This paper analyzes the competitive role of retail shopping experience in markets with consumer search costs.We examine how a retailer’s advantage in providing consumer shopping experience affects its equilibrium
pricing and price advertising strategies. We find that if the consumer valuation of a shopping experience issufficiently low, its effect on retailer strategy is similar to that of quality, and the retailer with the advantagein shopping experience then deploys higher levels of price advertising. On the other hand, when the shoppingexperience is valuable enough for consumers, it acts akin to price advertising in that it makes it optimal forthe retailer with the advantage in shopping experience to eschew price advertising. The optimal competitiveinvestments in consumer shopping experience can be higher than that of a monopoly. The profit impact ofshopping experience for a retailer depends on the level of shopping experience: for low levels, the profit impactdepends on the difference in the levels between the retailers, but for high enough levels, it depends only onwhether the retailer’s shopping experience level is higher than that of its competitor. In this case, even smalldifferences in shopping experience levels can result in large differences in equilibrium profits.
Key words : shopping experience; retail competition; search costs; price advertising; store atmospherics;game theory
History : Received: December 12, 2011; accepted: June 12, 2012; Preyas Desai served as the editor-in-chief and JohnZhang served as associate editor for this article. Published online in Articles in Advance September 20, 2012.
1. IntroductionThe creation of in-store shopping environments thatenhance consumer shopping experience is recognizedby retail practitioners as an important element ofmodern retail strategy. For example, a 2010 surveyof 479 industry managers and participants conductedby RetailWire finds that in-store shopper experienceis the most important driver of customer satisfac-tion ahead of product assortment, deals, and promo-tions.1 Retailers make long-term investments in storeatmospherics, which includes elements such as light-ing, merchandising, music, salesperson attractiveness,entertainment, and even scent (e.g., dispersing fra-grance in the air to make the shopping process enjoy-able) (see, for example, Chapter 18 in Levy andWeitz 2009). Starting in the 1990s, this focus becameexemplified by the increasing adoption of what theindustry calls “entertainment retailing” by brick-and-mortar retailers. Successful retailers, such as FAOSchwarz or Toys “R” Us in toy retailing and REI in
1 For details, see Ball and Jones (2010). A 2002 industry study oncustomer relationship management in 10 U.S.-based retailers by theIBM Institute for Business Value indicates that in-store and person-to-person experience have significantly greater impact on satisfac-tion than other areas of investment, including pricing and value,marketing and communications, and data integration/analytics(Chu 2002).
sporting goods, invest significantly in creating enjoy-able experiences for shoppers, which can help bringmore consumers to their stores.2 Other retailers suchas Home Depot, Pottery Barn, and Selfridges, whooperate in highly competitive markets, use the strat-egy of entertainment and education (labeled by theindustry as “edutainment”) to bring potential cus-tomers to the store. Home Depot, for example, offersfree home-improvement classes so that customers canlearn how to use home-building and decoration sup-plies (supplies that are also available at competingstores such as Lowe’s). Similarly, Pottery Barn hasdedicated in-store decorating centers that conductdecorating events and classes.
This paper analyzes the effects of consumer shop-ping experience on the retail competitive strategiesand identifies incentives for providing such value toconsumers. Several important characteristics of thein-store shopping environment guide our analysis.
2 For example, consumers contemplating a visit to the FAO Schwarzflagship store in Manhattan can let their kids ride in the store’s$300,000 three-dimensional motion simulator, watch dancers per-form musical numbers on a giant piano keyboard, or let their child“adopt” a baby-doll (complete with adoption papers) in the store’snursery (Scardino 2005). Toys “R” Us reportedly spent $35 millionin 2001 on its flagship Times Square store to make it the “personi-fication of every kid’s dream” (Prior 2001, p. 46).
Retailers make substantial and ongoing investmentsin creating enhanced store shopping environmentswith the goal of drawing customers to the storeand inducing them to purchase. However, it is alsoapparent from the examples above that consumerscan enjoy the shopping experience and entertainmenteven if in the end they do not purchase the product.In other words, unlike in the case of product quality,consumers can enjoy the consumption utility derivedfrom the shopping experience even if they do notpurchase anything from the retailer. This public good–like nature of shopping experience implies that con-sumers may free ride on a retailer’s investments andenjoy the shopping environment without purchase—a behavior that is commonly referred to as “win-dow shopping” or browsing. This raises the questionof why retailers would prefer to use investments inshopping experience rather than use lower advertisedprices in order to attract consumers to the store. Inanalyzing the effect of consumer shopping experience,we investigate the unique role played by this variablecompared with other retail strategies such as pricingand price advertising. We then analyze the optimalretail investment strategies in making shopping morepleasurable to consumers.
The analysis highlights the role of shopping experi-ence when consumers have search costs of visiting theretailers and are uncertain about retail prices unlessthey make a shopping visit. In such markets, there isa potential for market failure; i.e., given the unobserv-ability of prices, consumers may be averse to incur-ring the cost of going to a retailer, even though onceat the store, they may obtain positive purchase util-ity. A retailer’s strategy will therefore be affected notonly by the incentive to compete for customers atthe store but also by the incentive to draw customersto the store. Retailers can build store environmentsthat provide significant shopping experience valueobservable to customers before they decide whetheror not to incur the search costs. This may induceconsumers to search because the shopping experienceutility can compensate for the event that consumersdo not obtain sufficient ex post purchase utility oncethey have arrived at the store. In addition, shoppingexperience may act as an instrument that helps theretailer compete for consumers from the rival retailer.On the flip side, if shopping experience investmentsare costly, retailers may face the drawback that con-sumers may free ride and consume the experiencewithout purchasing the product. A formal analysisis therefore useful for understanding these trade-offsunder competition and for determining the role ofshopping experience investments.
When consumers have search costs and do notknow the prices (unless they visit the stores), retailers
can use another retail decision variable: price adver-tising. Indeed, the existing literature (e.g., Lal andMatutes 1994) has focused on the role of price adver-tising in inducing consumers to visit the store. There-fore, we investigate the role of shopping experiencewhen price advertising is also possible. Accordingly,our model considers equilibrium price advertisingand pricing strategies and how they depend onthe retail shopping experience levels. We then con-sider the optimal investment in shopping experience,thus considering both price advertising and shop-ping experience as endogenous competitive instru-ments available to retailers.
The model considers two retailers competing inprice, price advertising, and shopping experience. Toconsider the simplest possible market structure thatallows for competition in the above variables, we ini-tially assume that consumer preferences for the prod-ucts are homogeneous and identical across the tworetailers. All consumers have a common search cost forvisiting a store or for going between stores, and theyface uncertainty about the price at a store unless thestore advertises it. We first examine the case in whichthe shopping experience levels of the firms are givenand then the case in which firms endogenously choosethe levels of shopping experience. We also extend themodel to analyze the possibility that consumers areuncertain about the product fit independent acrossretailers. In this case, retailers are differentiated in theproducts carried, and so the analysis helps to highlightimplications of retail differentiation on retail invest-ment in consumer shopping experience.
Given the homogeneous market, in the absence ofsearch costs, firms would face Bertrand competitionand zero profits. With search costs, if firms adver-tise and eliminate consumer price uncertainty, thesame outcome would ensue, but retailers would notbe able to recoup the advertising expenses. There-fore, the optimal advertising strategy in the absenceof shopping experience results in partial (probabilis-tic) consumer price information and a distribution ofretail prices but still results in zero equilibrium profitsnet of advertising costs, as any positive profits wouldinduce the retailer to advertise more. However, weshow that if a retailer has an advantage in the pro-vision of shopping experience value to consumers, itcan make positive equilibrium profits even thoughthis value is available to consumers irrespective ofwhether or not they purchase.
An interesting result that emerges from the analy-sis is the relationship between shopping experienceand price advertising in competitive markets. Priceadvertising acts to attract consumers to the storeas well as to remove consumer uncertainty aboutprices, but it does not provide any direct consump-tion utility. Shopping experience attracts consumers to
the store but does not resolve consumer uncertaintyabout prices. Shopping experience is also unlike priceadvertising in that it creates direct consumption util-ity/value over and above the value of information.This raises the question of whether shopping expe-rience has a role that is more similar to a qualityimprovement (which creates extra consumer utility)or to price advertising (which provides information).When the shopping experience levels of retailers arelow, they do not fully compensate consumers fortheir search costs. Retailers therefore have to adver-tise prices in order to induce consumers to visit theirrespective stores. In this case, shopping experienceacts akin to product quality valuations, and conse-quently, the retailer with the advantage in shoppingexperience advertises at a higher level. It turns outthat this effect is discontinuous in the level of shop-ping experience: the retailer with just a slight advan-tage advertises with probability 1. In contrast, atsufficiently high levels, shopping experience acts as asubstitute for price advertising and helps to draw con-sumers to the store. In this case, the retailer with theadvantage does not advertise. However, the disad-vantaged retailer still finds it optimal to advertise nomatter how high its shopping experience level is, evenif all consumers find it optimal to visit both storesto consume the shopping experience value. Thusour analysis describes the conditions when shoppingexperience acts like a substitute for price advertisingas opposed to a complement.
Another important result is that when shop-ping experience levels are high enough, theadvertising and pricing strategy depend only onwhether or not the retailer has the higher shop-ping experience level and not on the level per se.This suggests that the competitive payoffs fromshopping experience differ qualitatively from thepayoffs from standard quality improvements. Unlikein the case of product quality, payoffs from shoppingexperience at high enough levels increase discontin-uously when the shopping experience of a retailerexceeds that of the rival. Thus at high levels of shop-ping experience, the competition in this instrument issimilar to what is observed in winner-take-all markets,in that even a small difference in the levels allowsthe firm with the higher shopping experience level toachieve disproportionately higher profits. Retailerscompete for the order of consumer visitation. Animplication of this is that firms may overcompete inshopping experience compared with the choice of amonopoly firm.
We then consider the case in which retailers makecostly investments in the levels of shopping experi-ence, thus endogenizing shopping experience levels.We find that when the cost of shopping experience
is sufficiently high, the equilibrium in pure shop-ping experience strategies has the two ex ante similarretailers endogenously differentiating in their choicesof shopping experience. This is reminiscent of themodels of quality differentiation. There is also a sym-metric equilibrium in mixed strategies. Furthermore,when the cost of shopping experience is sufficientlylow, only the symmetric mixed strategy equilibriumexists. Therefore, competition in shopping experienceis analytically distinct from quality competition (as in,e.g., Shaked and Sutton 1982, Moorthy 1988), wherethe differentiated pure strategy equilibrium alwaysexists. Note that when shopping experience levels arehigh enough so that consumers will visit both storesregardless of purchase, the retail competition is for theorder in which consumers would visit the stores, andthe retailer with the higher shopping experience levelwill be visited first. This makes the payoffs from shop-ping experience investments for a firm discontinuousaround the investment level chosen by the competitor(which is again in contrast to the payoffs from qualityinvestments). This nature of the payoffs also pointsout the importance of competitive intelligence aboutthe competitor’s planned level of investment.
If consumers face uncertainty about product fitindependent across retailers, then retailers are ex postdifferentiated in the products they offer. Therefore,one may expect the outcome of such competition to bebetween the outcomes of a monopoly and an undiffer-entiated competition. However, we find that shoppingexperience investments under some levels of prod-uct fit uncertainty, even in the absence of advertisingcosts, are higher than under the case of monopolyand the case of the undifferentiated retail competition.Furthermore, for an intermediate range of retailercosts of shopping experience, the optimal investmentlevels are, on average, higher than the socially optimalones, even when advertising costs are absent.3 Thisresult supports the observation that retailers in suchproduct categories as fashion goods or apparel, wherefit is uncertain, invest more in enhancing shoppingexperience than supermarkets, which sell frequentlypurchased goods.
It is also useful to compare the competitive incen-tive to invest in shopping experience to the monopolyone. Although it is known that competing firms may
3 When advertising costs are high, then the retailer (either inmonopoly or in competition) could use shopping experience invest-ment at a level that is at least sufficient to cover the consumersearch cost in order to avoid advertising costs. Therefore whenshopping experience costs are somewhat high as well, this invest-ment could be higher than socially optimal if one assumes that asocial planner need not advertise because she would set prices lowenough. However, when price advertising has no costs, this resultis no longer trivial and does not hold when retailers are not differ-entiated in the product fit.
overinvest in informative advertising, prior research(e.g., Tirole 1988) has also established the oppositein the case of product quality (i.e., competition mayattenuate the incentive to invest in quality).4 Our anal-ysis of endogenous investments in shopping experi-ence provides a useful perspective that complementsthe existing research. We show that competing firmsmay overinvest in shopping experience relative to amonopoly when shopping experience acts as a sub-stitute for advertising. This is similar to the findingsin existing research such as Grossman and Shapiro(1984). However, in contrast to the existing research,we show that in this case firms do not overinvest inshopping experience relative to what is socially opti-mal. This then also highlights our finding in the caseof uncertain product fit that firms may overinvest inshopping experience relative to even the socially opti-mal level, but only when shopping experience acts asa substitute for quality improvement and not when itacts as a substitute for price advertising.
1.1. Related ResearchThis paper is related to the literature on retail strate-gies to promote store traffic and attract consumersto the store. Some of the early research in this areaexamined the phenomenon of loss leader pricing, i.e.,pricing some products below marginal cost. Hess andGerstner (1987) were perhaps the first to formallyanalyze loss leader pricing and provide a full infor-mation rationale for such practice. They consideredconsumers shopping at retailers selling two typesof goods—impulse goods and shopping goods. Theformer are defined as those purchased “on sight”without any price comparison, whereas consumersconduct price comparisons for the latter. Hess andGerstner show that retailers may price the shoppinggood below its marginal cost to attract consumers tothe store and then recoup the profits through charginghigher prices for the impulse goods.
Another important issue related to the store traf-fic is that consumers may not have information aboutretail prices without incurring a costly shopping trip.In this context, informative advertising (Butters 1977,Grossman and Shapiro 1984, Lal and Matutes 1994,Soberman 2004, Iyer et al. 2005, Amaldoss and He2010) is a useful instrument because when consumersincur travel costs, the lack of retail price observabilitycan lead consumers to not shop at all.5 The closestapproaches to our research are the models of price
4 Brekke et al. (2010) show that competition could increase the sup-ply of quality in the presence of income effects.5 Alternatively, Desai et al. (2010) considered the possibility that inthe absence of retail price advertising, consumer valuations can bereduced as a result of consumers having to incur price discoverycosts, and therefore retailers may have to lower prices.
advertising in Lal and Matutes (1994) and Lal and Rao(1997). In their models, consumers are uninformedabout prices unless they are advertised, and thereforeconsumer price expectations and advertising play arole in explaining loss leader strategies. Rajiv et al.(2002) investigated how the service/quality level ofcompeting retailers is related to their price advertisinglevels. Lal and Matutes (1994) showed the existenceof a loss leader equilibrium, in which both retailersadvertise one of their goods at a low price and con-sumers rationally expect that the other good will beavailable only at the reservation price. Price adver-tising (of a sufficiently low price) attracts consumersto the store, but unlike shopping experience, it doesnot provide any direct independent-of-purchase con-sumption utility. Shopping experience enhancementsattract consumers to the store by providing consump-tion utility, but they do not directly resolve consumeruncertainty about prices. Our analysis highlights theconditions under which shopping experience invest-ments act akin to price advertising and help bringconsumers to the store and the conditions underwhich they act similarly to product quality and helpthe retailer to charge higher prices.
Given the central role of consumer shopping costsand the lack of price information as a justification forshopping experience, our research also relates to thelarge literature on the role of search costs in compet-itive markets with imperfect price information. Thisliterature starts with the seminal paper by Diamond(1971), which showed that in a homogeneous goodsmarket with positive search costs, the only equilib-rium is for all firms to charge the monopoly priceand for consumers not to search. This result—thateven very small search costs lead to a “no-trade”equilibrium—is known in the literature on searchcosts as the Diamond paradox. The argument for itis as follows: Suppose consumers search among theretailers sequentially and must pay a search cost s > 0to learn a retailer’s price. If the consumers expect theretailers to charge pe, they will only visit a retailer iftheir product valuation is V > pe +s. Once a consumerhas arrived at a store, she will purchase the product ifthe price is less than pe +s, because it is not optimal toincur an additional cost of s to visit another retailer.Therefore retailers will always charge more than con-sumer expectations unless the price is constrained bythe consumer valuation V . The only equilibrium is forconsumers to expect V and not visit any retailer evenfor very small search costs.
Not surprisingly, there is a significant line of re-search that proposes resolutions to this no-traderesult through different mechanisms. These includeproduct differentiation with uncertain match values(Anderson and Renault 1999), a variation in searchtechnology such that more than one price could
be learned upon incurring the search cost (Burdettand Judd 1983), a consumer segment with zerosearch costs (Salop and Stiglitz 1977, Varian 1980,Narasimhan 1988), or retailer heterogeneity in infor-mation they have about consumer valuation (Kuksov2006). In contrast to this research, the shopping expe-rience mechanism in this paper is an endogenousfirm strategy (like price advertising) that helps induceconsumer search rather than resolve uncertainty. Theagenda is therefore similar to the research examiningother types of firm strategies. Such research includesWernerfelt (1994b), who considered firms’ colocationstrategy as a mechanism to attract consumers withsearch costs and, more recently, Janssen and Non(2009), who analyzed the possibility of advertisinginforming consumers that a firm is selling a particu-lar product at a certain price and therefore reducingthe expected consumer search cost compared with anonadvertising firm.
There is also a stream of research beginning withTelser (1960) on retailing services such as presaleinformational services, which consumers demandprior to purchasing. The key distortion highlighted inthis research is the free-riding problem faced by retail-ers that invest in informational services: consumerscould get the information necessary for making a pur-chase decision from a full-service retailer and thenpurchase from a discount outlet that does not pro-vide informational services but offers a lower price(Mathewson and Winter 1983). More recently, Shin(2007) showed that free riding may actually end upbenefiting the full-service retailer under competition.6
Though shopping experience investments are alsosubject to free riding, their effects and the associatedmechanisms are distinct from informational servicesin retail markets. First, unlike informational services,once the consumer is in the store, the shopping expe-rience is not necessary for the consumer to make apurchase. For a consumer already in the store, theshopping experience provides a pure consumptionutility that can be enjoyed independent of purchase.Second, because the consumption utility from shop-ping experience is observable before the consumersincur the search cost for the store visit, shopping expe-rience can play a role that is akin to price advertisingin motivating consumers to travel to the store even ifthe price or utility of the product is uncertain.
Note that there is also a clear distinction betweenshopping experience and the point-of-sale retailservices that reduce consumer search costs or help
6 The incentives to provide information have also been consideredin the context of return policies (Shulman et al. 2011), quality infor-mation (Guo and Zhao 2009), and multichannel retailing (Ofek et al.2011), as well as when competing firms have asymmetric qualitylevels (Kuksov and Lin 2010).
consumers find the best product fit. Existing researchhas examined these services in the context of in-storesales assistance (Wernerfelt 1994a, Ofek et al. 2011),which can also be seen as mitigating the search costsof finding the correct product only after the consumeris in the store. However, unlike shopping experience,these point-of-sale services do not allow for the pos-sibility of consumer free riding and do not provideconsumers with observable independent-of-purchaseconsumption utility, which is the particular featurethat allows shopping experience to compensate forthe consumer cost of the shopping trip.
Finally, there is a substantial behavioral litera-ture that investigates the role of sensory factors andthe effects of atmospherics on consumer shoppingbehavior (for reviews, see Peck and Childers 2008,Turley and Milliman 2000). Much of this literatureinvestigates the effects of specific sensory factors inthe store environment such as music, smell, or coloron key dependent variables such as consumer pur-chase behavior and the time spent in the store. Gorn(1982) showed that hearing pleasant music can signif-icantly affect product evaluations, whereas later stud-ies (e.g., Gorn et al. 1993) produced results consistentwith Schwartz and Clore’s (1983) affect-as-informationhypothesis—that the mood created by music impactedupon product evaluations only when the subjectswere not made aware of the source of their feel-ings. A good mood induced by music would affectproduct evaluations positively only if the consumerswere not made consciously aware of the source ofthe mood.7 This finding has obvious relevance forthe use of ambient or background music in retailstores; studies such as Milliman (1982) and Bakeret al. (1992) have shown that music as a store envi-ronmental variable impacts purchase behavior andthe time spent in the store. Other studies such asSpangenberg et al. (1996) and Mitchell et al. (1995)have shown that the choice of the appropriate ambi-ent scent can significantly influence sales, processingtime, variety seeking, and the perceived time spentin a store. Taken together, these studies identify theeffects of some relevant store atmospheric variableson consumer shopping behavior. Although our paperis about the competitive role of shopping experienceas a retail strategy instrument, the behavioral litera-ture can be seen as providing prescriptions for whichtype of in-store stimuli (e.g., music, taste, smell) couldpositively affect consumer perceptions and shoppingbehavior.
7 Recently, Iyer and Kuksov (2010) examined firms’ strategic choicesto supply quality as well as activities that create mood or affectfor consumers unable to fully separate the true-quality evaluationsfrom their affect. Because of this information-processing problem,firms end up in equilibrium, choosing costly mood-creating activi-ties even if they hold no direct consumption value for consumers.
2. ModelConsider two retailers, each selling a product to amarket consisting of a unit mass of consumers. Themarginal cost of the product for each retailer is con-stant and normalized to 0. Each consumer valuesthe product of both retailers at V . Each consumerhas single-unit demand and has to incur travel orsearch costs s of visiting either store (from his or herhome) or has to travel between the stores. Consumersobserve the price offered if they incur the cost s andare at the store or if the store has advertised its price(in which case they observe the price even if they donot incur the search cost). Consumers also have to beat the store from which they wish to buy; i.e., it isnot possible for consumers to buy without incurringthe search cost (and possibly without observing theprice).
If retailer j has the shopping environment to theextent of mj , consumers will get a shopping experi-ence utility of mj from that store if it is the first storethey visit but only �mj if it is the second store theyvisit, where � < 1. The rationale for this assumptionis that shopping activities involve physical fatigueor emotional satiation for enjoyment, which impliesthat consumers’ enjoyment from the shopping expe-rience at the second store is attenuated. Alternatively,a smaller value of � might also capture the fact thatthe shopping experiences in different stores are closersubstitutes and that consumers perceive less varietyin the type of shopping experience across the stores.8
We allow consumers to visit the same store morethan once, but we will assume that consumers cannot“double up” on the enjoyment of the shopping experi-ence by shopping at the same retailer more than once.This is reasonable because once a consumer comesto a retailer, she should enjoy the shopping expe-rience until the marginal benefit of doing so is nolonger positive. This would then imply that if the con-sumer returns to a previously visited store, there isno more residual enjoyment. The consumer may stillconsume the experience at a retailer not previouslyvisited because the experience at a retailer is differentfrom the experience at other retailers.
We assume that the shopping experience level of astore can be observed by consumers before they incurthe search cost, whereas prices cannot be observed
8 The analysis holds for � arbitrarily close to 1. However, �= 1 willresult in multiple equilibria when both mj values are sufficientlyhigh because then an arbitrary assumption can be made aboutwhich store the consumers will want to visit first. The assumption� < 1 forces consumers to prefer to visit the store with higher mj
first if they are otherwise indifferent. This by itself seems a naturalassumption and could be used as a refinement to analyze the caseof � = 1. Substituting � = 1 in the equilibrium that we derive inthis paper also leads to the equilibrium of the game with �= 1 thatsatisfies such a refinement.
until consumers incur the search cost, unless theretailer advertised the price. The reason for thisassumption is that shopping experience is a long-term decision that is not as easy to change as price(this is certainly true in decisions such as creatingan attractive storefront, designing the physical layoutand space, using infrastructure to create ambiance,hiring attractive salespeople, etc.). Thus the enjoy-ment a consumer has from the shopping experienceat a particular store is more stable than prices, andtherefore, over time, the consumer may have knowl-edge of the level of shopping experience but not ofthe easily changeable prices. Accordingly, we modela two-stage game in which the retailers first make the(long-term) shopping experience decisions, and thenin the next stage, they make the (short-term) pric-ing decisions along with the decision on whether toadvertise the price.9 If a retailer advertises, consumersknow its price before making the shopping trip; if itdoes not advertise, consumers only observe the priceif they incur the search (travel) cost s and visit thatretail store. The cost of advertising (to the full market)for each retailer is A.
In many situations retailers must compete givenpredetermined levels of shopping experience. Thiscould be due to constraints placed by historical cor-porate decisions or due to constraints placed by thenature of the retail site. Accordingly, we start withan analysis of retail price competition for given levelsof mjs. This then allows us to analyze how shoppingexperience interacts with price advertising to bringconsumers to the store and to establish the role ofshopping experience when prices can be communi-cated through advertising. We assume that V is highenough relative to s and A, so that even if mj = 0, amonopoly firm would prefer to be in the market. Inother words, a monopolist would prefer to advertiseprice p = V −s and achieve unit sales rather than havezero sales and profits. The explicit condition for thisis V >A+ s.
As we have noted above, consumers observe mj ofeach store prior to deciding on the shopping trip andobserve the price of store j if and only if it advertised.Given this information, consumers maximize theirexpected utility by deciding whether or not to travelto the store(s) and whether or not to buy at a partic-ular store. Note that consumers can enjoy the utilityfrom the shopping experience mj either before or after
9 Although in some situations, one could assume that advertising isdone in a stage prior to price selection, in the case of price adver-tising, the price has to be selected for it to be advertised. Therefore,we assume that the price and price advertising are decided at thesame stage and simultaneously across retailers. Because of this, theprice and advertising decisions of a given retailer can be (and inequilibrium, are) interrelated, but these decisions across the tworetailers are independent.
purchase, and the utility received is not conditionalon purchase. Therefore, the model allows for whatis commonly called window shopping: i.e., a shoppingtrip made for the purpose of obtaining (nonpurchase-related) enjoyment at the retail location and withoutthe intention (or sometimes even the possibility) ofbuying.
After considering the model with exogenous butpossibly different across-retailer levels of shoppingexperience, we endogenize the levels of shoppingexperience. Specifically, we assume that in a stage pre-ceding the price and price advertising choices, theretailers choose the levels of shopping experience,either simultaneously or sequentially. We assume thatthe cost of developing shopping experience at thelevel m for either retailer is a fixed cost cm2 indepen-dent of the number of units subsequently sold. Thiscaptures many of the important long-term invest-ments in store appearance, layout, and store atmo-spherics variables such as merchandising, lighting,music, sales staff, and in-store entertainment activitiesthat enhance consumer experience.
3. Analysis and SolutionWithout loss of generality, we assume that m2 > m1,i.e., that retailer 2 has the advantage in retail shop-ping experience.10 Let us denote retailer j’s price by pjand the probability of retailer j’s advertising by �j . Asalready noted, a retailer’s pricing strategy is interre-lated to its advertising strategy. We will show that inequilibrium, a retailer’s pricing strategy involves ran-dom sales from a certain maximum price (i.e., a mixedstrategy with a possible mass point at the top of theprice distribution) either when the retailer advertisesor when it does not, but not both. Let Mj denote themass point of retailer j’s mixed strategy at the topof its distribution, and let Fj4p5 denote the cumula-tive distribution function (cdf) of that distribution. Atsome value of price difference p2 − p1, consumers areindifferent between going to retailer 1 and retailer 2on the first visit. When shopping experience levels aresmall enough so that consumers prefer to go to oneretailer only, it is logical that the consumers prefer togo to the retailer with the higher shopping experiencelevel, i.e., retailer 2, unless retailer 1 compensates forthe consumer’s lower shopping experience value with
10 We do not explicitly consider the case of m2 = m1 because (i) aswe will show in the following section, where the shopping expe-rience levels are endogenous, the retailers’ choice of mj will notresult in m1 =m2 with positive probability; and (ii) the equilibriumin this case is not unique because the limit of the equilibrium strate-gies as m2 → m1 is also an equilibrium for m2 = m1, and it is notnecessarily symmetric. In other words, there are three equilibria foreach m2 =m1, two of which are mirror images of each other andthe third that is symmetric.
a correspondingly lower advertised price. As we willsee from the subsequent analysis, this price differenceis indeed positive in all cases, but it is not alwaysequal to the difference m2 −m1.
3.1. Effect of Shopping Experience onConsumer Shopping Strategy
To determine the retailers’ optimal price advertisingstrategy, it helps to first consider how shopping expe-rience and advertising could influence the consumershopping trip strategy. One can develop the logic asfollows: If the level of shopping experience at bothstores is low enough (namely, mj < s for j = 112), con-sumers would only go to a retailer if they expect thepossibility of buying there. The standard holdup argu-ment presented in §1.1 then implies that consumerswould expect any retailer that did not advertise to set aprice that is too high for consumers to benefit from vis-iting that retailer. Therefore, in this case, price adver-tising is essential for a retailer to bring customers to itsstore and to possibly achieve positive sales. The opti-mal consumer strategy in this case is to decide, basedon the retailers’ price advertising (if any), from whichstore to buy while at home and to go only to the storethat provides the higher utility (if this utility is pos-itive). Note that the higher utility is determined bycomparing V − pj + s +mj for j = 112.
At the opposite extreme, if shopping experienceat each retailer is high enough (�mj > s), consumerswould go to both retailers regardless of their pur-chase intention. To obtain the highest total enjoymentfrom the shopping experience, consumers will preferto first visit the retailer with the higher m (retailer 2)and then the other retailer. In this case, if retailer 1were to not advertise, consumers will infer that itsprice is at V , and therefore they will buy at retailer 2as long as its price is below V . Therefore, retailer 1will have to advertise in order to achieve positivesales. On the other hand, retailer 2 does not needto advertise because consumers will first visit thisretailer regardless.11
In between the above two cases, consumers wouldlike to visit at least one retailer for the shopping expe-rience even if they do not expect to buy. Of course, ifneither retailer advertises and consumers expect highprices (p = V or slightly below), they will then visitthe retailer with the highest m (retailer 2). In contrast,suppose both retailers did advertise, and consumershaving observed both prices decide that it is best tobuy at retailer 1. Then if they only go to retailer 1,
11 Conceivably, retailer 1 could refrain from advertising in the hopethat consumers would visit it first to make sure they are makingthe right choice of buying at retailer 2. But it is easy to see thatthis cannot be an equilibrium because it would then be optimal forretailer 2 to price just below V , and consumers would be strictlyworse off sacrificing 41 −�5m2 for such a strategy.
they obtain the utility of V − p1 +m1 − s; if they visitretailer 2 first (just for the shopping experience) andthen retailer 1 (to buy the product and for the extrashopping experience), they will obtain the utility ofm2 − s + V − p1 + �m1 − s.12 The second option aboveis preferred to the first one if and only if m2 > 41−�5 ·m1 + s. Therefore, if this condition holds, retailer 2will have no incentive to advertise, because even ifconsumers are convinced by retailer 1’s advertisingto buy there, they will still visit retailer 2 first (atwhich time they will know p2 whether or not retailer 2advertised), whereas retailer 1 does have the incen-tive to advertise (otherwise, consumers will not visitit after visiting retailer 2). If the above condition doesnot hold, then consumers will go to retailer 1 firstif it advertises a sufficiently low price (relative tothe expected or observed price at retailer 2). There-fore, retailer 2 also has the incentive to advertise. Aswe show in the analysis, the equilibrium in this caseindeed involves both retailers advertising with posi-tive probability.
To summarize the above discussion, we have thefollowing possibilities of consumer shopping behav-ior as a function of the retail shopping experience val-ues m1 <m2.
1. m2 < s: Consumers compare values V − pj +
mj − s based on advertised prices and go to theretailer that provides the higher value (as long as it ispositive).
2. s < m2 < s + 41 − �5m1: If retailer 1 does notadvertise, consumers will go to retailer 2. If retailer 1advertises a price low enough relative to consumerexpectation of the price at retailer 2 (which, of course,equals the actual price if retailer 2 did advertise), con-sumers only go to retailer 1 and buy from there.
3. s + 41 − �5m1 < m2 and �m1 < s: Consumersalways go first to retailer 2 regardless of advertisingand then go to retailer 1 if it advertised a low enoughprice.
4. �m1 > s: Consumers always go to retailer 2 andthen to retailer 1 regardless of advertising.Given these optimal consumer strategies, retailersdecide on the price and advertising strategies. Thenext section analyzes the equilibrium retailer strate-gies for each of the cases above.
3.2. Retail Price and Advertising StrategiesCase 1: m1 < m2 < s. As already noted, in this case,because the levels of the shopping experience forboth retailers are lower than the search costs, con-sumers do not have the incentive to visit either ofthe retailers in order to just consume the shopping
12 Consumers could also consider buying at retailer 1 and then goto retailer 2 just for the shopping experience, but this option isdominated by going to retailer 2 first.
experience. Therefore, when retailer j advertises, itwill receive zero demand unless the price is at orbelow p̄j = V − s +mj , and so this is the upper boundon retailer j’s advertised price. Further, if a retailerdoes not advertise, consumers will rationally expect ahigh price of V and will not visit the store. Thus theretailer’s demand and profit without advertising willbe zero. When both retailers advertise, the one withthe greater level of shopping experience (retailer 2)should have a higher profit, and so retailer 2’s profitwith advertising should be positive. This implies thatretailer 2 strictly prefers advertising to not advertis-ing, and hence, �2 = 1. Given this, retailer 1 will notadvertise with probability 1 because doing so wouldlead it to price competition resulting in zero price andnegative net-of-advertising profits (while retailer 2would charge price m2 − m1 and achieve a profit ofm2 −m1 −A). As we derive below, retailer 2’s pricingstrategy will involve a mass point M2 at the top ofits price distribution (p̄2), whereas retailer 1—when itadvertises—will charge up to p̄1 but without a masspoint.
Depending on the price advertising received, con-sumers will either compare the “offers” across thetwo retailers when both of them advertise or (ifretailer 1 did not advertise) buy from retailer 2, whichalways advertises as long as it advertises a price ator below p̄2. Because the best offer is determined bycomparing V − pj − s + mj across the two retailersand to 0 (the outside option of not buying), a con-sumer prefers retailer 1 to retailer 2 if and only ifp1 < p2 −ãm, where ãm =m2 −m1. The mixed strategyequilibrium is characterized by the following systemof equations representing the indifference of retail-ers 1 and 2, respectively:
The first equation represents retailer 1’s indifferencebetween advertising and not, and the second one rep-resents retailer 2’s indifference between pricing at theupper bound and a putative lower price. Solving (1),we obtain the equilibrium retailer advertised pricestrategies as
F14p15=1�1
−41 −�15p̄2
�14p1 +ãm5for p1 ∈ 4A1 p̄153
F24p25=
1 at p2 = p̄21
1 −A
p2 −ãm
1 for p2 ∈ 4A+ãm1 p̄250
(2)
The above equilibrium price distribution for retailer 2implies that there is a mass point M2 = A/p̄1 at p̄2.
Retailer 1 advertises with probability �1 such thatp1 = p2 −ãm =A; therefore,13
�1 = 1 −A+ãm
p̄2= 1 −
A+ãm
V − s +m20 (3)
In this case, retailer 1’s advertising is complemen-tary to its own shopping experience, and it increaseswith m1 but decreases with the shopping experiencelevel of the competitor m2. Notice also that retailer 1has zero expected profits but retailer 2 earns posi-tive expected profit of ãm =m2 −m1. Thus retailer 2’sprofit advantage results from the difference in shop-ping experience levels.
Given the assumption that retailers are not differen-tiated in the product dimension, and that one of themhas a (strict) advantage in the shopping experience,it is not surprising that the disadvantaged retailerreceives zero profits. Note that the disadvantagedretailer still operates with positive expected demand.Of course, one should not interpret this outcome asthe prediction that one of the retailers in observedmarkets will end up having zero profit, as in practicethe retailers are at least somewhat differentiated in theproducts they carry. In §6.1, we show how differentia-tion leads to positive profits for both retailers withoutinvalidating the other predictions of our model.
Case 2: s < m2 < s + 41 − �5m1. This is the casein which given no price advertising, consumers visitretailer 2 for the shopping experience. Furthermore,as argued above, the condition s < m2 < s + 41 − �5m1implies that m2 is low enough so that if consumersare induced (by price advertising) to visit retailer 1first, they will not go to store 2 just to consume theshopping experience. Because m2 > s, one may pre-sume that consumers could possibly go to retailer 2even if it never advertises because retailer 1 doesnot always advertise. But this presumption would beinvalid, because if retailer 2 does not advertise atall, then retailer 1 will advertise a price just belowV +m1 −m2, and retailer 2 would never receive anydemand. Hence, in the equilibrium, both firms willadvertise with some positive probability.
In equilibrium, retailer 1 has zero expected profit(which is equal to the profit if it does not adver-tise), and retailer 2’s indifference condition is given byequating the profit when advertising at any price tothe profit when not advertising but charging p2 = V(in which case it has positive demand only whenretailer 1 does not advertise).14 Note that if retailer 2does not advertise and retailer 1 does, retailer 1 can
13 Note that this equilibrium holds as long as V >A+ s−m1 (whichwas assumed), i.e., if it pays for the retailer with lower shoppingexperience to advertise if it were a monopoly.14 One could also imagine a possibility that when retailer 1 adver-tises some specific price, retailer 2 could charge a lower price in
convince consumers to visit its store first as long asit sets the price at or below p̄1 = V − ãm, which istherefore the highest price retailer 1 can advertise toobtain positive sales. Therefore, the equilibrium priceand price advertising probabilities must satisfy{
where the price distributions are conditional onadvertising. To determine the advertising probabili-ties, note that at p1 = p̄1, retailer 1 has demand equal tothe probability that retailer 2 does not advertise andzero profit net of advertising (because it is indiffer-ent between advertising or not), which yields �2 = 1−
A/p̄1. Further, retailer 2’s lowest possible price mustbe higher than retailer 1’s lowest possible price by ãm,which then determines �1. Solving Equations (4) withthe above conditions on �j , we obtain
F14p15=1�1
−A+ 41 −�15V
�14p1 +ãm5for p1 ∈ 4A1V −ãm51
F24p25=1�2
−A
�24p2 −ãm5for p2 ∈ 4A+ãm1V 51
with
{
�1 = 1 −ãm/V 1
�2 = 1 −A/4V −ãm50(5)
Substituting p̄1 in F14p5 above, we obtain thatretailer 1 will have a mass point of M1 = A/4V�15 atp̄1 = V −ãm. This mass point is consistent with thefact that upon advertising, retailer 2 will have a higherprofit. Retailer 2 has no mass point at the top ofits distribution of prices p̄2 = V . Substituting �1 intoEquation (4), we obtain that the expected profits ofthe retailers in this case are the same as in the casem1 <m2 < s; i.e., �1 = 0 and �2 =ãm =m2 −m1.
Case 3: s+ 41 −�5m1 <m2 and �m1 < s. As reasonedbefore, in this case, consumers always go to retailer 2first. Furthermore, if retailer 1 does not advertise, con-sumers buy at retailer 2 if and only if p2 ≤ V and
the hope that consumers will come to its store expecting to get thislower price even when it does not advertise. To show that suchconsumer behavior cannot be an equilibrium, consider the lowestprice p∗ advertised by retailer 1 that results in consumers going toretailer 2 first (of course, p∗ < V , because advertising a price at orabove V has no benefit for the retailer). Given this, retailer 2 knowsthat if a consumer came to its store, either she did not observe p1
or she observed p1 ≥ p∗. Then retailer 1 will optimally set its pricea shade lower than p∗ + s−�m1, which (under the conditions of thecurrent case) results in the consumer being strictly worse off fromvisiting retailer 2 first instead of retailer 1 when observing p1 = p∗.Therefore, there is no minimal advertised price p1 at which, in equi-librium, consumers could go to retailer 2 first given that retailer 2does not advertise. Therefore, if consumers go to retailer 2, then itimplies that retailer 1 did not advertise, and then it is optimal forretailer 2 to set price p2 = V .
do not visit retailer 1. When retailer 1 does advertise,consumers still visit retailer 2 first, but they buy thereonly if it is not better to go to and buy at retailer 1.It follows that consumers buy at retailer 2 when p2 <p1 + ãs , where ãs = s − �m1 is the price differenceunder which consumers are indifferent between buy-ing at retailer 2 once already there and making theadditional visit to retailer 1 and buying there. Thiscase is thus qualitatively different from the previoustwo cases because here consumers consider whetherto buy at retailer 1 after they have already visitedretailer 2, whereas in the previous two cases, con-sumers consider which retailer to visit before incur-ring any search cost.
Retailer 1’s indifference between advertising andnot and retailer 2’s indifference between any feasibleprice and the price p̄2 = V it can charge when theother retailer does not advertise implies
Finally, the equilibrium advertising level �1 isobtained from the condition that the lower bounds onthe price distributions must differ by ãs , which yields
�1 = 1 − 4A+ãs5/V 1 (8)
and from the condition that at the upper boundF 4 · 5= 1, we obtain that price distribution of retailer 2has a mass point of M2 =A/4V −ãs5 at p̄2 = V .
Thus in the current case, the retailer 1 and 2 prof-its are �1 = 0 and �2 = A+ãs = A+ s − �m1, respec-tively. The key difference between this case and theprevious ones is that retailer 2 saves on the advertis-ing cost because it has such a high level of shoppingexperience that consumers always go there first, and itachieves discontinuously higher profits independentof shopping experience level.Case 4: �m2 > �m1 > s. Finally, consider the pre-
vious case but when �m1 > s. Now consumers willalways visit both retailers just to enjoy the shop-ping experience utility. This represents consumerwindow shopping. Furthermore, since the conditionm2 > s + 41 −�5m1 always holds when �m2 > �m1 > s,consumers will always want to visit retailer 2 first(to consume the nondiscounted utility m2) and thenretailer 1. As in Case 3, retailer 2 does not need to
advertise because price advertising does not have anyincremental benefit for a retailer that is always vis-ited by consumers first. Further, if retailer 1 does notadvertise, it will have an incentive to set a higherprice (because if a consumer reaches it without buy-ing at retailer 2, the consumer would have to incurthe search cost s to go back).
In summary, consumer and retailer behavior is sim-ilar to that in the previous case, except that in this caseconsumers go to retailer 1 even if they have alreadypurchased at retailer 2. Therefore, if retailer 1 adver-tises its price, consumers are indifferent between thetwo retailers only given equal prices. Similar to theprevious case, we can solve for the equilibrium asfollows:
F14p5=1�1
−41 −�15V
�1p= 1 −
41 −�154V − p5
�1pfor p ∈ 4A1V 51
F24p5= 1 −A
pfor p ∈ 4A1V 50
(9)
It follows from the above that retailer 2 has a masspoint of M2 =A/V at p̄2 = V . Retailer 1’s equilibriumprice advertising is �1 = 1−A/V . The retailer, 1 and 2profits are �1 = 0 and �2 =A, respectively. Comparedwith Case 3, here one can see that the profits are thesame as the ones in the limit of the previous case as�m1 → s. Indeed, the last case is analogous to and canbe recovered from Case 3 by noting that, in general,ãs = max801 s −�m19 and so when �m1 > s, ãs = 0.
4. Effects of Retailer Advantage inConsumer Shopping Experience
Table 1 summarizes the equilibrium advertisingstrategies and profits as functions of shopping expe-rience levels mj (j = 112). Before analyzing the effectsof retailer advantage in shopping experience, it is use-ful to consider the benchmark case in which both
Table 1 Equilibrium Price Advertising and Profits 4m1 <m25
retailers offer zero shopping experience. In this case,if advertising costs are prohibitively high (A>V − s)and there is no price advertising, there will be a mar-ket failure leading to zero retailer demand and profits.Consumers will rationally expect the highest price Vand will therefore not want to incur the search cost tovisit either retailer. However, if advertising costs arelow enough (A< V − s), retailers will advertise withpositive probability. If a retailer advertises in equilib-rium with probability less than 1, the retailer mustbe indifferent between advertising and not. There-fore, its equilibrium profits in this homogeneous con-sumer market will be zero. It is also not possiblein equilibrium for both retailers to always adver-tise, because if they do, the standard Bertrand argu-ment applies, leading to zero prices and negativenet-of-advertising profits. Therefore the symmetricequilibrium is such that both retailers advertise withpositive probability lower than 1, which turns out tobe �= 1 −A/4V − s5.15 In the extreme, if advertisingcosts are zero, both retailers always advertise, and theequilibrium profits are still zero because retailers com-pete head-to-head for all consumers.
This means that in a homogeneous consumermarket with symmetric advertising costs, price adver-tising alone does not lead to positive equilibriumprofits. This sets up the rationale for the use of shop-ping experience as a competitive tool in markets withunobservable prices and search costs. How does pos-itive and asymmetric levels of shopping experienceacross retailers affect advertising decisions and prof-itability? We analyze this question in the remainder ofthis section and discuss the conceptual implicationsof the results derived in the previous section.
4.1. Shopping Experience and Price AdvertisingProposition 1 demonstrates the manner in whichshopping experience interacts with price advertisingand suggests when shopping experience should actas a substitute for price advertising as opposed to acomplement.
15 This can be derived as follows: Following the standard Diamondparadox argument presented in §1.1, if retailer j does not advertise,it gets no demand and profits. Therefore, given advertising proba-bilities �j , the solution then is similar to one in Narasimhan (1988):with probability 1−�3−j (i.e., that the other retailer does not adver-tise), any advertised retailer j’s price at or below V − s results insales to all consumers, whereas with probability �3−j , retailer j sellsto all consumers if and only if it advertises prices lower than theadvertised price of retailer 3 − j . The equilibrium pricing is then inmixed strategies with each retailer being indifferent between pric-ing at V − s and lower prices, and the expected equilibrium profitof retailer j must be the same as its expected profit at the priceV − s, which is 41 − �j 54V − s5. To solve now for equilibrium �j ,recall that if advertising is probabilistic, the advertising cost mustbe equal to the profit given advertising, i.e., A= 41−�j 54V − s5, andthus, �j = 1 −A/4V − s5.
Proposition 1. (i) For low levels of shopping experi-ence (m1 <m2 < s), the advantaged retailer always adver-tises while the disadvantaged retailer advertises with someprobability less than 1.
(ii) For intermediate levels of shopping experience (s <m2 < s + 41 − �5m1), both retailers advertise with a posi-tive probability less than 1. Furthermore, the probability ofadvertising for each retailer depends only on the differencein shopping experience levels, not on their absolute levels.
(iii) Finally, when the shopping experience level of atleast one firm (i.e., retailer 2) is sufficiently high (m2 >s + 41 − �5m1), the advantaged retailer never advertises,and only the disadvantaged retailer advertises with someprobability less than 1.
When the levels of shopping experience of bothfirms are sufficiently small, as in Case 1 4m1 <m2 < s5,shopping experience alone cannot compensate for thesearch costs consumers incur to visit the store. Thisimplies that both retailers have to advertise to bringconsumers to the store. With the advantage, retailer 2advertises with probability 1. Shopping experiencelevels in the presence of price advertising affect theat-home consumer value, which determines the con-sumers’ decision regarding which retailer they visitand buy from. In this case, shopping experience actsakin to an improvement in product quality and allowsretailers to charge higher prices. This is also consis-tent with the relationship between the equilibriumadvertised prices and retailer 2’s advantage in shop-ping experience ãm = m2 − m1: the expected equilib-rium price of retailer 1 is decreasing in this advantagewhereas that of retailer 2 is increasing. When shop-ping experience investments continue to be small butlarge enough for one firm that it can compensate forthe consumer search costs as in part (ii) of Proposi-tion 1, the equilibrium price and advertising behaviorare qualitatively similar to the previous case. How-ever, because retailer 2 can now also rely on shoppingexperience to induce consumers to come to its store,it need not always advertise its price.
When retailer 2 has a sufficiently high level of shop-ping experience as in part (iii) of Proposition 1, con-sumers will not only be ensured a high enough utilityto compensate for their search costs, but they willalso have the incentive to visit retailer 2 first in orderto consume the undiscounted shopping utility there.Once consumers are at retailer 2, then dependingon retailer 1’s pricing and advertising strategy (andif m1 < s/�), they decide whether to further travelbetween the retailers. Thus, because consumers willalways visit retailer 2 first, it does not need to relyon price advertising to bring consumers there. So incontrast to part (i) of Proposition 1, retailer 2 does notadvertise at all. The disadvantaged retailer 1 needs toadvertise to induce consumers to buy at its store.
Finally, at the extreme of m2 > m1 > s/�, shoppingexperience levels are high enough not only to com-pensate consumers for the search cost of visiting eachretailer but also to make traveling from any retailerto its competing retailer just to consume the shop-ping experience optimal. This characterizes window-shopping behavior on the part of the consumers. Itis for this reason that equilibrium prices are indepen-dent of shopping experience levels. When the shop-ping experience level at a retailer is sufficiently largeand higher than that at the other retailer, it substi-tutes for price advertising. Because shopping experi-ence acts akin to price advertising in this case, higherlevels of mj lead to lower mean levels of advertisingspending in the industry.
Note that in the monopoly case, m > s acts likeadvertising, whereas in the competitive case, onlym2 > m1 + max801 s − �m19 has an effect on profits,which results from savings on advertising costs. Onecan also note that even though consumers search bothstores in Case 4, and the market is fully homogeneous,the equilibrium does not end up being perfectly com-petitive. Further, even though consumers visit bothretailers, the disadvantaged retailer 1 still finds it nec-essary to use price advertising to induce consumersto buy at its store.
4.2. Shopping Experience and ProfitabilityAnother interesting implication of the advantagein shopping experience is that at high levels, anincrease in shopping experience has discontinuouspayoffs. More fully, the profit impact is summarizedin Proposition 2.
Proposition 2. The retailer with the advantage inshopping experience earns positive equilibrium profits. Fur-thermore, for low levels of shopping experience, the profitsare continuous in the offered levels of shopping experience,but for high enough levels, a retailer’s profits increase dis-continuously when its shopping experience exceeds that ofits competitor.16
The presence of shopping experience investmentsleads to positive profits for the retailer with theadvantage, whereas the profits of the disadvantagedretailer are competed away. For relatively low lev-els of shopping experience, retailer 2’s equilibriumprofit is precisely equal to the advantage in its shop-ping experience level over its competitor. The quality-like interaction between shopping experience (at lowlevels) and price advertising is further seen fromthe equilibrium profit functions in Table 1. One can
16 For an intermediate range of m of the competitor, the retailer’sprofits discontinuously increase when its shopping experiencesexceeds that of its competitor by a sufficiently high level (specifi-cally, s −�m1).
observe that for small values of shopping experience,as represented by Cases 1 and 2 (when mj < s orm2 −m1 < s − �m1), its effect on profits is similar to aquality improvement. For example, if we considereda model with no search costs, but consumer valua-tions for products at the two retailers are V +m1 andV +m2, where m1 and m2 can be seen as increases inquality, we would then obtain the same retailer profitsas in Cases 1 and 2.
In contrast, when shopping experience levels arehigh enough as to draw consumers to the store, thedifference in retailer profits (i.e., retailer 2’s profitadvantage) becomes a function of the advertisingcosts. This is most clearly seen in Case 4, whereretailer 2’s profit advantage is exactly equal to A.In this case, consumers shop around and visit bothstores regardless of purchase. This free riding resultsin shopping experience levels having no effect onprices and profits as long as the order of these levels(i.e., which retailer has the higher level) is unchanged.Because consumers always shop around both retail-ers, retailer 2’s profit advantage comes from its abilityto outbid its rival in the pricing game. This, in turn,is dependent on how easily retailer 1 can advertise,i.e., on the level of its advertising cost.
An implication of the above analysis is that thecompetitive payoffs from shopping experience invest-ments differ qualitatively from the payoffs associ-ated with product quality investments. Specifically,a retailer’s profit increases discontinuously by theorder A when its level exceeds that of the competi-tor. This is in contrast to when shopping experience isakin to product quality, in which case the firm’s pay-off increases continuously (and is the difference in thelevels) when the shopping experience level exceedsthat of its competitor. This discontinuous increase inretailer payoffs at high levels of shopping experienceimplies that even small differences across firms leadto significant differences in payoffs. This feature isreminiscent of the literature on rank-order tourna-ments (Lazear and Rosen 1981) and that on winner-take-all competition, where even small differencesin input characteristics can lead to a disproportion-ately large difference in (output) payoffs for players.An implication is that firms might overcompete inshopping experience compared with the choice of amonopoly firm. The market outcome is not perfectlycompetitive despite the fact that consumers can shoparound both retailers. Further, the fact that small dif-ferences in shopping experience can lead to signifi-cant differences in profits highlights the value of pre-cise competitive intelligence and consumer researchin the optimal design of the store environment. Asthe survey of retail industry managers conducted byRetailWire indicates (Ball and Jones 2010), the abilityto properly design in-store shopper experience is one
of the most important drivers of customer satisfactionin competitive retail markets.
We have considered a competitive environmentwhere consumers face search costs and are uncer-tain about prices, and where retailers can use priceadvertising and shopping experience enhancementsto attract consumers to the store. Whereas advertisinga sufficiently low price acts to both attract consumersto the store and alleviate consumer uncertainty aboutprices, shopping experience enhancements only serveto attract consumers to the store and do not resolveconsumer price uncertainty. The model captures con-sumer uncertainty about prices in the sense thatalthough rational consumers can infer the equilib-rium retailer strategies, the mixed pricing strategiesemployed by the retailers do not allow consumersto perfectly infer the price, and thus the strategiesthat resolve consumer uncertainty in this model servenot only the strategic (commitment) role but also aninformative role. Alternatively, from the point of viewof consumer utility (above and beyond the value ofinformation), shopping experience creates consump-tion utility, whereas price advertising has no directconsumption utility. This evokes the question as towhether shopping experience is more similar to qual-ity improvement (which creates consumer value) orto price advertising. As we argued above, the answeris that in some cases (namely, low levels of shop-ping experience), it acts like a quality improvement,whereas in other cases (high levels), it acts as a substi-tute for price advertising. This point may be seen ascontrary to expectation, as it is precisely at high levelsthat shopping experience does not correlate with theprice range or with the expected price charged by aretailer.
5. Optimal Investment inShopping Experience
Consider now the endogenous choice of investmentsin shopping experience and assume that the costs ofmaking these investments are increasing and convexand are given by the quadratic cost form C4m5= cm2.In this case, retailers in a first stage make their shop-ping experience investments. Then in the next stage,these investments are observable to consumers as wellas retailers, and the game proceeds as in the previoussection, with retailers simultaneously choosing theiradvertising and pricing strategies. We now examinesimultaneous retailers’ investment in shopping expe-rience levels and then ask how results would be dif-ferent when the investments are made sequentially.
5.1. Simultaneous InvestmentSimultaneous investment can also be interpreted asthe case in which retailers decide on their investment
in shopping experience without any knowledge of theplans of the competitor. In Proposition 3 we presentthe equilibrium in simultaneous investment strategiesand the manner in which they depend on the cost ofinvestment.
Proposition 3. When retailers simultaneously choosetheir shopping experience investments, then
(i) When c is sufficiently high,17 there exist a symmet-ric mixed strategy equilibrium in which m is uniformlydistributed on 6011/42c57 and two pure strategy asymmet-ric equilibria in which one of the retailers chooses zero mwhile the other one chooses m at the level of 1/42c5 or s,depending on which one yields the higher profits.
(ii) When c is sufficiently low, the unique equilibrium isthe symmetric mixed strategy equilibrium in m. The prob-ability with which m> s/� is P= 1 − 442c5/�564A+ s5−√
4A+ s52 − s2/�7, and the equilibrium distribution ofm starts from a uniform distribution with density 2con the initial interval 601 m̂ = 41 −P5/42c57 and endswith a “triangular” distribution, which starts from s/�and conditional on m> s/�, has cdf G4m5= cm2/4PA5−41 −P544c4A+ s5−�41 −P55/44cPA5.
Proof. See the appendix.
When creating shopping experience is sufficientlycostly for the retailers, there exist pure strategy asym-metric equilibria in which ex ante symmetric retailersendogenously differentiate in their choices of shop-ping experience in a manner reminiscent of the mod-els of quality competition (e.g., Shaked and Sutton1982, Moorthy 1988).18 However, unlike standardquality competition, in this model there is a differenti-ated retail equilibrium (with positive demand for bothretailers) even if the consumer market is homoge-neous. The presence of consumer search costs for vis-iting retailers and the possibility of price advertisingallows the firms to differentiate even in a market withhomogeneous consumers, and the extent of retail dif-ferentiation in shopping experience can increase withconsumer search costs. One can also see from part (i)of Proposition 3 that when the retailer’s cost of shop-ping experience is sufficiently large, then the level ofshopping experience can be decreasing in the extentof the retailer’s cost c. In this case shopping experi-ence has an effect that is similar to product qualityimprovements.
Given that the retailers are ex ante symmetric andthat the retailer with the advantage in shopping expe-rience earns higher profits, one may ask how, in prac-tical terms, retailers could arrive at the asymmetric
17 The explicit condition on c is c > max81/42s51 41/42s2554A + �s +√
A2 − s2 + 2A�s59.18 Here, retailers differentiate by providing more value, but theycan also do so through reducing assortment (Dukes et al. 2009) orthrough providing bundle discounts (Balachander et al. 2010).
pure strategy equilibria in shopping experience. Oneway to understand this coordination issue is thatretailers make initial plans regarding shopping expe-rience, and as a result of some exogenous randomshocks, one of the retailers is observed by both retail-ers as being ahead. This then leads to the expec-tation about which retailer would end up with thehigher shopping experience. This expectation can alsobe formed through the knowledge of past behavior(which are presently payoff irrelevant) of the rival’smanagers or their background or temperament, or asa result of small differences in costs or initial “endow-ment” of entertainment.19 This reasoning is the basisfor the literature in marketing and economics thatconcentrates on considering pure strategy asymmetricequilibria in product quality/location strategies whenthey exist and even when the firms are ex ante sym-metric (e.g., Shaked and Sutton 1982, Moorthy 1988,Kuksov 2004).
Consider now part (ii) of Proposition 3, when theretailer’s cost of shopping experience becomes suf-ficiently small. Now there exists only a symmetricmixed strategy equilibrium in shopping experience.Given the equilibrium in mixed shopping experi-ence investment strategies, one may wonder aboutthe interpretation of the assumption that consumersknow mj but not pj (unless advertised). As we havealready mentioned, shopping experience is a longer-term decision and not as easy to change (unlikeprices), and so the randomization of shopping experi-ence is completed and fixed before the pricing stage,whereas the pricing game can be seen as being shortterm and being played repeatedly.20 To elaborate onthe mixed strategy outcome in part (ii) of Propo-sition 3, consider when m is sufficiently low andm< s. Then the profit of the retailer with the advan-tage is m2 −m1, and the probability density functionof the mixed strategy has density 2c on an inter-val starting from 0, where the retailers are indifferentbetween investing at a given level or not investingat all. Retailers choose the higher levels of shop-ping experience m> s/� with the probability P. Thisprobability of choosing higher shopping experiencelevels decreases as c increases, and thus when thecosts of shopping experience are sufficiently high,the equilibrium distribution of m becomes uniform.As c decreases, the width of the lower interval with
19 Of course, such behavior has to be an equilibrium; i.e., neitherof them would want to deviate from their respective equilibriumstrategies.20 Note that it is still appropriate to use a one-period pricing modelas in the price promotions models of Varian (1980) and Narasimhan(1988), because without across-time interaction, the outcome of arepeated game would just be a repetition of the outcome of theone-period game, and the shopping experience value would thenbe known and fixed for each repetition.
uniform distribution decreases, and the firms choosehigher levels of m> s/� with greater probability andwith the conditional cdf G4m5.
As one can see from Table 1, higher levels of shop-ping experience lead to lower average levels of adver-tising when the levels are high enough. The retailerwith the higher level of shopping experience doesnot advertise at all, whereas the other firm adver-tises with some positive probability. Indeed, in thisendogenous investment case, if the cost c becomessufficiently small (so that the probability of m> s/�increases), then the firms advertise prices with lowerprobability. Taken together, these observations sug-gest an interesting implication for retail strategy:retailers that find it easier to supply shopping expe-rience will likely use less price advertising as a com-petitive instrument to attract consumers to their store.Within an industry, the more upscale retailers that sellhigher-margin assortments will likely have greaterincentive and ability to invest in higher levels ofshopping experience, and our analysis suggests thatthese are precisely the type of retailers that will useless price advertising. This is consistent with upscaledepartment stores such as Nordstrom or NeimanMarcus using less price advertising than such retailersas Macy’s or J.C. Penney.
In the equilibrium above, it is clear that whenshopping experience levels are high and equal to s/�,consumers will visit both retailers. One can then ques-tion what the benefit of further increasing the level ofshopping experience would be given that it is alreadyhigh enough to attract consumers. The answer is sim-ple: retailers now compete for the order in which con-sumers visit their stores. The retailer with the highershopping experience level will be visited first, whichthen allows this retailer to preclude price advertis-ing. The retailer with the lower level of shoppingexperience, who has lost the competition for the firstvisit, then responds in equilibrium by price advertis-ing. Price advertising cannot change the order of con-sumer shopping visits, but it can convince consumersto postpone their purchases until they visit the retailerthat advertises its price (if the price at the first retailerwas not low enough).
It is also useful to compare shopping investmentlevels under competition to the equivalent case of amonopolist selling two goods. When the cost of shop-ping experience is sufficiently high so that in the equi-librium, the levels are below s, a monopoly would setits level to 1/42c5 for one good and 0 for the other,which is then the same as what is provided by thecompeting retailers. But when the costs of shoppingexperience are lower, then competing retailers maysupply a level of shopping experience higher thanthat of a monopoly. Shopping experience now alsoacts like price advertising in inducing consumers to
visit the store. A firm facing competition—and awareof consumer search costs—cares about whether con-sumers would visit it first. Therefore retailers over-compete in the provision of shopping experience andprovide higher levels when compared to what wouldbe provided by a monopoly. Thus, in general, com-petitive retailers provide the same or higher levelsof shopping experience compared with a monopoly.This comparison is also interesting in the context ofconsumers’ potential free riding on the investmentsin shopping experience. One might ask whether thisfree riding might lead to lower investments in a com-petitive environment. The interesting conclusion isthat competitive investments are higher than those ofa monopoly precisely for the case when consumersactually do end up free riding on the retail invest-ments in shopping experience. For low levels of shop-ping experience, consumers only visit the store withthe intention to buy (and will always buy if theydid visit the store). But for high levels of shoppingexperience, consumers visit the store even if theyhave already bought. As already indicated, this fea-ture of free riding on retail shopping experience canbe thought of as realistically capturing the notion ofconsumer window shopping. From Proposition 3, wenote that the extent of window shopping can be rep-resented by P2 or the probability with which bothretailers will choose m > s/�. The extent of windowshopping increases with the advertising costs. This isintuitive because as advertising costs go up, the retail-ers have the incentive to supply even higher levelsof shopping experience, and this leads to higher inci-dences of consumers visiting a retailer just for theshopping experience. The extent of window shoppingdecreases with the extent of consumer search costs(as long as advertising costs are large enough) butincreases when � is higher (or in other words, con-sumers experience less fatigue when shopping acrossthe stores).
5.2. Sequential InvestmentsThe simultaneous timing assumption can also be seenas a process of shopping experience investments inwhich firms do not have information about theircompetitor’s investment actions while making theirdecision. Frequently, however a retailer will haveinformation on the investments made by an incum-bent. Or one of the retailers might already havedecided its layout or locate itself closer to complemen-tary assets such as a popular café or movie theater.In these cases the sequential timing of investmentsseems to be more relevant. With sequential timing,the equilibrium is in pure strategies and asymmetricwith the first mover choosing the higher investmentlevel. When costs of shopping experience are suffi-ciently high, the equilibrium is the same as the pure
strategy one reported in Proposition 3 for high c (withthe first mover selecting the positive level of m). How-ever, when the cost of shopping experience is low,then there continues to exist a pure strategy equilib-rium in which the retailers differentiate even more.In this case the retailers are able to invest at a levelgreater than s/�. The first retailer (say, retailer 2) theninvests at a level such that retailer 1 does not havethe incentive to “outbid” its investment. In this case,�2 = s and �1 = 0, and so the profit advantage of thefirst mover is exactly equal to the search cost.
6. ExtensionsIn this section, we discuss some additional factors thatare observed in retail markets where shopping expe-rience plays a role, and we analyze how these factorsinfluence the strategies of retailers.
6.1. Product Fit Uncertainty and DifferentiationUntil now, the analysis assumed that retailers haveundifferentiated products. We now ask how retaildifferentiation would affect our results. Furthermore,although we considered consumer uncertainty aboutprices, we did not consider uncertainty in consumerproduct valuations. Both of these possibilities can bejointly addressed by assuming that consumers faceuncertainty about the product fit prior to their visitinga retail store. The fit is an independent variable acrossretailers and, like price, is known only at the point ofpurchase. Specifically, assume that the probability offit of either product is f , so that upon inspection of theproduct at a retailer, with probability f , a consumerrealizes that the product fits her preferences and thushas a value of V , and with probability 1 − f , the con-sumer realizes that the product does not fit and has avalue of 0.
The analysis and the subcases of this model are sim-ilar to the subcases that we analyzed in §3, which is aspecial case of this model when f = 1. To understandthe relation of the outcomes with uncertain productfit to the previous outcome with f = 1, it is useful tothink of s/f as the adjusted search cost when con-sumers are searching for a product (with uncertainvaluations) at a retailer, whereas the search cost iss if the consumers are contemplating a visit to theretailer in order to consume the shopping experience.The former is because when searching for a prod-uct, a consumer compares the search cost s to theexpected benefit of f 4V − p5, which is equivalent tothe comparison of s/f and V − p. Further, for eachretailer, a fraction of (1−f ) of consumers who did notfind a fit at the rival retailer are potentially loyal con-sumers. Among these consumers, those who have notyet visited the retailer would have an expected valu-ation of fV , whereas those who have already visitedknow whether or not the retailer’s product fits their
preferences. Thus if f 41 − f 5V is high relative to Aand s, then retail competition is not as intense, andboth retailers will have positive profits regardless ofthe values of the shopping experience they provide tocustomers.
As we know from the main model, when the costof shopping experience is high (so that the equilib-rium levels of m are below s), competition resultsin similar levels of shopping experience comparedwith a (two-product) monopoly. On the other hand,when the cost is low, competition results in choicesof m that are higher than that of a monopoly butlower than what is socially optimal. One could thenhypothesize that in the model with uncertain prod-uct fit, the equilibrium shopping experience would bebetween that of the competitive and the monopolycases and always below what is socially optimal.However, this hypothesis turns out to be not true.In fact, as long as fV is high relative to s, uncertainfit may increase competing firms’ incentives to investin shopping experience, and the expected equilibriuminvestment in shopping experience can exceed notonly the monopoly but also the socially optimal level.This happens for some parameter values when thecost of shopping experience provision is high enough(so that the equilibrium levels are below s). Specifi-cally, we have the following result.
Proposition 4. When there is product fit uncertainty(f < 1), the competitive incentive to invest and the equi-librium investment in m are not monotone in f . In par-ticular, while m → 0 when f → 0, we also have that insome parameter range, the equilibrium investment in mis, on average, higher than both the monopoly and sociallyoptimal levels. For example, for small enough A and largeenough V and c (relative to all other parameters and 1/f ),the average expected equilibrium m is higher than thesocially optimal one by 41 −�541 − f 5/44c5.21
Consider the consumers’ decision of which retailerto visit first for the case when they would notvisit either retailer for the shopping experience util-ity alone. If both stores advertised, and consumersobserve the prices, then they should plan their shop-ping strategy as follows: First, they should go to theretailer that provides the higher expected utility andbuy there if the product is a fit with their preferences.If the product at the first retailer turns out to not bea fit, then they should go to the other retailer andbuy there if the product there is a fit. This strategy isoptimal as long as the prices are low enough, so thatgoing to the rival store if there was no fit at the first
21 As we have noted above, although there are pure and mixedstrategy equilibria in m when c is large, the average expected equi-librium m turns out to be the same in all equilibria. The optimal(two-product) monopoly m is equal to the socially optimal one inthis case.
store is optimal (which we have assumed holds inthe equilibrium as long as the prices are advertised).The relevant condition for consumer indifference inthe order of visitation is
f 4V − p15+m1 − s + 41 − f 54−s +�m2 + f 4V − p255
= f 4V − p25+m2 − s
+ 41 − f 54−s +�m1 + f 4V − p1551 (10)
which implies that the price differential that makesconsumers indifferent is
ãf =4m2 −m1541 −�+ f�5
f 23 (11)
i.e., the low-m store must price lower by ãf to makeconsumers indifferent if both prices are advertised atlevels sufficient enough to entice a consumer to visitthe store instead of forgoing the purchase opportunityaltogether. For example, when � = f = 1/2, then ã =
24m2 − m15; i.e., the price differential needed for thelow-m store to entice consumers to visit its store firstis twice as large as in the case of certain fit.
Given that a retailer benefits from attractingconsumers to visit its store first, we have that whenf < 1, it is harder for a retailer to overcome the advan-tage in m of competing retailer through charging alower price. Therefore, the advantage in m is morebeneficial to the retailer when f < 1, and under someparameter ranges, this may lead to investments thatare greater than the socially optimal one.22 Of course,when f tends to 0, the consumer expected valuefV tends to 0 as well and eventually does not jus-tify the expenditure on shopping experience; i.e., thearguments will no longer apply since fV is not highenough relative to A and/or s, and in this case, theoptimal m will tend to 0.
Whereas the main effects of consumers’ shoppingexperience on the firm’s strategy are robust to thefit uncertainty consumers face, Proposition 4 showsthat competition can lead to excess investment inshopping experience when product fit is uncertain.Recall that the investment was never higher thanthe socially optimal level in the main model withperfect fit. Proposition 4 suggests that one can findexcessive investment in shopping experience com-pared with the socially optimal level in retail envi-ronments where consumers are sufficiently uncertain
22 Note that for a full proof, one also needs to consider how theretailer’s benefit of enticing consumers to come to its store firstdepends on f . This effect is a product of the market share benefit,which is f − f 41 − f 5 = f 2 and the margin (which increases as fdecreases). Since ãf is of the order of f 2, the declining benefit fromthe market share effect would exactly cancel out the price advan-tage effect ãf if the margin effect were not to exist. Note that themargin effect comes from the incentive of each retailer to go afterthe f 41 − f 5 “loyal” consumers.
about product fit. This is consistent with the observa-tion that stores in shopping malls, which sell productcategories such as fashion goods or apparel (where fitis uncertain), invest more in atmospherics and con-sumer shopping experience than stores such as super-markets, which sell frequently purchased goods (e.g.,grocery products). For frequently purchased groceryproducts, although retail prices may not be observedwithout advertising, product values would be morecertain for consumers.
6.2. Heterogeneous Product QualitiesThe parameter V , which represents the valuation ofthe product, has been uniform across retailers. Let usconsider how relaxing this assumption would affectthe role of shopping experience and retailers’ strate-gies. Accordingly, suppose that retailers are differen-tiated in quality, and the consumer valuation for theproduct quality at retailer i is Vi.
We assume that consumers observe retailers’ qual-ity levels before the pricing stage, and so quality deci-sions are assumed to be more long term than theprice, just as in the case of the shopping experiencedecision. Consumers know the quality before visit-ing the retailers but only know the price if it wasadvertised. Note that in the extension with uncertainfit, we had assumed that fit was not known beforevisiting the store. This is to account for the fact thatwhereas the overall quality of a store (e.g., Macy’s andJ.C. Penney versus Nordstrom and Neiman Marcus)is a long-term decision, the fit is product specific andshort term.
At the one extreme, we have Case 1 of the mainmodel, where the levels of shopping experience atboth retailers are low enough so that they do not com-pensate for the search costs (m1 < m2 < s). In otherwords, the cost of shopping experience is sufficientlyhigh enough. Recall that retailers now compete for theconsumers at home to induce them to visit the store. Ifthe high-quality retailer now also has the higher shop-ping experience (i.e., V2 >V1), then the equilibrium issimilar to that in §3: the equilibrium profits are �2 =
ã+ãV 3�1 = 0, where ãV = V2 −V1 is retailer 2’s qual-ity advantage. Retailer 2’s equilibrium profits nowreflect the advantage not only in shopping experi-ence but also in quality, and the two variables actin a similar fashion on profits. Further, as mentionedbefore, retailer 2—because of its higher profitability—always advertises, whereas retailer 1 advertises withprobability less than 1. Suppose now that V2 < V1and ãV < 0, and the high-quality retailer 1 has alower level of shopping experience (and −ãV nowdenotes retailer 1’s quality advantage). The equilib-rium now depends on the relative magnitudes ofthe quality and shopping experience advantages. If−ãV <ã, then retailer 2’s advantage in shopping
experience dominates retailer 1’s quality advantage,and the equilibrium profits are once again of the form�2 = ã+ãV 3�1 = 0. However, if −ãV >ã and if it isretailer 1’s quality advantage that dominates, we havethat �2 = 03�1 = −4ãV +ã5. The overall point is thatadvantages in shopping experience and quality aresubstitutes in retail profits, and the retailer with thegreater overall advantage is the one that has positiveequilibrium profits.
Next consider the other extreme of Case 4 (�m2 >�m1 > s) in which shopping experience levels at bothretailers are so high that consumers visit them regard-less of purchase. Now if V2 > V1, then the natureof the equilibrium is similar to the one in the basicmodel except that retailer 2’s equilibrium profits alsoreflect the quality advantage (�2 = A + ãV > A3�1 =
0). However, if V2 < V1, then we will still have thatretailer 2 has higher equilibrium profits if the qualityadvantage for retailer 1 is not too high or V1 −V2 <A.In this case retailer 1 still advertises with probabil-ity less than 1, and the equilibrium profits are �2 =
A+ãV <A3�1 = 0. Notice, however, that retailer 2’sprofits are lower than A, reflecting retailer 1’s qualityadvantage. In contrast, if retailer 1’s quality advan-tage is sufficiently large and −ã = V1 − V2 > A, wehave a (pure) equilibrium in which the high-qualityretailer 1 can have higher equilibrium profits despitehaving a lower level of shopping experience. Giventhis high level of V1, retailer 1 has the incentive toalways advertise and charge a price of −ãV , and theequilibrium profits are �1 = −4ã + A53�2 = 0. Thepoint is, despite the fact that retailer 2 has won thecompetition in having consumers visit its store first,retailer 1 has enough of a quality advantage to alwaysprice advertise and win over the consumers.
7. Discussion and ConclusionRetail activities and investments that enhance theshopping experience of consumers are ubiquitousand an important aspect of competitive retail strat-egy. These investments range from store atmospher-ics variables such as lighting and music to providingenjoyable experiences through entertainment andeducation. And although the recognition of the impor-tance of shopping experience and store atmosphericsas a retail instrument dates back at least to Kotler(1973), to the best of our knowledge, there exists noresearch that analyzes the competitive implications ofthis element of retailing. In this paper we examinehow retailer advantage in consumer shopping experi-ence affects consumer shopping and competitive retailmarket outcomes.
We analyze two aspects of investments in shoppingexperience: First, retail investments in shopping expe-rience have a public good–like nature and are sub-ject to consumer free riding. Consumers can enjoy
the shopping experience and entertainment opportu-nities even if they do not purchase the product. Sec-ond, if the role of consumer shopping experience is toattract consumers, then another commonly used strat-egy of price advertising would also perform this rolefor retailers. This raises the question of why retail-ers would prefer to use shopping experience invest-ments rather than lower advertised prices to attractconsumers to the store. In analyzing the effects ofshopping experience, we investigate the role playedby this variable compared with retail price advertis-ing and highlight the essential economic differencesbetween these variables. Whereas price advertisingacts to attract consumers to the store by alleviat-ing consumer price uncertainty, shopping experienceenhancements only serve to attract consumers to thestore but do not directly resolve consumer uncertaintyabout prices. Further, shopping experience invest-ments create direct consumer utility, whereas priceadvertising has no immediate consumption utility. Inthis context, we ask the question as to whether shop-ping experience is more similar to price advertisingor to quality improvement (which creates direct con-sumption utility).
Our analysis shows that if the consumer valueof shopping experience is sufficiently low, its effecton retailer strategy is similar to that of quality, andthen the retailer with the advantage in shoppingexperience also deploys higher levels of price adver-tising. On the other hand, when consumers valueshopping experience at high enough levels, it actsakin to price advertising in that it makes it opti-mal for the retailer with the advantage in shoppingexperience to do away with price advertising. Athigh levels, competition in shopping experience hasa winner-take-all characteristic, and a small advan-tage can lead to large increases in equilibrium profits.Upon considering the optimal investment strategies,we show that competitive investments in shoppingexperience may be higher than that of a monopolywhen the costs are low enough. Further, there is asuggestion that competition can lead to greater invest-ments in shopping experience compared with boththe monopoly and the socially optimal levels in mar-kets where consumers are sufficiently uncertain aboutproduct fit.
Our analysis of the interaction between shoppingexperience and other critical retail variables such asprice advertising and quality provides testable impli-cations for competitive retailing strategy. First, wefind that when the cost of providing shopping expe-rience goes down, or alternatively, when retailersfind it easier (or have greater incentives) to supply
shopping experience, then the incentives to deployprice advertising would decrease, and so retailerswould advertise with lower probability. One wayto empirically test this insight would be to identifyacross-market variations in cost and in ease of sup-plying shopping experience for retailers that operatein multiple markets. For example, the supply costof shopping experience and entertainment may berelated to real estate costs, the costs of hiring trainedstore staff, and the costs of operating experience-related activities, which can vary across different citiesor markets. It might then be interesting to investigatehow within-retailer pricing and advertising strategiesinteract with the provision of shopping experienceacross these markets.
A more nuanced prediction involves the correla-tion between which one of the competing retailers hasthe higher level of shopping experience and whichone uses more price advertising. Our prediction isthat this relationship is not monotonic: for low lev-els of shopping experience, the advantage in shop-ping experience would correlate positively with priceadvertising, whereas at the high enough levels, itwould correlate negatively. Note that to test this pre-diction, one does not need to estimate the costs (oreffectiveness) of price advertising and the investmentin shopping environment. Instead, what one needs isa measure of what constitutes a high enough level ofshopping experience as well as a measure of which ofthe competing retailers has the advantage in shoppingexperience. The latter could be potentially obtainedthrough a survey, whereas an empirically measurableproxy for the former could be the propensity of con-sumers to window shop, i.e., visit a store without theintention to buy.23
The discontinuity of the advantage in shoppingexperience on profits could also be empirically testedthrough different specifications of the functional formof retail profitability as a function of the advantagein shopping experience and whether the level is highenough to result in window shopping. Note that insuch an empirical analysis, one has to be careful aboutaccounting for consumer heterogeneity in the enjoy-ment of shopping experience and shopping costs.
Our analysis also provides testable implications forhow the degree of product fit uncertainty affects the
23 Another prediction, albeit a more straightforward one, is that theextent of window shopping increases when shopping experiencebecomes easier to supply and also when the shopping environ-ment is such that consumer enjoyment at the stores that are visitedlater is not significantly attenuated. Further, window shopping canincrease as search costs go down or if the firms find it costlier toadvertise. For an empirical analysis of this phenomenon, one mayneed observational or survey data that record consumers’ shoppingbehavior (over and above the purchase decision) once they are in aretail location, data on retail pricing and advertising, and informa-tion on retail or mall characteristics.
competitive incentives to invest in shopping experi-ence activities. As we argue in §6.1, one may expectretailers or shopping malls specializing in apparel andother goods—where fit is uncertain—to have higherincentives to invest in shopping experience. First, inthe presence of product fit uncertainty, retailers wouldbe more likely to overinvest in shopping experienceactivities in markets where they face competitioncompared with markets where they have monopolypower. Second, the increased incentive to provideshopping experience when fit is more uncertain couldbe seen as an interesting prediction by itself, becauseone could have hypothesized that the uncertain fitwould facilitate search, and retailers would not needto invest in shopping experience.
The welfare implications of shopping experienceinvestments also are interesting to investigate fur-ther. As we noted before, retailers’ investments inshopping experience not only encourage search butalso provide consumption utility independent of pur-chases. This latter aspect can lead to window shop-ping when retailers compete in the provision of storeentertainment activities. An important policy matteris the optimality of the extent of window shoppingfrom the social welfare standpoint and its implica-tions for the household economy (for example, howit substitutes for other productive household activi-ties). Empirical analysis of this phenomenon seemsespecially relevant for fast-growing retail markets incountries such as China and India, where, at leastanecdotally, the consumption of shopping experienceis on the increase.
AcknowledgmentsThe authors thank the editor-in-chief, the associate editor,and two anonymous reviewers for their comments. Theyalso thank seminar participants at the Cheung Kong Grad-uate School of Business, National University of SingaporeBusiness School, University of Texas at Dallas, and Univer-sity of Toronto for comments.
Appendix
Proof of Proposition 3(i). When m1 <m2 < s+ 41 −�5m1,then �1 = 03�2 = m2 − m1. Retailer 2 would then maxi-mize ç2 = m2 − m1 − cm2
2. Consider a candidate pair m2 =
1/42c53m1 = 0 as a possible equilibrium. This would bethe case as long as 1/42c5 < s and if retailer 1 has noincentive to deviate from 0. Note that 1/42c5 < s impliesc > 1/42s5. Given m2, retailer 1 has no incentive to devi-ate to any md
1 ≤ s as it makes negative profit. We canalso rule out any deviation to md
1 > s + 41 − �5/2c if c >4A+�s +
√
A2 − s2 + 2A�s5/42s25.There also exists a symmetric mixed strategy equilibrium
in m. Firm i’s profit function equals
�i4m5=
∫ m
04m− x5f 4x5dx− cm20 (12)
From the first-order condition d�i4m5/dm = 0, we have thatthe symmetric equilibrium distribution is F 4m5= 2cm or thatf 4m5 = 2c. The symmetric equilibrium distribution will nothave any mass point, because otherwise, a firm can alwaysdeviate and move some of the mass to a slightly higher mand be better off. The equilibrium profits can be derived bynoting that the profits for every choice of m in the strategyspace should be the equal to that which is obtained for thechoice of m= 0 and is therefore equal to 0. �
Proof of Proposition 3(ii). When 1/42c5 > s (or c <1/42c5), then there exists no pure strategy equilibrium. Con-sider the case in which the costs are so small that retailerscan choose a level of shopping experience that ensures thatconsumers will visit the retailer regardless of purchase, ormi ≥ s/�. In this case, the mixed strategy equilibrium willinclude an interval that is close to 0—say, 601 m̂7—as wellas an interval mi ≥ s/�.24 Let P be the probability of thefirms choosing m> s/�. Given that in the symmetric equi-librium there are no mass points, the probability of m= s/�is 0. As in part (i) above, in the interval close to 0, the pay-off for retailer i with higher mi is 4mi −mj5, and the profitfunction is similar to (12). Then the distribution must havedensity 2c on 601 41−P5/2c7 and probability P that m> s/�with conditional cdf G4m5. We already checked the indiffer-ence between the points of the interval and 0. We can nowwrite the indifference condition between any m > s/� andm= 0 as
−cm2+
∫ 41−P5/2c
04A+s−�x5·2cdx+P
∫ m
s/�AdG4x5=00 (13)
From (13), we can obtain G4m5 = cm2/4PA5 − 41 − P5 ·
44c4A+ s5− �41 −P55/44cPA5. Then, P can be determinedfrom the condition G4s/�5 = 0, and we can derive it to beP= 1 − 42c/�564A+ s5−
√
4A+ s52 − s2/�7 �.
Proof of Proposition 4. To prove the proposition, it isenough to consider the equilibrium in the parameter rangewhere V and/or 41 − f 5 are high enough so that a retailerwould always like to attract consumers who shop first atthe other store and encountered lack of product fit there. Inother words, we assume that the highest advertised price ofretailer j is at most
Bj = V −s −�mj
f0 (14)
This pertains to the condition for maximum market cov-erage. We provide only the derivations for the casem1 <m2 < s, because that is enough to prove the claims inthe proposition (of course, the equilibrium investment levelwill not fall out of this range when c is high enough).The derivations in this case are similar to ones in the mainmodel, and we can show that retailer 2 always advertises(�2 = 1), whereas retailer 1 will either advertise with prob-ability less than 1 if f 41 − f 5V < A or will always adver-tise otherwise. Recall that the price differential necessaryto make consumers at home indifferent between going toretailer 1 first and going to retailer 2 first is derived in the
24 Note that for simplicity of the analysis, we are assuming that c issmall enough so that there is no density on s <m< s/�.
main text as ãf = 4m2 − m1541 − � + f�5/f 2 (note that thisassumes that both retailers’ prices are advertised at Bj orbelow).
Consider first the case when advertising cost is high(A > f 41 − f 5B1), so that retailer 1 does not always adver-tise. We have that B2 − ãf < B1, and so the highest priceof retailer 1 is then p̄1 = B2 − ãf , and the highest price ofretailer 2 is p̄2 = B2. Furthermore, similar to Case 1 of themain model, retailer 2 always advertises and has a masspoint at p̄2, whereas retailer 1’s pricing distribution has nomass points. The indifference between advertising/pricingand zero profits for retailer 1 and the indifference betweendifferent prices for retailer 2 lead to the following equilib-rium conditions:
f 41 − F24p1 +ãf 55p1 + f 41 − f 5F24p1 +ãf 5p1 −A= 01f 41 −�15p2 + f�141 − F14p2 −ãf 55p2
+�1f 41 − f 5F14p2 −ãf 5p2 −A
= f 41 −�15p̄2 +�1f 41 − f 5p̄2 −A1
(15)
where �1 is determined from the condition that the priceat which F24 · 5 = 0 is ãf higher than the price at whichF14 · 5= 0. Solving the above equations, we obtain
F14p5=1
f�1−
p̄241 − f�15
f�14p+ãf 5
for p ∈ 4A1 p̄151 where �1 =1f
−A− f p̄2
f 2p̄23
F24p5=1f
−A
f 24p−ãf 5for p ∈ 4A+ãf 1 p̄250
(16)
Furthermore, retailer 2 has a mass point of M2 at p̄ = V −
4s − �m25/f . Thus, in this case, the profit of retailer 1 is 0,and the profit of retailer 2 is qãf . To revisit the V highenough condition, note that for the above to be the equilib-rium, we need
V −s −�m2
f>A+
4m2 −m1541 −�+ f�5
f 2
so that the lower bounds on the price distributions wouldbe lower than Bj . This condition is satisfied when, say,V >A+ 4m2 −m15/f
2, or when V > A+ s/f 2, for example.Also, for the maximum coverage condition to be satisfied,in addition to V sufficiently high, 41 − f 5V should be inthe order of 41 − �5m or larger. One can show that if this isnot true, the price distributions derived above would not becorrect, but the profits would still be the same as derivedabove.
Now, consider the case when A is low (A< f 41 − f 5B1).Then, both retailers always advertise. In this case, retailer 1has a mass point at its upper bound p̄1 = B1, and retailer 2also has a mass point at its upper bound p̄2 = B2 (at theupper bound of its price distribution, retailer 2 receives thedemand from consumers who went to retailer 1 first but didfind product fit). Solving for the equilibrium price distribu-tions in a manner similar to the above analysis, we obtainthat the profit of retailer 1 is f 41 − f 5p̄1 −A and the profit
of retailer 2 is fãf higher than that of retailer 1. Thus, theprofits are
�1 = max801 41 − f 54fV − s +�m15−A91
�2 = �1 + fãf =�1 +4m2 −m1541 −�+ f�5
f
= �1 + 4m2 −m15
(
1 +41 −�541 − f 5
f
)
1
(17)
as long as V >A+ s/f 2.To prove that the equilibrium investment in shopping
experience may be above the socially optimal one, considerthe equilibrium investment in m when A = 0 and V highenough. The above equation on equilibrium profits impliesthat the marginal benefit of increasing m by � for the retailerwith the lower m is 41−f 5��, whereas the benefit of increas-ing m by � for the retailer with the higher m is �41 − � +
f�5/f . Given the cost 2c� of increasing m, we then obtainthat the symmetric equilibrium m is uniformly distributedon 641 − f 5�/42c51 41 − � + f�5/42fc57 as long as c is highenough so that the equilibrium levels of m do not exceed s.There is also an asymmetric equilibrium in pure strategiesin which one retailer chooses m= 41 −�+ f�5/42fc5 and theother chooses m= 41−f 5�/42c5. In all equilibria, the averageacross stores expected level of m is the same.
Now, consider the optimal monopoly investment in mwhen A= 0. To make the most efficient investments in shop-ping experience of the two stores under its control, it isoptimal for the monopoly to induce all consumers to searchin one store—say, retailer 2—first and search in retailer 1only if they did not find product fit in retailer 2. Then themarginal benefit of increasing m by � in retailer 2 is the ben-efit of increasing price by �/f while keeping the same salesof f ; i.e., it is �. Therefore, the optimal m2 for a monopolyis 1/42c5. For retailer 1, the benefit is one of increasing theprice by ��/f while keeping the same sales of f 41 − f 5.Therefore, the optimal m1 for a monopoly is �41 − f 5/42c5.
Thus, the expected average m under competition is largerthan the monopoly average m by 41 − �541 − f 5/44c5 > 0.For example, when c = 1 and �= f = 1/2, the average com-petitive m is 004375 and the average monopoly m is 003125.Note also that the lowest competitive m is the same as thelowest monopoly m, but the highest competitive m is higherthan that of monopoly. Finally, note that the monopoly pro-vides the socially optimal level of m in this case because themonopoly appropriates all the social surplus. �
ReferencesAmaldoss W, He C (2010) Product variety, informative advertising,
and price competition. J. Marketing Res. 47(1):146–156.
Anderson SP, Renault R (1999) Pricing, product diversity, andsearch costs: A Bertrand-Chamberlin-Diamond model. RANDJ. Econom. 30(4):719–735.
Baker J, Levy M, Grewal D (1992) An experimental approachto making retail store environmental decisions. J. Retailing68(4):445–460.
Balachander S, Ghosh B, Stock A (2010) Why bundle discounts canbe a profitable alternative to competing on price promotions?Marketing Sci. 29(4):624–638.
Ball B, Jones R (2010) Investing in the shopper experience: Weign-ing customer satisfaction vs. ROI. RETAIL:NEXT Study 3,RetailWire/Dechert-Hampe, San Juan Capistrano, CA. http://www.retailwire.com/public/sponsors/intel/downloads/20100401/ShopperExp_ExecSum.pdf.
Brekke KR, Siciliani L, Straume OR (2010) Price and quality in spa-tial competition. Regional Sci. Urban Econom. 40(6):471–480.
Burdett K, Judd K (1983) Equilibrium price dispersion. Econometrica51(4):955–969.
Butters G (1977) Equilibrium distributions of sales and advertisingprices. Rev. Econom. Stud. 44(3):465–491.
Chu J (2002) What top-performing retailers know about satisfyingcustomers: Experience is key. IBM Institute for Business Valuereport, IBM, Somers, NY.
Desai PS, Krishnamurthy A, Sainam P (2010) “Call for prices”:Strategic implications of raising consumers’ costs. MarketingSci. 29(1):158–174.
Diamond PA (1971) A model of price adjustment. J. Econom. Theory3(2):156–168.
Guo L, Zhao Y (2009) Voluntary quality disclosure and marketinteraction. Marketing Sci. 28(3):488–501.
Gorn GJ (1982) The effects of music in advertising of choicebehavior: A classical conditioning approach. J. Marketing46(1):94–101.
Gorn GJ, Goldberg ME, Basu K (1993) Mood awareness and prod-uct evaluation. J. Consumer Psych. 2(3):237–256.
Grossman G, Shapiro C (1984) Informative advertising with differ-entiated products. Rev. Econom. Stud. 51(1):63–81.
Hess J, Gerstner E (1987) Loss leader pricing and rain check policy.Marketing Sci. 6(4):358–374.
Iyer G, Kuksov D (2010) Consumer feelings and equilibrium prod-uct quality. J. Econom. Management Strategy 19(1):137–168.
Iyer G, Soberman D, Villas-Boas JM (2005) The targeting of adver-tising. Marketing Sci. 24(3):461–476.
Janssen MCW, Non MC (2009) Going where the ad leads you: Onhigh advertised prices and searching where to buy. MarketingSci. 28(1):87–98.
Kotler P (1973) Atmospherics as a marketing tool. J. Retailing49(12):48–64.
Kuksov D (2004) Buyer search costs and endogenous productdesign. Marketing Sci. 23(4):490–499.
Kuksov D (2006) Search, common knowledge, and competition.J. Econom. Theory 130(1):95–108.
Kuksov D, Lin Y (2010) Information provision in a vertically differ-entiated competitive marketplace. Marketing Sci. 29(1):122–138.
Lal R, Matutes C (1994) Retail price and advertising strategies.J. Bus. 67(3):345–370.
Lal R, Rao R (1997) Supermarket pricing: The case of everyday lowpricing. Marketing Sci. 16(1):60–80.
Lazear E, Rosen S (1981) Rank-order tournaments as optimum laborcontracts. J. Political Econom. 89(5):841–864.
Levy M, Weitz BA (2009) Retailing Management, 7th ed. (McGraw-Hill, New York).
Mathewson GF, Winter R (1983) The incentives for resale pricemaintenance under imperfect information. Econom. Inquiry21(3):337–348.
Milliman RE (1982) Using background music to affect the behaviorof supermarket shoppers. J. Marketing 46(3):86–91.
Mitchell DJ, Kahn BE, Knasko SC (1995) There is something in theair: Effects of congruent and incongruent ambient odor on con-sumer decision-making. J. Consumer Res. 22(2):229–238.
Moorthy KS (1988) Product and price competition in a duopoly.Marketing Sci. 7(2):141–168.
Narasimhan C (1988) Competitive promotional strategies. J. Bus.61(4):427–449.
Ofek E, Katona Z, Sarvary M (2011) “Bricks and clicks”: The impactof product returns on the strategies of multichannel retailers.Marketing Sci. 30(1):42–60.
Peck J, Childers TL (2008) Effects of sensory factors and consumerbehavior. Haugtvedt CP, Herr PM, Kardes FR, eds. Handbookof Consumer Psychology (Taylor and Francis Group, PsychologyPress, New York), 193–220.
Prior M (2001) Toys “R” Us Times Square store raises retailtainmentbar. DSN Retailing Today 40(1):46.
Rajiv S, Dutta S, Dhar SK (2002) Asymmetric store positioning andpromotional advertising strategies: Theory and evidence. Mar-keting Sci. 21(1):74–96.
Salop S, Stiglitz J (1977) Bargains and ripoffs: A model of monop-olistically competitive price dispersion. Rev. Econom. Stud.44(3):493–510.
Scardino E (2005) Revamped FAO tempers pedigree with fun. DSNRetailing Today 44(3):22–27.
Schwartz N, Clore GL (1983) Mood, misattribution, and judgmentsof well-being: Informative and directive functions of affectivestates. J. Personality Soc. Psych. 45(3):513–523.
Shaked A, Sutton J (1982) Relaxing price competition through prod-uct differentiation. Rev. Econom. Stud. 49(1):3–13.
Shin J (2007) How does free riding on customer service affect com-petition? Marketing Sci. 26(4):488–503.
Shulman JD, Coughlan AT, Savaskan RC (2011) Managing con-sumer returns in a competitive environment. Management Sci.57(2):347–362.
Soberman DA (2004) Additional learning and implicationson the role of informative advertising. Management Sci.50(12):1744–1750.
Spangenberg ER, Crowley AE, Henderson PW (1996) Improvingthe store environment: Do olfactory cues affect evaluations andbehaviors? J. Marketing 60(2):67–80.
Telser LG (1960) Why should manufacturers want fair trade? J. LawEconom. 3(October):86–105.
Tirole J (1988) The Theory of Industrial Organization (MIT Press,Cambridge, MA).
Turley LW, Milliman RE (2000) Atmospheric effects on shoppingbehavior. J. Bus. Res. 49(2):193–211.
Varian H (1980) A model of sales. Amer. Econom. Rev. 70(4):651–659.Wernerfelt B (1994a) On the function of sales assistance. Marketing
Sci. 13(1):68–82.Wernerfelt B (1994b) Selling formats for search goods. Marketing
![[EXCERPT] Holiday 2016: Consumer Shopping Survey](https://static.documents.pub/doc/80x56/588045791a28abfd0a8b6c6f/excerpt-holiday-2016-consumer-shopping-survey.jpg)