, an increase in d generates a first-order stochas- prices that both firms charge in equilibrium, firm also generates the highest profit that it can when firm tic dominance shift toward higher prices in each firm’s 6 equilibrium mixed strategy. selects a fully revealing demonstration. Although the share of the market that has a favorable view of the innovation (’s maximum market share) 4. Flexibility in Demonstration Design In this section, we consider the case where firm has decreases with demonstration informativeness, the loss the flexibility to adjust its demonstration policy after of market share is offset by an increase in the price. prices are observed. This is consistent with settings in Interestingly, the price effect dominates. which retail managers, sales staff, or dealers have the Corollary 2. When demonstrations are established ability to adjust demonstrations to maximize sales. It is before prices, firm ’s profit is weakly increasing in infor- also consistent with the firms committing to prices up mativeness and strictly increasing when informativeness front, by establishing a policy to not discount items at passes a threshold. In equilibrium, firm chooses a fully the point of sale.7 There is ample evidence that prices informative demonstration. tend to be sticky, with firms reluctant to change their prices too often (e.g., Rotemberg 1982, Blinder 1994), This result is driven by the product differentiation suggesting that this timing is often reasonable. effect of informative demonstrations, which allows the innovator to set higher prices for the group of 4.1. Demonstration Informativeness favorable consumers, offsetting the lost market share When demonstration informativeness is chosen after from the larger group of consumers who view the prices, its strategic role is significantly different. Here, product negatively. the demonstration responds to the prevailing market To highlight the importance of the product differenti- prices and it cannot be used to soften price competition, ation effect in generating a profit function that increases because prices have already been set. Instead, the with informativeness, we briefly consider a benchmark innovating firm adjusts the informativeness of its version of our model in which the product differentia- demonstration in order to maximize its market share: tion effect is artificially turned off. Specifically, suppose the demonstration is designed to persuade consumers that the willingness of favorable consumers to pay ∗ to buy the innovation, not to dampen competition. is fixed at rather than increasing in d. The charac- In the previous section, both the market division terization of the pricing equilibrium in Proposition1 and product differentiation effects reduce competi- also applies here, with ∗ replacing 4d5. Hence, when ∗ tive pressure and increase profits. In this section, in 4d5 < , firm ’s expected profit in the benchmark contrast, the product differentiation effect increases ∗ = ∗ − ∗ is 4d54 4d55. If < 2, then this function is the innovating firm’s profit, but the market division nonmonotonic in d for d ∈ 601 17, and indeed, when ∗ < 2, it is decreasing over this domain. Therefore, the 6 product differentiation effect is an essential component The established firm prefers to select a fully informative demonstration and accommodate the innovator, rather than to select a

Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. in these results. less informative demonstration, which reduces consumer value for When prices respond to demonstrations, an interest- product , but also leads to more fierce price competition. This is ing alignment of interest arises between the competing reminiscent of the “puppy dog ploy” described by Fudenberg and firms: both firms’ profits are highest when firm Tirole (1984). chooses a fully informative demonstration. To see this 7 For example, Canada Goose never discounts its jackets and does alignment, note that whenever 4d5 > 4d5, firm ’s not allow any authorized retailer to do so either. Similarly, Apple − and other electronics manufacturers rarely or never discount their equilibrium profit (1 4d5) is also increasing in demon- current generation of productions. Luxury clothing and accessory stration informativeness. Therefore, for values of d such makers such as UGG, Hermés Birkin, and Louis Vuitton have similar that 4d5 > 4d5 (which always exist), the established reputations. Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition 8 Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS

effect reduces it. With these effects working against than needed to persuade consumers with favorable each other, the optimal demonstration is no longer impressions to purchase its product. Consequently, fully informative, but rather maintains some consumer the consumers that purchase product are indifferent uncertainty about their values. between the two products. To explore this difference in more detail, suppose that Proposition 2. When demonstrations respond to prices, with an uninformative demonstration, the innovation the innovative firm uses demonstrations to maximize its offers consumers a higher expected payoff than the market share given prices. When feasible, it prefers the established product. In this case, increasing informa- least informative demonstration policy such that favorable tiveness only reduces ’s market share, because doing consumers (weakly) prefer the innovative product. so increases the mass of consumers with unfavorable • If prices are sufficiently favorable for firm (i.e., impressions and no interest in product . In this case, p − p < 1 − ), then firm always captures the entire an uninformative demonstration is optimal.8 However, market, regardless of demonstration informativeness. if the innovation offers a smaller expected payoff than • If prices are sufficiently favorable for firm (i.e., the established product, when the demonstration is 1 − ≤ p − p ), then firm captures the entire market in uninformative, then increasing informativeness can be equilibrium. beneficial by generating product differentiation. Indeed, • For intermediate levels of price differences (i.e., 1 − ≤ increasing informativeness increases the valuation of p − p < 1 − ), firm chooses a partially informative consumers with favorable impressions, and firm demonstration with d ∈ 401 15, firm sells to consumers prefers demonstration that is informative enough to who receive an unfavorable impression of product , and convince those with favorable impressions to buy its firm sells to consumers who receive a favorable impression. product. However, increasing informativeness also Therefore, causes market division, which reduces the fraction of favorable consumers and the firm’s market share. = − = Therefore, when demonstrations follow prices, the 1 p and p. 1 − p + p 1 − p + p firm prefers an intermediate level of demonstration, maximizing the share of consumers willing to buy its The expression for the equilibrium d∗ is given in product at the given prices. Equation (4) in the appendix. When chosen prior to

Formally, for given product prices p and p, firm ’s price competition, demonstrations play a strategic role optimal choice of d solves of dividing the market and minimizing competition between the firms when setting prices. When chosen max 4d5 s.t. 4d5 − p ≥ 1 − p. after price competition, demonstrations persuade con- d∈601 17 sumers to purchase product . Clearly, the product Firm chooses d to maximize the mass of consumers differentiation effect is essential for demonstrations to who receive a favorable impression of its product, have any value for the innovating firm after prices are subject to the constraint that the demonstration is set; instruments that only generate market segmenta- sufficiently informative that consumers with a favorable tion (like capacity limits) are not beneficial after prices impression prefer the innovation. As demonstrations are set. become more informative, a larger portion of consumers have unfavorable impressions of the product as they 4.2. Price Competition realize that the product does not meet their needs. When choosing prices, the firms anticipate how their This decreases firm ’s market share at the given choices influence the subsequent design of demon- price. Therefore, firm prefers as low of d as possible strations and its impact on their market share. When while meeting the constraint. The expected value of deriving equilibrium pricing strategies, we focus on product , given a favorable impression, is increasing the case where in d. Thus, firm prefers the minimum d such that > 40 (A1)

4d5 − p ≥ 1 − p. This assumption can be viewed in one of two ways: When firm ’s price advantage is sufficiently large, (1) The product is a “breakthrough,” offering a large = Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. consumers will purchase its product even if d 0. In added value over the existing product ( is big). (2) The that case, it provides uninformative (or no) demonstra- product is a niche product, appealing to a relatively tions and captures the entire market. When neither small portion of the market ( is low). The assumption firm has a sufficiently large price advantage, firm is mainly for tractability, allowing us to characterize implements a partially informative demonstration. In the equilibrium in closed form. The detailed analysis in this case, its demonstration is no more informative the online appendix establishes that (A1) is necessary and sufficient for the existence of a pure strategy 8 With an uninformative demonstration, all impressions are favorable, equilibrium in the pricing game, characterized in the i.e., 405 = 1. following proposition. Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS 9

Proposition 3. When demonstrations respond to market is extracted by the ﬁrms despite competition. prices: Second, the expected total surplus in the market is • A pure strategy equilibrium of the pricing stage exists determined by the match between the consumers and if and only if (A1). the products. Because demonstrations are only partially

• In any pure strategy equilibrium, firm sets p = 1, revealing, some consumers purchase the innovation and firm selects any price p ∈ 6pL1 pH 7 ⊂ 61 7. although they would be better off purchasing the estab- • On the equilibrium path, the innovative firm chooses lished alternative, generating inefficient matches in a partially informative demonstration, those with favor- equilibrium. As such, total surplus would be higher able impressions purchase the innovation, and those with under a system of fully informative demonstrations. unfavorable impressions purchase the established product. That the innovator is able to obtain the monopoly

• Firm expects the monopoly profit = , consumer profit has interesting implications for innovation policy, surplus is 0, and the established firm’s profit is smaller the goal of which is to ensure that innovators receive than the monopoly profit in the uncontested fraction of the enough compensation for their innovations that they

market, < 1 − . allocate sufficient resources toward developing new products. A major concern is that innovations are pro- To understand the structure of this equilibrium, note tected from imitators, a concern that we do not address. that when p = 1, firm is indifferent between all However, another important issue is that the value of prices inside 61 7, which deliver the innovating firm developing an innovation may be diluted by competi- the monopoly profit (see Proposition2). Intuitively tion from inferior products.10 In our analysis, when the p = we see that when 1, the established product offers innovating firm has developed a breakthrough (so that consumers 0 payoff, and it is therefore not really 9 (A1) holds) and it can design product demonstrations competing with the innovator. Thus, if some p inside in response to prices, this concern does not arise. The = this interval could be found for which p 1 is a best innovating firm captures the monopoly profit, despite response, then these would constitute an equilibrium the presence of an inferior established alternative, and of the pricing stage. Firm considers two types of therefore, its incentives to innovate are not reduced by = deviations from p 1: large price cuts to capture the competition from the inferior product. entire market (independent of firm ’s subsequent Finally, the analysis suggests that a firm may be demonstration strategy), or a smaller price cut that better off retaining flexibility in its demonstration incentivizes to select a more informative signal. As policy, adjusting its consumer information strategy to we argue in the online appendix, precluding these account for price differences between the products. This deviations imposes bounds on feasible prices p (the set highlights a way in which firms in innovative industries of prices satisfying these conditions is nonempty when may benefit from a reputation for not changing prices. (A1) is satisfied). The calculations are technical, but the It also suggests that consumer surplus and total surplus effects that generate the bounds are intuitive. When p may be higher in industries where prices are less sticky, is low, firm ’s profit per unit sold is low, and the firm which (as we showed in Section3) leads firms to adopt has an incentive to increase its price, forfeiting some more informative demonstrations. market share, but increasing overall profit. When p is too high, firm does not need to drop its price 5. Conclusion much from p = 1 to capture the entire market and will choose to do so. Therefore, an intermediate value We consider strategic information provision in a model of price competition. Our model of demonstrations of p is needed for p = 1 to be a best response and for firm not to prefer a price increase. may represent product trials, samples, return poli- When demonstrations are more flexible than prices, cies, review platforms, or any other means by which the innovating firm selects a partially revealing demon- firms give consumers exposure to products before stration designed to persuade the maximum number of the consumers commit to purchase decision. A firm consumers to purchase its product. This has significant releases an innovative product, which may benefit normative implications. First, the equilibrium demon- only some consumers. By providing demonstrations, stration leaves consumers (nearly) indifferent between the firm gives consumers an opportunity to better Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. purchasing either product, and the firms extract all learn about their own value for the innovation. More surplus as profits: the entire consumer surplus in the information simultaneously increases the expected valuation of those who receive favorable impressions of the new product (the product differentiation effect), 9 Given that the established product offers 0 expected payoff, the innovator can achieve the monopoly profit by charging any price 10 p ∈ 61 7 and choosing demonstration informativeness such that This effect arises when demonstrations are determined before 4d5 = p. In expectation, the probability of generating a favorable prices, for example, where the innovator’s equilibrium payoff is impression and the increase in the price exactly offset each other to − 2, which is less than the ex ante social surplus generated by generate a constant expected profit. the innovation, . Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition 10 Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS

while also decreasing the share of consumers with or establishing online review platforms to help poten- favorable impressions (the market division effect). tial consumers learn from the experiences of others. Depending on whether its demonstration policy is This is broadly consistent with the behavior of Ama- chosen before or after prices are set, the innovating zon.com, for example: online prices are easily adjusted, firm either designs its demonstration policy to reduce and the company has established a generous return subsequent price competition or to persuade consumers policy and extensive review system to help consumers to purchase its product given the prevailing prices. learn about new products (e.g., Lecher 2015). When prices respond to the demonstration policy, the firm prefers to make its demonstrations as informative Supplemental Material as possible, generating the greatest amount of product Supplemental material to this paper is available at http://dx differentiation and reducing the contested portion of the .doi.org/10.1287/mnsc.2016.2449. market as much as possible, minimizing the intensity of competition in the pricing stage. In contrast, when Acknowledgments the firm adjusts its demonstration policy in response to The paper greatly benefited from valuable comments made by prices, the product differentiation effect can increase department editor J. Miguel Villas-Boas, the associate editor, and two anonymous referees, as well as Heski Bar-Isaac, Justin demand for the innovation, while the market division Johnson, Harish Krishnan, Pino Lopomo, Mahesh Nagarajan, effect reduces it. Consequently, the innovating firm and Curt Taylor. C. S. Cotton is grateful for financial support prefers only a partially informative demonstration, provided through his position as the Jarislowsky-Deutsch designed to maximize its market share. In this case, Chair in Economic and Financial Policy at Queen’s University. the ability to offer demonstrations can lead to the H. Gurnani’s research is partially supported by the Benson- innovating firm collecting the monopoly profit. Pruitt endowed professor position at Wake Forest University. This result implies that an innovating firm may The research started while he was Leslie O. Barnes Professor benefit from the ability to adjust its demonstration at the University of Miami. strategy in response to market prices. If feasible, a firm Mathematical Appendix may prefer to commit to prices, while allowing retailers In addition to this appendix, an online appendix provides a or sales agents to choose in-store demonstrations, more detailed analysis of the flexible-demonstrations envi- samples, or test drives to maximize sales, once prices ronment and considers a version of the model in which the are set. This suggests that the returns to innovation firm can simultaneously use demonstrations and capacity may be higher in industries in which manufacturers constraints in an effort to limit competition. control prices, or prices are otherwise sticky and slow to change (e.g., Rotemberg 1982, Blinder 1994). Such A.1. Upfront Demonstration Design strategies are generally consistent with automobile Derivation of equilibrium of the pricing subgame. When choosing prices following the choice of d, firm ’s best response manufacturers giving dealers relatively little flexibility to p is either p = 1, which earns profits = 1 − 4d5, or in adjusting prices, and a lot of flexibility in the design p = 1 − 4d5 + p (or “just under” this value when 4d5 > 1), of test drives and showroom experiences (e.g., Cato which results in capturing the entire market and earning

2014), and Apple, which is known for adjusting its = 1 − 4d5 + p. The problem is similar to asymmetric prices on its iPhone, iPad, and Mac computers very Bertrand price competition, except that when p is low infrequently (often only once per year), and which enough, firm prefers to avoid competition all together, set establishes its own stores where it controls many details p = 1, and focus on its role as a monopolist provider to those of the consumer shopping experience (e.g., Kane and with unfavorable impressions of the innovative product. ≤ Sherr 2011). Case 1. Suppose 4d5 4d5. Then the equilibrium involves firm setting p = 1 − 4d5, and firm setting p = 0. Firm In settings with a higher degree of price flexibility, an captures the entire market, earning = 1 − 4d5. Given that innovative firm is better off providing more-informative 4d5 ≤ 4d5, this payoff is at least as large as the firm’s profit demonstrations. Although demonstrations may lead from setting p = 1 and earning 1 − 4d5. some consumers to realize a new product is not for Case 2. Suppose 4d5 > 4d5. In this case, there is no pure them, this potential loss in market share is offset by an strategy equilibrium. Consider the possibility of a mixed increase in the willingness to pay off those with favor- strategy Nash equilibrium in which Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. able impressions, and by a decrease in the incentives • firm mixes using a smooth continuous distribution min that competitors have to cut their prices in response to over a continuum 4p 1 15 according to F and a mass point = the entry of the new product. In such settings, firms on p 1 with weight ; and • firm mixes using a smooth continuous distribution should help potential customers to learn as much as over a continuum 6pmin1 4d55 according to F . possible about their values. Where feasible, this may When firm sets p = 1, doing so results in = 1 − 4d5. involve offering generous return policies, satisfaction Thus, any other strategy played with positive probability

guarantees, or unrestricted product trials. It may also by the mixed strategy must also give = 1 − 4d5. The involve providing professional reviewers with samples, minimum p that returns such a proﬁt is p = 1 − 4d5, and Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS 11

only when setting such a price leads to a market share A.2. Flexible Demonstrations of 1 for firm . This implies a lower bound for ’s mixing Firm chooses a demonstration that is just informative distribution, since 1 − p = 1 − 41 − 4d55 = 4d5 must be enough that those with favorable realizations buy its product. higher than 4d5 − p for all p. Thus, p > 4d5 − 4d5. (This Doing so maximizes the number of consumers with suffi- could be negative, a possibility we rule out later.) ciently favorable impressions to purchase the product. Then, min min This implies that p = 1 − 4d5 and p = 4d5 − 4d5. firm ’s best response demonstration to prices p and p is In turn, this implies that firm can achieve a profit of = − ∗ 1 − p + p − 44d5 4d554d5 from setting a price at this lower d = 1 (4) bound, and thus the profits from other prices in the mixing 41 − p + p541 − 5

distribution must equal this amount. ∗ ∗ when d > 0. When 1 − p ≤ − p, it follows that d ≤ 0, Firm must be indifferent between all p ∈ 41 − 4d51 17. and the preferred demonstration involves d = 0. When d∗ > 1, An arbitrary p in this range results in capturing the entire there does not exist a feasible demonstration policy that leads market if 1 − p > 4d5 − p, which is true if p > 4d5 − 1 + p. Thus, to ﬁrm selling to any share of the market. Detailed derivation of the upfront pricing strategies, as well 4p5 = F44d5 − 1 + p541 − 4d55p as the necessary and sufﬁcient conditions for the existence of pure strategy equilibria, are included in the online appendix. + 41 − F44d5 − 1 + p55p

= p − F44d5 − 1 + p54d5p0 This has to equal the equilibrium payoffs 1 − 4d5. Thus, References setting the above expression equal to 1 − 4d5 and solving for Bagwell K, Ramey G (1988) Advertising and limit pricing. RAND J. Econom. 19(1):59–71. F44d5 − 1 + p5 gives Bagwell K, Riordan MH (1991) High and declining prices signal p − 41 − 4d55 product quality. Amer. Econom. Rev. 81(1):224–239. F 44d5 − + p 5 = 1 1 Blinder AS (1994) On sticky prices: Academic theories meet the real 4d5p world. Mankiw NG, ed. Monetary Policy (NBER and University which implies a distribution of p such that of Chicago Press, Chicago), 117–154. Boleslavsky R, Cotton C (2014) Limited capacity in project selection: p − 44d5 − 4d55 F4p5 = 0 (1) Competition through evidence production. Working paper, 4d54p − 44d5 − 155 University of Miami, Coral Gables, FL. Note that if p = 4d5 − 4d5 then F 44d5 − 4d55 = 0, and if Boleslavsky R, Cotton C (2015) Grading standards and education quality. Amer. Econom. J. Microeconom. 7(2):248–279. = = p 4d5 then F44d55 1. Cato J (2014) Car dealers profit—but not as much as you’d think. Similarly, firm must be indifferent between all p ∈ Globe and Mail (January 4), http://www.theglobeandmail.com/ 64d5 − 4d51 4d55. An arbitrary p in this range results in globe-drive/news/cato-car-dealers-profit—but-not-as-much -as-youd-think/article16188656/. = − − + 4p5 41 F41 4d5 p554d5p1 Che Y-K (1996) Customer return policies for experience goods. which must equal payoffs 44d5 − 4d554d5. Setting the J. Indust. Econom. 44(1):17–24. expression for 4p 5 equal to 44d5 − 4d554d5 and solving d’Aspremont C, Gabszewicz JJ, Thisse J-F (1979) On Hotelling’s “stability in competition.” Econometrica 47(5):1145–1150. − + the implied equality for F41 4d5 p5 gives Davis S, Gerstner E, Hagerty M (1995) Money-back guarantees in 4d5 − 4d5 retailing: Matching products to consumer tastes. J. Retailing F41 − 4d5 + p5 = 1 − 0 71(1):7–22. p Fudenberg D, Tirole J (1984) The fat-cat effect, the puppy-dog ploy, Thus, and the lean and hungry look. Amer. Econom. Rev. 74(2):361–366. p − 41 − 4d55 Gardete PM (2013) Cheap-talk advertising and misrepresentation in F 4p 5 = 0 (2) 4d5 − 1 + p vertically differentiated markets. Marketing Sci. 32(4):609–621. Gelman JR, Salop SC (1983) Judo economics: Capacity limitation and Note that F41 − 4d55 = 0 and F415 = 4d5/4d5. Thus, the coupon competition. Bell J. Econom. 14(2):315–325. mass on p = 1 equals = 1 − F415 or Gill D, Sgroi D (2012) The optimal choice of pre-launch reviewer. 4d5 − 4d5 J. Econom. Theory 147(3):1247–1260. = 0 (3) Gorodnichenko Y, Sheremirov V, Talavera O (2014) Price setting 4d5 in online markets: Does it click? NBER Working Paper 20819, In equilibrium of the pricing subgame, firm mixes over National Bureau of Economic Research, Cambridge, MA. Grossman SJ (1981) The informational role of warranties and private all p ∈ 64d5−4d51 4d55 according to distribution (2). Firm ∈ − disclosure about product quality. J. Law Econom. 24(3):461–483. mixes over all p 41 4d51 17 according to distribution (3), Heiman A, Muller E (1996) Using demonstration to increase new with mass point on p = 1 given by (3). Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved. product acceptance: Controlling demonstration time. J. Marketing Optimal demonstration policy. It is always feasible for firm Res. 33(4):422–430. to set a demonstration strategy d ∈ 601 17 such that 4d5 > 4d5. Heiman A, McWilliams B, Chen Z, Zilberman D (2001) Demon- Given that 4d5 ≤ 4d5 results in = 0 and 4d5 > 4d5 strations and money-back guarantees: Market mechanisms to reduce uncertainty. J. Bus. Res. 51(1):71–84. results in > 0, firm always prefers such a d. The optimal d such that 4d5 > 4d5 maximizes Iyer G, Soberman D, Villas-Boas JM (2005) The targeting of advertising. Marketing Sci. 24(3):461–476. 2 = 44d5 − 4d554d5 = − 41 − d41 − 55 1 Johnson JP, Myatt DP (2006) On the simple economics of advertising, marketing, and product design. Amer. Econom. Rev. 96(3):756–784. which is strictly increasing in d ∈ 601 17. Thus, fully informative Kamenica E, Gentzkow M (2011) Bayesian persuasion. Amer. Econom. demonstrations are optimal for firm . Rev. 101(6):2590–2615. Boleslavsky, Cotton, and Gurnani: Demonstrations and Price Competition 12 Management Science, Articles in Advance, pp. 1–12, © 2016 INFORMS

Kane YI, Sherr I (2011) Secrets from Apple’s genius bar: Full loy- Moorthy S, Srinivasan K (1995) Signaling quality with a money- alty, no negativity. Wall Street Journal (June 15), http://on.wsj back guarantee: The role of transaction costs. Marketing Sci. .com/iAQhQR. 14(4):442–466. Kuksov D, Lin Y (2010) Information provision in a vertically Moscarini G, Ottaviani M (2001) Price competition for an informed differentiated competitive marketplace. Marketing Sci. 29(1): buyer. J. Econom. Theory 101(2):457–493. 122–138. Rayo L, Segal I (2010) Optimal information disclosure. J. Political Lecher C (2015) Amazon is overhauling its reviews system. The Econom. 118(5):949–987. Verge (June 20), http://www.theverge.com/2015/6/20/8818621/ Rotemberg JJ (1982) Sticky prices in the United States. J. Political amazon-reviews-system-changes/. Econom. 90(6):1187–1211. Lewis TR, Sappington DEM (1994) Supplying information to facilitate Schlee EE (1996) The value of information about product quality. price discrimination. Internat. Econom. Rev. 35(2):309–27. RAND J. Econom. 27(4):803–815. Meurer M, Stahl D (1994) Informative advertising and product match. Shaked A, Sutton J (1982) Relaxing price competition through product Internat. J. Indust. Organ. 12(1):1–19. differentiation. Rev. Econom. Stud. 49(1):3–13. Milgrom P, Roberts J (1986) Price and advertising signals of product Villas-Boas JM (2004) Communication strategies and product line quality. J. Political Econom. 94(4):796–821. design. Marketing Sci. 23(3):304–316. Downloaded from informs.org by [75.183.55.49] on 07 February 2017, at 15:02 . For personal use only, all rights reserved.