Impact of Web-based e-Commerce on Channel Strategy in Retailing

You don't have the latest version of Adobe Flash Player.

Please update your flash player.

Please login or register to make a comment!

International Journal of Electronic Commerce / Spring 2004, Vol. 8, No. 3, pp.

103 3130. Copyright © 2004 M.E. Sharpe, Inc.

All rights reserved. 1086-4415/2004 $9.50 + 0.00. Impact of Web-based e-Commerce on Channel Strategy in Retailing Ruth C.

King, Ravi Sen, and Mu Xia ABSTRACT: Web-based channels are fast becoming an integral part of the channel strat- egy of traditional off-line retailers. This paper uses a game-theoretic approach to study the impact of Web-based e-commerce on retailers 9 choices of distribution channel strategies. At present most firms adopt a multi-channel strategy that includes both Web-based channels and pre-existing off-line channels.

The analysis identifies this trend as an equilibrium of the game resulting from competitive pressure by other retailers, but shows that it is not the only possible short-run outcome. Other possible outcomes are: all sellers continue to sell via traditional off-line channels; some sellers adopt a coordinated dual-channel strategy (i.e., both on-line and off-line) when faced with competition from Web-based entrants; off-line sellers continue to sell via off-line channels when faced with purely on-line competition. The conditions for these equilibriums depend on the proportion of buyers unlikely to buy on-line and their level of discomfort with on-line channels.

Traditionally off-line sellers can pre- empt ... more. less.

competition from on-line channels by improving such features as convenience, per- sonalized after-sale service, and trust. In so doing, they will also increase the factor of buyer discomfort associated with relatively new and unknown on-line channels. KEY WORDS AND PHRASES: Distribution, e-commerce, electronic markets, off-line chan- nel, on-line channel, Web-based channel, Web-based selling. Web-based retailing is rapidly gaining acceptance by American Internet users as a valid retail outlet. A growing number of households are connected to the Internet, and thus the potential for shopping on the Internet is greater than ever. Jupiter Communications [20] estimated that more than a quarter of U.S. adult Internet users made an on-line purchase in 1997. More recent numbers by eMarketer.com (2002) suggest that the number of U.S. on-line buyers has grown from 53.2 percent of all Internet users in 2001 to 60 percent in 2004 (90 million people purchasing on-line). Furthermore, U.S. revenues from on-line B2C buying are predicted to reach $190 billion in 2004, with an average on-line shopper spending over $300 per quarter. In The State of Retailing Online 6.0 , a more recent Shop.org study of 130 retailers conducted by Forrester Research, on-line retail sales soared to $76 billion in 2002, an increase of 48 percent from the previous year [12]. Given the increasing acceptance of on-line chann els, it is not surprising that more sellers are including Web-based channels in their general channel strategy. This paper defines a seller as any business en- tity that sells goods or services directly to end-consumers. Under this defini- tion, both traditional retailers and manufacturing companies are sellers if they sell their final products and goods to consumers. Besides providing sellers with an opportunity to reach a wider audience, Web-based sales channels also give them better control of their sales and mar- keting activities [34], support existing off-line channels (e.g., Marshall Indus- tries [34]), divert competition from new Web-based entrants in their market (termed creintermediation d by Chircu and Kauffman [11]), and reduce transac- 104 KING, SEN, AND XIA tion costs [2, 6, 16, 22]. However, incorporating Web-based channels in an over- all channels strategy raises some concerns. For example, the lack of prior expe- rience in direct selling (because the Web virtually makes the seller a direct seller) may increase rather than decrease the transaction cost of selling a product. This is especially critical for sellers that have traditionally sold through retail stores (e.g., Toys-R-Us, Best Buy, Circuit City). Additionally, the cost of managing the channel conflicts that may arise from a multichannel strategy may be prohibi- tive for some sellers. Given the various advantages and disadvantages of Web- based channels for sellers, it is necessary to address several questions: " Under what conditions should traditional off-line sellers (retailers) incorporate Web-based channels in their overall channel strategy? " Should a new channel strategy consist exclusively of on-line chan- nels, or would a dual-channel strategy that combines on-line and off- line channels be more desirable? " If a seller decides to adopt a dual-channel strategy, how much coordination is needed between the on-line and off-line channels? The answers to these questions become critical when one considers that whereas many sellers have adopted a multichannel strategy that includes both Web-based channels and pre-existing off-line channels, most of them have not yet found this strategy to be successful [19]. The channel literature from industry and marketing sheds light on how the marketing environment af- fects a firm 9s channel choice and performance [24, 32]. However, the research on direct channels has focused on the optimization of operations within the firm [26, 28], whereas the research on retail channels has generally focused on the competition between retailers and between retailers and direct sellers [3, 35, pp. 277 3280]. Little effort has been made to study the subject of retailers switching to a direct-seller mode (Web-based, in the present case) or to a mul- tichannel strategy (i.e., both direct and retail) and the conditions that would justify such changes from the traditional channel strategy. Lal and Sarvary formulated guidelines for firms expanding their distribution network to the Internet [21]. Their analysis, however, compares two simple channel strate- gies 4the purely off-line channel strategy and the hybrid on-line/off-line chan- nel strategy. The discussion in this paper goes further, for it includes the exclusively on-line channel strategy and two types of multichannel strategies. The Model The analysis begins by looking at a market with two sellers selling their prod- ucts to heterogeneous buyers via traditional off-line channels (i.e., non-Web channels). The elements of the model are explained below. Buyers Stone classified shoppers as falling into one of four categories: economic, apa- thetic, personalized, and ethical [33]. Several researchers have tested and con- INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 105 firmed Stone 9s categories [13, 23]. This paper considers personalized and ethical shoppers, and also recreational shoppers, as defined by Belenger and Kargaontar [4], to be consumers with low intentions of using on-line channels (hereinafter called cLows d). Economic and apathetic shoppers were considered to be buy- ers with high intentions of shopping via Web-based on-line channels (herein- after called cHighs d). A detailed classification of buyers follows. " Highs : According to Bellman and colleagues, individuals who shop on-line are more like traditional catalog shoppers [5]. The prototypi- cal Internet shopper is time-starved and looks for convenience in buying. Stone defined the economic shopper as one who views shop- ping as an economic activity and shops for the best bundle of quality and price, and the apathetic shopper as one who does not enjoy shop- ping and tries to minimize buying effort [33]. Both types of buyers also qualify as buyers with comparatively high intentions of buying via on-line channels. In addition, Highs do not particularly care whether they can try the product before they buy it as long as they can maximize their utility from the trade. Finally, they believe that the benefits offered by Web-based channels generally outweigh the costs associated with using these channels. " Lows : These buyers have a low intention of purchasing from Web- based channels for various reasons. For example, they may be easily distracted by complicated Web sites due to inconvenience and mental fatigue, or they may not have great trust in Web-based channels [17, 25]. Stone defined the personalized shopper as one who enjoys develop- ing a close relationship with store personnel and tends to shop close to home, and the ethical shopper as one who feels a moral obligation to shop at local stores [33]. Both types would qualify as buyers with low intentions to buy via on-line channels, as would recreational shoppers [4], who view shopping as an enjoyable activity and therefore want to spend time in off-line retail outlets. All of these Low buyers believe that the benefits of Web-based channels do not amount to much when compared to the cost of buying via these channels. In addition to segmenting buyers in terms of their intentions to buy on- line, the model incorporates several buyer-related parameters. Here is a brief description of the parameters and the assumptions associated with them. Assumption B1. Product Valuation : All potential buyers have the same intrinsic valuation, V , for the product because the model only applies to undifferentiated products. Assumption B2. Discomfort Cost : Both Lows and Highs have a discomfort factor that reflects their belief about the benefits and costs of using on-line channels. For any buyer, the discomfort cost is the product of the discomfort factor and the product valuation, V . The decision to relate the discomfort cost to the product valuation is justified by the assumption that buyers feel more discomfort if they have to buy an expensive product from an on-line channel, and less discomfort if they buy a lower-value product from an on-line 106 KING, SEN, AND XIA channel. The discomfort factor is manifested, for example, in the lack of immediate gratification, the perceived risk associated with buying on-line [25], and the absence of a practical demonstration of the product. One assumes that sellers can estimate the discomfort cost associated with on-line channels through consumer surveys and other market research exercises. On- line sellers can use personalization technology to get an idea of consumers 9 concerns about on-line channels and the cvalue d they attach to these con- cerns. The discomfort factor of Highs is denoted by ´ H , and the discomfort cost of Lows is denoted by ´ L , where ´ H < ´ L . It is assumed that each consumer will purchase either one or zero units of the product during the period under study. Assumption B3. Demand : Without any loss of generality, the total number of buyers is assumed to equal 2n, and the segment sizes of Lows to be 2n(x), with the remaining 2n(1 3 x) representing the segment size of Highs. The model makes the following additional assumptions about the buyers: Assumption B4. Probability of Returns : Parameter ² denotes the probabil- ity that the consumer will need to access the on-line channel again after buying the product in order to exchange it or get it repaired. Assumption B5. Cost of Returns : Parameter ¸ denotes the cost, as a percentage of the selling price, incurred by any buyer who needs to access the on-line channel again after buying the product. For example, a buyer who needs to get the product exchanged or repaired may have to pay the cost of shipping the product to the seller if the purchase was made via a Web-based channel. On the other hand, if the product was bought via an off-line channel, the buyer can simply take it to the retail outlet at a minimal cost (i.e., as compared to the cost of shipping the product to the on-line seller). Although the buyer has to go to an off-line store and stand in a queue to make the return or exchange, this cost is canceled by the cost the buyer incurs from driving to the post office and waiting in line there when making a return to an on-line store. Besides postage, on-line returns also incur extra costs like packaging the merchandise for shipping, and the psychological cost associated with the delay in getting the refund or the replacement (as compared to almost no delay with off-line returns), which are reflected in the choice of the variable ¸ . Assumption B6. Off-line Transportation Cost : The transportation cost incurred by a buyer who buys off-line is assumed to be zero in the model. One reason for this is that the cost is negligible when compared to the personal valuation of the product. A second reason is the assumption that the inherent benefits associated with traditional off-line channels (e.g., recreational value of off-line shopping, low risks associated with these channels, instant gratifi- cation allowed by immediate trade completion) compensate for the nominal transportation cost incurred during off-line shopping. Finally, off-line transportation costs are generally used to differentiate off-line sellers. The focus of this paper, however, is to differentiate channel strategies and not sellers using the same channel strategy. The model uses the discomfort factor associated with on-line channels to differentiate channel strategies. INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 107 If p A and p B denote the unit price of the product when sold via off-line and on- line channels respectively, the utility functions for Highs (i.e., U H ) and Lows (i.e., U L ) are as follows (see Appendix A for derivation): The parameters defined so far are summarized in Table 1. Sellers Sellers are profit-maximizing firms, selling a commodity product suitable for sale via both on-line and off-line channels. Some important assumptions about sellers are listed below: Assumption S1. Duration of the Game : The sellers know they are going to be in the market for a very long time and therefore have to choose a price that maximizes their long-run profits. Assumption S2. Discount Rate : The discount rate is assumed to be sufficiently large (i.e., closer to 1) because information technology has enabled firms to implement rapid price changes in any period. This is especially true for Web-based markets, where price changes can be implemented very fast. The ability to introduce rapid price changes shortens each time-period. For instance, the time-periods may be reduced from days to hours. Shorter time- periods imply that if a seller gets a payoff of $P in time period t, the net present value of the payoff in time-period t 3 1 remains essentially the same, $P, because the time elapsed between t 3 1 to t is very short. Assumption S3. Retaliation Strategy : Both sellers adopt a trigger strategy [35, pp. 209 3276]. This implies that both sellers continue to charge certain prices greater than the Bertrand price (i.e., price equals the production cost) in every period. Deviation (i.e., a price cut) from the focal price by either of the sellers, in a period, results in the harshest retaliation from the other seller in subsequent periods. This retaliation results in the Bertrand price (and therefore zero profits) for both sellers and therefore zero profits in subsequent periods. Assumption S4. Production Cost : The production costs (or the costs at which the seller might have bought the product from another seller/producer) (1a) (1b) (1 ) (1 ).................... (1 ) ................................. .......................................................... 2 2 + 2 2 = 2 H B H B H A V p buy on-line/exchange on-line V p buy on-line/exchange on-line U V p ´ ²¸ ´ ............... 0...................................................................................... buy off-line not buy (1 ) (1 ).................... (1 ) ................................. .......................................................... 2 2 + 2 2 = 2 L B L B L A V p buy on-line/exchange on-line V p buy on-line/exchange on-line U V p ´ ²¸ ´ ............... 0...................................................................................... buy off-line not buy 108 KING, SEN, AND XIA for both sellers are assumed to be zero. As a result, all the prices in the following analysis have to be interpreted as deviations from zero. The model excludes any variable costs associated with processing on-line and off-line orders because they are assumed to be too low to have a major impact. The inventory management cost is assumed to be about the same for on-line and off-line channels, since both can use IT to collaborate with their suppliers and minimize the cost. Assumption S5. Location of Off-line Sellers : The off-line transportation cost for buyers is assumed to be the same, zero, for all off-line sellers because off-line retail stores are often located near each other in actual markets. Various reasons have been cited for the close proximity of sellers: (a) demand is concentrated around a few poles (e.g., location of shopping malls), (b) common installations and trade centers provide better economic incentives to locate near one other, (c) close location of sellers results in lower search costs and increases aggregate demand for sellers, (d) price competition is prevented due to tacit collusion among the sellers [10]. The sellers have four fundamental channel choices: (a) to continue selling via their traditional off-line channels, (b) to switch completely to Web-based channels, (c) to sell via both the on-line and off-line channels without much coordination between the two, and (d) to sell via well-coordinated on-line and off-line channels. Coordination of channels would indicate whether buyers who buy the product in one channel are allowed to return or exchange it in the other channel. In cases of minimum coordination, this is not allowed (e.g., Boscovs): Buyers are allowed to return or exchange the product only via the channel from which it was purchased. In cases of maximum coordination, buyers Possible No. Parameter value Description 1 V positive Buyer 9s valuation of product 2 ´ H 0 31 Discomfort factor of buyers with high intentions to buy from online channel: a percentage of valuation V of product purchased. 3 ´ L 0 31 Discomfort factor of buyers with low intentions of buying from on-line channel: a percentage of valuation V of product purchased. 4 x 0 31 Number of buyers with low intentions of buying on- line as fraction of total demand 5 p A positive Price of product in off-line channel 6 p B positive Price of product in on-line channel 7 ² 0 31 Probability that consumer will need to access channel again after buying product 8 ¸ 0 31 Cost, as percentage of selling price, incurred by any buyer who needs to access on-line channel again after buying product Table 1. Summary of Buyer-Related Parameters Used in the Model. ´ L > ´ H INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 109 can purchase the product from on-line channels but return or exchange it at off- line channels of the retailer that made the sale (e.g., Wal-Mart, Best Buy, Gateway). Given the four channel strategies available to each of the two sellers, sixteen scenarios will have to be analyzed in order to identify the optimal channel strategy for both sellers. These scenarios are defined in Table 2. We will now define an infinite-period repeated game in which each period consists of two stages, with sellers and consumers as players [35, pp. 209 3276, 277 3280] ( see Figure 1 ). In this game, the sellers get to move in the first stage, during which they decide on their channel strategy simultaneously. The consumers move only in the second stage, in which they have to decide on their preferred channel and the price they are willing to pay for the product. Timing of Decisions The model assumes that in each period, sellers and buyers make decisions in two sequential stages: Uncoordi- Coordi- Seller 2 nated nated Sell Sell dual dual Seller 1 off-line on-line channels channels Sell off-line Scenario 1 Scenario 2 Scenario 3 Scenario 4 Sell on-line Scenario 5 Scenario 6 Scenario 7 Scenario 8 Uncoordinated dual channels Scenario 9 Scenario 10 Scenario 11 Scenario 12 Coordinated dual channels Scenario 13 Scenario 14 Scenario 15 Scenario 16 Table 2. All the Possible Scenarios for Channel Strategies Adopted by Sellers 1 and 2. Figure 1. Sequence of Moves by the Players Involved in the Game Note: The various scenarios are defined in Table 2. Scenario 16 Scenario 15 Scenario 14 Scenario 13 Scenario 12 Scenario 11 Scenario 10 Scenario 9 Scenario 8 Scenario 7 Scenario 6 Scenario 5 Scenario 4 Scenario 3 Scenario 2 Scenario 1 Scenario 16 Scenario 15 Scenario 14 Scenario 13 Scenario 12 Scenario 11 Scenario 10 Scenario 9 Scenario 8 Scenario 7 Scenario 6 Scenario 5 Scenario 4 Scenario 3 Scenario 2 Scenario 1 Stage 1: Sellers decide their channel strategy Stage 2: Buyers make their Purchase decision Buy Offline Do not buy Buy Online Buy Off-line Buy On-line Do not buy 110 KING, SEN, AND XIA Stage I The sellers choose the channels they want to sell on and the price at which they want to sell. The price will be denoted by p i , where i = A for off-line prices and i = B for on-line prices. In deciding on a channel strategy, the sellers are aware of the discomfort cost incurred by buyers with low and high intentions to buy via on-line channels. In deciding on a pricing strategy, the sellers try to maximize the discounted value of their long-run profits. Stage II Each consumer decides whether to purchase one unit of the product. In mak- ing this decision, the consumer treats the distribution channel and the price of the product as given. After the consumers make their purchase decisions, the sellers collect their revenues from the consumers and realize profits. Solving for the Equilibrium The equilibrium channel strategy for each seller will be the strategy that maximizes the present discounted value of its profit in an infinite period. One of the price equilibriums of a repeated game is the Bertrand equilib- rium repeated infinitely (i.e., both firms sell at their production cost in every period) [35, p. 244]. However, when both sellers adopt a trigger strategy (As- sumption S3) and the discount factor is sufficiently large (Assumption S2), then any price greater than the Bertrand price but less than or equal to the monopoly price could be a possible equilibrium [35, pp. 209 3276]. Given that any of an infinite number of prices could be a possible equilibrium, the assumption used here is one commonly used in the literature on repeated games, namely, cin a symmetric game the focal equilibrium must be Pareto optimal from the viewpoint of the two firms (i.e., must yield a payoff on the frontier of the attainable set of per-period profits), d to identify the equilib- rium price [35, pp. 209 3276] . In the model defined here, these assumptions clearly result in both the sellers selling at the monopoly prices for each of the possible channel strategies. This equilibrium price can be enforced by the strategy of charging monopoly prices as long as both sellers have charged monopoly prices earlier and are aware that in case of a deviation from these prices, the rival firm would charge the Bertrand price, resulting in zero prof- its in subsequent periods of the game. This result is a formalization of tacit collusion that is enforced through a purely non-cooperative mechanism, which ensures that price competition does not take place [10]. Tacit collusion in the model can be sustained by the threat of future losses in a price war [10]. Chamberlin suggests that in an oligopoly producing a homogenous product, firms recognize their interdependence and, therefore, might be able to sustain the monopoly price without explicit collusion [9]. The threat of a vigorous price war would be sufficient to deter the temptation to cut prices. Hence, the oligopolies might be able to collude in a purely non- cooperative manner. Chamberlin even suggested that under certain conditions, INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 111 the monopoly price was the most likely outcome [10]. The conditions are (a) price changes can be observed very quickly by the competition, (b) under symmetric conditions the price at which tacit collusion occurs is the monopoly price, whereas under asymmetric conditions the collusion is more complex but still occurs at prices higher than the marginal cost (i.e., the Bertrand equilibrium price), (c) multi-market contact among sellers is generally thought to blunt the incentive for rivalry [7], and (d) tacit collusion is easier to sustain in markets with low concentration (e.g., a duopoly, as in the case of the present model). In light of these reasons, it is argued that the long- term profit-maximizing prices charged by the sellers in each of the scenarios will be the maximum price the buyers are willing to pay, that is, the monopoly price. Therefore, all one need do to solve for the equilibrium price in any period is to identify the monopoly prices for various possible channel strategy choices ( see Table 2 ) and compute the single-period profits. A comparison of these profit functions for different channel strategies would give the optimal channel strategy for the period. This will also be the optimal channel strategy for the whole game because the prices for this strategy would maximize the present discounted value of the profits (in the infinite period) for both sellers. The following section solves for the monopoly prices for each possible channel strategy in any period. This is computed by solving the game backward for a single period 4that is, Stage II is solved, and then Stage I. Stage II: Consumer Purchasing Decision: Estimation of Demand At this stage, each consumer has the following information: (a) whether the product is available for sale on-line, off-line, or on both channels, (b) the price of the product, (c) a valuation of the product, and (d) the discomfort cost of buying on-line. In this stage, one estimates the demand for the vari- ous channel choices open to the sellers. Given the four channel strategies available to both sellers, the buyers could encounter any of the 16 scenarios shown in Table 2. The discussion that follows will identify the total on-line and off-line demand for Highs and Lows for all of the possible scenarios. At this stage there is no need to consider the distribution of demand between the two sellers. It will be computed once the profit-maximizing prices that result from tacit collusion are identified for the 16 scenarios. The demand for all the scenarios is computed simply by ensuring that the buyer 9s utility is (a) some positive value when the purchase is made, and (b) more for the channel from which the purchase was made than from the other channel. For example, a buyer would buy from the on-line channel if his or her utility is positive for on-line purchases and this utility is greater than the one for off-line purchase. Thus, by solving for U H ,U L e 0 ( Eqs. 1a and 1b ), and com- paring U H and U L and for different channel-strategy options, one can deter- mine the channels preferred by Low and High buyers, and the range of on-line and off-line prices they are willing to pay when buying from their preferred channel. 112 KING, SEN, AND XIA Scenario 1: Both sellers sell only via existing off-line channels. The total number of buyers according to Eqs. (1a) and (1b) will be AL AH A AL AH A q 2xn.;q x n p V q q p V 2(1 ) ................... 0....................................... = = 2 d = = > (2) Scenario 2 (Scenario 5): Seller 1 (2) sells via an off-line channel, and Seller 2 (1) sells via an on-line channel. In all these scenarios, the buyers have to decide between buying on-line and buying off-line. The total demand for the two channels is given in Table 3. Scenario 6: Both sellers sell only via on-line channels. The total demand for this on-line channel is given in Table 4. Scenario 3 (Scenario 9), Scenario 7 (Scenario 10), and Scenario 11: If Seller 1 (2) sells via an off-line channel and Seller 2 (1) sells via uncoordi- nated dual channels, or both sellers sell via uncoordinated dual channels, or one seller sells via an on-line channel and the other via uncoordinated dual channels, the buyers have to decide between buying on-line or off-line. The total on-line and off-line demand is given in Table 5. Scenario 4 (Scenario 13), and Scenario 16: Seller 1 (2) sells via an off-line channel and Seller 2 (1) sells via coordinated dual channels, or both sellers sell through coordinated dual channels. In all these scenarios, the buyers have to decide between off-line and coordinated dual channels (i.e. buy on-line and exchange off-line). The total on-line and off-line demand for these scenarios is given in Table 6. Scenario 8 (Scenario 14), Scenario 12 (Scenario 15): The buyers have a choice between an on-line channel, an off-line channel, and coordinated dual channels. No on-line buyer would ever buy from a seller who sells only via an on-line channel or via uncoordinated dual channels when there is an option of buying from a coordinated dual channel, if the on-line price charged by all the sellers is same. This is because for the same price, the buyer 9s utility from using a coordinated dual channel is always greater than the utility from using a purely on-line channel. The on-line buyers would be indifferent between the purely on-line channel and the coordinated dual channel when 2´ 2 +²¸ = 2´ 2 i B i B V p V p 1 2 (1 ) (1 ) (1 ) ; where, i = 1 or 2, p 1 B is the on-line price charged by seller who sells via the on-line channel or the uncoordinated dual channel, and p 2 B is the on-line price charged by the seller who sells via the coordinated dual channel. This gives the following relation between the on- line prices p 1 B and p 2 B : = +²¸ B B p p 2 1 . 1 (2a) With this pricing strategy, all the off-line demand goes to the seller selling via a coordinated dual channel (in Scenario 8 (14)), and the on-line demand is divided equally between the two. In scenario 12 (15), this on-line pricing strategy results in both sellers getting equal on-line and off-line demand. The INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 113 Pricing constraints q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Table 4. Demand Functions for Scenario 6. Pricing constraints A A B p V V p L and (1 ) d 2 ´ d +²¸ Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x )n q AL 0 2 xn 2 xn Table 3. Demand Functions for Scenario 2 (Scenario 5). A A L A H B p V p V p V p and (1 ) (1 ) d 2 ´ 2 ´ < d +²¸ +²¸ A A H B p V p V p and (1 ) d 2 ´ > +²¸ L B V p (1 ) (1 ) 2´ d +²¸ L H B V V p (1 ) (1 ) (1 ) (1 ) 2´ 2´ < d +²¸ +²¸ H B V p (1 ) (1 ) 2´ > +²¸ Table 5. Demand Functions If One Seller Sells via Off-line or On-line Channel and the Other via Uncoordinated Channels or Both Sell via Uncoordinated Channels. A A L B p V p V p and (1 ) d 2´ d + ²¸ A A L A H B p V p V p V p and (1 ) (1 ) d 2´ 2´ < d + ²¸ +²¸ A A H A B p V p V p p and (1 ) d 2´ e > + ²¸ Pricing constraints Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn Table 6. Demand Functions If One Seller Sells via Off-line Channel and the Other via Coordinated Dual Channels or Both Sell via Coordinated Dual Channels. Pricing constraints A B A L p V p p V and d d 2´ A A H B A L p V p V p p V and d 2´ e > 2´ A B A H p V p p V and d > 2´ Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn 114 KING, SEN, AND XIA on-line and off-line price constraints, corresponding to demand for on-line, and off-line channels are presented in Table 7. Stage 1: Sellers Select Their Channel and Price to Maximize Their Profits In this stage, the sellers select their respective profit-maximizing price subject to the consumers 9 demand functions, as derived in the preceding section. The sellers are aware of the discomfort cost of the two types of buyers (i.e., when the buyers buy on-line), the percentage of Lows and Highs in the buyer population, and the buyers 9 common valuation of the product. The remaining part of this section derives the profit functions for the two sellers for various scenarios. Scenario 1 : The profit-maximizing price, given that the product is sold off- line by both sellers, is p A = V for both sellers ( see Eq. 3 ). The off-line demand will be divided equally between the two sellers. The profits for each seller for this scenario are given by À i = nV for i = 1, 2 (3) Scenario 2 (Scenario 5): The profit-maximizing prices for these scenarios can be derived from Table 3. In this case, Seller 1 (2) will get all the off-line demand, and Seller 2 (1) will get all the on-line demand. Without loss of generality, the profit functions will be computed only for Scenario 2. The resulting profits for the two sellers are given in Table 8. Scenario 3 (Scenario 9): Total demand and the profit-maximizing prices for these scenarios are computed from the price constraints given in Table 5. In Scenario 3 (Scenario 9) Seller 1(2) gets half the off-line demand, and Seller 2 (1) gets all the on-line demand and half the off-line demand. With this demand distribution, their respective profit functions are given in Table 9. Scenario 4 (Scenario 13): The profit-maximizing prices for these scenarios are derived from the price constraints given in Table 6. Without loss of Table 7. Demand Functions If One Seller Sells via On-line Channel and the Other via Coordinated Dual Channels. Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn A B A L B B p V p p V p p 2 2 1 and 1 d d 2´ = + ²¸ A A L B A H B B p V p V p p V p p 2 2 1 and 1 d 2´ d < 2´ = +²¸ A B A H B B p V V p p 2 2 1 and 1 d > 2´ = +²¸ Pricing constraints INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 115 Table 8. Profit Functions for Scenario 2 (Seller 1 Sells via Off-line Channel, and Seller 2 Sells via On-line Channel). Note: The pricing options given in column 4a and 4c result in zero profits for Seller 1 and Seller 2 respectively. The profit functions for these pricing strategies are not sustainable because the seller with zero profit can always change the channels strategy that gives it a positive profit. Therefore, we will ignore these pricing options in subsequent analysis, leaving us with the focus on the profit functions given in column 4b. A L B p = V V p and (1 ) (1 ) 2 ´ = + ²¸ A H B p = V V p and (1 ) (1 ) 2 ´ = + ²¸ A H B p = V V p and (1 ) (1 ) 2 ´ > + ²¸ L nV 1 2 1 2 ´ + ²¸ Profit-maximizing prices Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2xn 2 xn Profits for seller 1 0 2 xnV 2 nV Profits for seller 2 0 4a 4b 4c Table 9. Profit Functions When Seller 1 Sells via Off-line Channel and Seller 2 Sells via Uncoordinated Dual Channels. Note: Since the pricing strategy in column 5a results in zero profits for Seller 1, and Seller 1 can always change its channel strategy to get positive profits, the profits in column 5a will be ignored in subsequent analysis. This leaves the profit functions given in columns 5b and 5c. For Seller 1, 5c is always greater than 5b. For Seller 2, however, 5c > 5b only when 2²¸ ´ > H 1 2 . Therefore, 5c will be used when this condition is assumed to be true. Otherwise, there is no equilibrium for these scenarios. A H B p V V p and (1 ) (1 ) = 2 ´ = + ²¸ A H B p V V p and (1 ) (1 ) = 2´ = + ²¸ A H B p = V V p and (1 ) (1 ) 2´ > + ²¸ Profit-maximizing prices Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn Profits for Seller 1 0 xnV nV Profits for Seller 2 nV 5a 5a 5a 5a 5a 5b 5b 5b 5b 5b 5c 5c 5c 5c 5c H x nV (1 )(1 ) 2 (1 ) 2 2´ + ²¸ L nV 1 2 1 2´ +²¸ H x nV x 2(1 )(1 ) [ ] 1 2 2´ + +²¸ 116 KING, SEN, AND XIA generalization, only Scenario 4 will be considered. For this choice of channels, Seller 1 will get half the off-line demand, and Seller 2 will get the other half of the off-line demand and all the on-line demand. The profit functions for the two sellers are given in Table 10. Scenario 6 : The profit-maximizing prices for this scenario, when the product is sold on-line by both sellers, can be derived from Table 4. In addition, the total on-line demand will be shared equally between the two sellers. The profits for the two sellers are given in Table 11. Table 10. Profit Functions When Seller 1 Sells via Off-line Channel and Seller 2 Sells via Coordinated Dual Channels. Note: Since the pricing strategy in column 6a results in a zero profit for Seller 1, the profit functions given in this column will be ignored in subsequent analysis. For Seller 1, 6c is always greater than 6b. For Seller 2, however, 6c > 6b is true when x + 2(1 3 x )(1 3 ´ H ) < 1, i.e., when ´ H > 1/2. In subsequent analysis, the profits given in column 6c will be used when ´ H > 1/2. Otherwise, there is no equilibrium Profit-maximizing prices A B L p = V p V and (1 ) = 2 ´ A B H p V V and (1 ) = > 2 ´ Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn Profits for Seller 1 0 xnV nV L nV 2 (1 ) 2´ H nV x x [ 2(1 )(1 )] + 2 2´ nV Profits for Seller 2 6a 6b 6c Table 11. Profit Options for Both Sellers When They Sell via On-line Channels. Profit-maximizing price L B V p (1 ) (1 ) 2´ = +²¸ H B V p (1 ) (1 ) 2 ´ = + ²¸ H B V p (1 ) (1 ) 2 ´ > + ²¸ Total on-line demand q AH 2(1 3 x ) n 2(1 3 x ) n 0 q AL 2 xn 0 0 Profits for each seller L nV (1 ) 1 2 ´ + ²¸ H x nV (1 )(1 ) (1 ) 2 2 ´ + ²¸ 0 7a 7b Note : 7b > 7a when x ´ 2´ d 2´ L H H 1 . Therefore, when x ´ 2´ d 2´ L H H 1 the profits in column 7b are used for further analysis. Otherwise, the profits in column 7a are used. A B H p = V p V and (1 ) = 2 ´ INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 117 Scenario 7 (Scenario 10): The profit-maximizing prices for these scenarios are derived from the price constraints given in Table 5. Without loss of generalization, only Scenario 7 will be considered. In this case, the on-line demand will be divided equally between Seller 1 and Seller 2, whereas all the off-line demand would go to Seller 2. The profit functions for the two sellers are given in Table 12. Scenario 8 (Scenario 14): The profit-maximizing prices for these scenarios are derived from the price constraints given in Table 7. Because of the symmetric nature of these two scenarios, the profit functions will only be computed for Scenario 8. Seller 2 is better off charging = 2´ B i p V 2 (1 ) [where i = H or L]. Substituting p 2 B in (2a) gives 2´ = + ²¸ i B V p 1 (1 ) 1 , which incidentally is the maximum on-line price the buyers are willing to pay in a purely on-line channel. Neither seller can increase the price any more Table 12. Profit Functions When Seller 1 Sells via On-line Channel and Seller 2 Sells via Uncoordinated Dual Channels. Note : Column 8c will be ignored in subsequent analysis because the zero profit for Seller 1 ensures that this scenario is not sustainable. For Seller 1, 8b > 8a when x L H H 1 ´ 2´ d 2´ . For Seller 2, 8b > 8a is always true. Therefore, when x L H H 1 ´ 2´ d 2´ the profits in column 8b will be used. Profit-maximizing price A L B p = V V p and (1 ) (1 ) 2´ = +²¸ A H B p = V V p and (1 ) (1 ) 2´ = + ²¸ A H B p = V V p and (1 ) (1 ) 2´ > + ²¸ Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn Profits for Seller 1 L nV 1 1 2´ +²¸ H x nV (1 )(1 ) 1 2 2´ +²¸ 0 Profits for Seller 2 L nV 1 1 2´ +²¸ H x nV x (1 )(1 ) [2 ] 1 2 2´ + +²¸ 2 nV 8a 8b 8c 118 KING, SEN, AND XIA because it is the maximum price the on-line sellers are willing to pay to Seller 1(2) and Seller 2(1). Neither seller can decrease the price because this would lead to Bertrand prices and zero profits in subsequent periods. The profits for the two sellers in this case are given in Table 13. Scenario 11: The profit-maximizing price and the total on-line and off-line demand, when both sellers sell the product via uncoordinated dual channels, are given in Table 5. The on-line and off-line demand will be equally divided between the two sellers. The corresponding profit functions are given in Table 14. Scenario 12 (Scenario 15): The analysis is similar to the one for Scenario 8 (Scenario 14) because the buyers again have a choice between buying on-line, buying via coordinated dual channels, and buying off-line. In all these scenarios, on-line buyers would be indifferent between on-line and the coordinated channels only when the relationship between p 1 B (i.e., the on-line price charged by the seller selling via the uncoordinated dual channel) and p 2 B (i.e., the on-line price charged by the seller selling via the coordinated dual channel) is given by Eq. (2a). Because of the symmetric nature of these Profit-maximizing price A B L L B p = V p V V p 2 1 and (1 ) (1 ) 1 = 2´ 2´ = +²¸ A B H H B p = V p V V p 2 1 and (1 ) (1 ) 1 = 2´ 2´ = +²¸ Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn Profits for Seller 1 L nV 1 1 2´ +²¸ H x nV (1 )(1 ) 1 2 2´ +²¸ 0 Profits for Seller 2 L nV (1 ) 2´ H x nV x (1 )(1 ) [2 ] 1 2 2´ + +²¸ 2 nV 9a 9b 9c A B H H B p = V p V V p 2 1 and (1 ) (1 ) 1 > 2´ 2´ > + ²¸ Note : Column 9c will be ignored in subsequent analysis because the zero profit for Seller 1 ensures that this scenario is not sustainable. For Seller 1, 9b > 9a when x L H H 1 ´ 2´ d 2´ . For Seller 2, 9b > 9a is always true. Therefore, when x L H H 1 ´ 2´ d 2´ the profits in column 9b will be used. Table 13. Profit Functions When Seller 1 Sells via On-line Channel and Seller 2 Sells via Coordinated Dual Channels. INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 119 Profit-maximizing price Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn Profits for each of the sellers L nV 1 1 2´ +²¸ L x nV x (1 )(1 ) [ ] (1 ) 2 2´ + + ²¸ nV 10a 10b 10c Table 14. Profit Options for the Two Sellers When They Both Sell via Uncoordinated Dual Channels. Note : It can easily be shown that 10c > 10b and 10a is always true. Since both sellers intend to maximize their long-run profits and fear severe retaliation in response to any price cut, p A = V and any 2 ´ > + ²¸ (1 ) (1 ) H B V p will be the result of the tacit collusion . Therefore, the profit function given in column 10c will be used in any further analysis. A L B p = V V p and (1 ) (1 ) 2´ = + ²¸ A H B p = V V p and (1 ) (1 ) 2´ > + ²¸ A H B p = V V p and (1 ) (1 ) 2´ = + ²¸ scenarios, it is only necessary to compute the profit functions for Scenario 12. The profits for the two sellers in this case are given in Table 15. Scenario 16: The on-line and off-line demand is equally divided between the two sellers, and the resulting profit functions are given in Table 16. Stage I: Sellers 9 Channel Strategy Decision The discussion in the preceding section identified the profit functions resulting from tacit collusion . These are summarized in Table 17. Analysis of the payoffs for both sellers given in Table 17 leads to several propositions. Proposition 1 : One seller sells via an off-line channel and the other (a) sells via an off-line channel, or (b) adopts an uncoordinated or coordinated dual channel, when the discomfort factor associated with on-line channels is more than 0.5 for Highs. Without loss of generality, let it be assumed that Seller 1 chooses to continue with its off-line channel strategy. In this case, Seller 2 is indifferent between selling through its traditional off-line channel, adopting an uncoordinated channel strategy, and adopting a coordinated channel strategy when ´ H > 1 2 ( see Tables 9 and 10 ). However, even when Seller 2 adopts a dual-channel strategy, its pricing policy encourages the buyers to buy from the off-line channel. An intuitive explanation for this equilibrium is that when the 120 KING, SEN, AND XIA Profit-maximizing price A B L L B p = V p V V p 2 1 and (1 ) (1 ) 1 = 2´ 2´ = +²¸ A B H H B p = V p V V p 2 1 and (1 ) (1 ) 1 = 2´ 2´ = +²¸ A B H H B p = V p V V p 2 1 and (1 ) (1 ) 1 > 2´ 2´ > +²¸ Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn Profits for Seller 1 L nV 1 1 2´ +²¸ H x nV x (1 )(1 ) [ ] (1 ) 2 2´ + +²¸ nV Profits for Seller 2 L nV (1 ) 2´ H nV x x [ (1 )(1 )] + 2 2´ nV 11a 11b 11c Note: It can easily be shown that 11c > 11b and 11a is always true because ´ H < ´ L and ´ , ² , ¸ , and x lie between 0 and 1. Since both sellers intend to maximize their long-run profits and fear severe retaliation in response to any price cut, p A = V and H B V p (1 ) (1 ) 2´ > + ²¸ will be the result of the tacit collusion. Therefore, the profit function given in column 11c will be used in any further analysis. Table 15. Profit Functions When Seller 1 Sells via Uncoordinated Dual Channels and Seller 2 Sells via Coordinated Dual Channel. Table 16. Profit Functions When Both Sellers Sell via Coordinated Dual Channels. Note: It can easily be shown that 12c > 12b and 12a is always true. Since both sellers intend to maximize their long-run profits and fear severe retaliation in response to any price cut, p A = V and any p B > V (1 3 ´ H ) will be the result of the tacit collusion. Therefore, the profit function given in column 12c will be used in any further analysis. Profit-maximizing price Total on-line demand q BH 2(1 3 x ) n 2(1 3 x ) n 0 q BL 2 xn 0 0 Total off-line demand q AH 0 0 2(1 3 x ) n q AL 0 2 xn 2 xn Profits for each seller L nV (1 ) 2´ H nV x x [ (1 )(1 )] + 2 2´ nV 12a 12b 12c A B L p = V p V and (1 ) = 2´ A B H p = V p V and (1 ) = 2´ A H B p = V V and (1 ) 2´ < INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 121 discomfort cost of buyers with high intention to buy on-line is high, then the discomfort cost of buyers with low intention to buy on-line will be higher. This could be an indication that the on-line channels are sufficiently developed or advanced to make most buyers feel comfortable about using them, resulting in a relatively high proportion of buyers with low intention to buy on-line. Under such conditions, the sellers would prefer to continue with traditional off-line channels, since most buyers would still be using them. Even if one seller includes an on-line channel in its channel strategy, it would still adopt an on-line pricing strategy such that buyers would prefer to buy off-line. This may explain the slow growth of on-line channels in many industries (e.g. personal banking) that could be characterized by a relatively high discomfort cost for most buyers. Proposition 2 : When the proportion of Lows is less than H H L ´ 2 2 1 , the equilibrium outcomes are (a) Both sellers sell via coordinated dual channels, Table 17. Profits of Seller 1 and Seller 2 When H 1 1 2 2 2 ²¸ ´ > > and L H H x 1 ´ 2´ d 2´ . Note : When the conditions H 1 1 2 2 2 ²¸ ´ > > and x L H H 1 ´ 2´ d 2´ are not true, no equilibrium exists for several of the scenarios. Therefore, the subsequent analyses assume that these conditions hold. Sell via Sell via Seller 2 uncoordinated coordinated Sell Sell dual dual Seller 1 off-line on-line channels channels Sell off-line nV, nV nV, nV nV, nV Sell on-line 2 xnV Sell via uncoordinated dual channels nV, nV nV, nV nV, nV Sell via coordinated dual channels nV, nV nV, nV nV, nV H xnV x nV 2 , (1 )(1 ) 2 (1 ) 2 2´ +²¸ H H x nV nV x x (1 )(1 ) , (1 ) [2 (1 )(1 )] 2 2´ +²¸ + 2 2´ H H x nV x nV x (1 )(1 ) , 1 (1 )(1 ) [2 ] 1 2 2´ +²¸ 2 2´ + +²¸ H H x nV x x nV (1 )(1 ) [2 ], 1 (1 )(1 ) (1 ) 2 2´ + +²¸ 2 2´ +²¸ H H nV x x x nV [2 (1 )(1 )], (1 )(1 ) (1 ) + 2 2´ 2 2´ +²¸ H H x nV x nV (1 )(1 ) , (1 ) (1 )(1 ) (1 ) 2 2 ´ + ²¸ 2 2 ´ + ²¸ H x nV (1 )(1 ), 2 (1 ) 2 2 ´ + ²¸ 122 KING, SEN, AND XIA (b) Both sellers sell via uncoordinated channels, or (c) one seller adopts a coordinated dual-channel strategy and the other adopts an uncoordinated dual-channel strategy. Without loss of generality, let it be assumed that Seller 1 chooses to continue with its off-line channel strategy. In this case, Seller 2 is indifferent between an off-line channel strategy, an uncoordinated dual-channel strategy, and a coordinated dual-channel strategy ( see Table 17 ). When Seller 2 decides to opt for a coordinated dual-channel strategy, Seller 1 is indifferent between continuing with an off-line channel strategy and adopting either an uncoordinated or a coordinated dual-channel strategy ( see Table 17 ). If Seller 1 decides to adopt a dual-channel strategy, Seller 2, in response, is better off with a coordinated or uncoordinated dual-channel strategy. When buyers with a high intention to buy on-line have a sufficiently low discomfort factor, on-line channels are likely to attract enough buyers to make it profitable for sellers to incorporate an on-line channel in their overall channel strategy. This equilibrium is observed in several industries. For example, some consumer retailers of ready-to-wear clothing sell via highly coordinated on-line and off-line channels (e.g., Gap), whereas others sell via relatively less coordinated on-line and off-line channels (e.g., Boscovs). Another example of this equilibrium would be the consumer electronics industry, where most sellers sell via a highly coordinated dual-channel strategy (e.g., Best Buy, Gateway). Proposition 3 : Both sellers adopt the dual-channel strategy because the market competition forces them to do so, and not necessarily because they are economically better off. As can be seen, when one seller adopts a coordinated dual-channel strategy and the other adopts an uncoordinated dual-channel strategy, the profits for the two are the same as when they were both selling via off-line channels, that is, nV. One might ask why the sellers do not sell only via the off-line channel if this is the case. Competition is a possible answer. As the analysis shows, if one seller decides to continue selling off-line, one of the possible equilibriums is that the other seller is better off by switching to a dual-channel strategy. Proposition 4 : When the traditionally off-line seller (e.g., Seller 1) faces competition from a Web-based new entrant (e.g., Seller 2), and the proportion of Lows is less than then Seller 1 is better off selling via a coordinated dual-channel strategy, as long as Seller 2 continues to sell via an on-line channel. Before deriving this equilibrium, it will be useful to present a brief intuitive discussion of the inequality ´ 2 ´ d 2 ´ L H H x , 1 L H H , 1 ´ 2 ´ 2 ´ INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 123 where x is the proportion of buyers with low intentions to buy on-line. The expression on the right-hand side of the inequality, L H H , 1 ´ 2 ´ 2 ´ is the ratio of the additional discomfort cost of buyers with low intentions to buy on-line and the comfort level of buyers with high in- tentions to buy on- line. Simply expressed, it can be written as Additional Discomfort of Buyer with Low Intentions to Buy On-line Comfort Level of Buyer with High Intentions to Buy On-line . If we call this ratio the discomfort ratio, Proposition 4 applies when the proportion of buyers with low intentions to buy on-line is less than the discomfort ratio. The discomfort ratio itself varies in direct proportion to the discomfort of buyers with low intentions to buy on-line and inversely with the comfort level of buyers with high intentions to buy on-line. For instance, if for any product the buyers with high intentions to buy on-line are very comfortable with the on-line channel, whereas the buyers with low intentions to buy on-line are not very comfortable with the on-line channel, then the proportion of low-intention buyers should be very small for the equilibrium described by Proposition 4 to come into play. The following paragraph presents the derivation of the equilibrium described in Proposition 4. Without loss of generality, let it be assumed that Seller 1 has to decide whether to switch on-line or not in response to Seller 2 9s on-line presence. To assume a more realistic scenario, Seller 1 faces competition from a purely Web- based Seller 2 and has to decide whether it should continue with its traditional off-line channel strategy or switch to another channel strategy. Now Seller 2 would enter the market and sell via an on-line channel only if the proportion of buyers with low intention to buy on-line satisfies the condition \x4 \x4 \x3 \x2 2 2 2 d H H x ´ ²¸ ´ 1 2 1 2 1 ( see Appendix B for proof ). In this case, Seller 1 is better off adopting a coordinated-channel strategy ( see Appendix C for proof ) . This outcome, although visible in many industries faced with new Web-based retailers, is not sustainable in the end. For example, in consumer electronics, the traditionally off-line sellers have switched to dual-channel strategies (e.g., Wal-Mart, Best Buy, Gateway) in the face of purely Web-based retailers (e.g., Amazon.com, CDW). In the end, however, as the Web-based new entrants incorporate off- line channels, the equilibrium outcome will be that the traditionally off-line seller sells via a coordinated dual-channel strategy, whereas the new entrant sells via a coordinated or uncoordinated dual-channel strategy. Discussion Analysis of the model makes it possible to identify the conditions under which (a) all sellers continue to sell via off-line channels, (b) traditionally off-line 124 KING, SEN, AND XIA sellers continue to sell via off-line channels even when faced with the threat of on-line competition, (c) all sellers sell via coordinated and/or uncoordinated dual-channels, and (d) some sellers sell via on-line channels and the others sell via coordinated dual channels. All these equilibriums are observed in one or the other industry. The conditions supporting one or the other equilibriums depend on the size of the market segments that favor on-line and off-line chan- nels, and the discomfort factors associated with on-line channels. As was noted earlier in this paper, sellers adopt a dual-channel strategy because of competition and not because they will be economically better off by doing so. This is supported by a Gartner report stating that roughly 75 percent of 375 retailers surveyed said that they already had a multichannel retailing strategy or were planning to implement one [29]. The main reason given for this, according to the report, was that enterprises had to incorporate such a strategy cto play in a multichannel world, d which is to say, a competitive environment. However, whether the dual-channel strategy is successful is still under investigation [19]. A research study conducted for Shop.org, the J.C. Williams Group, and BizRate.com, for example, based on more than 48,000 interviews with shoppers in all channels and in-depth interviews with more than 40 retail executives, reported that many retail executives readily acknowledge that they are far from realizing rewards from a multichannel strategy. There are three possible reasons for a failure to achieve the expected benefits from a multichannel strategy: (a) faulty assumptions about the behavior of on-line consumers, (b) technological concerns, and (c) operational concerns. Faulty Assumptions About Consumer Behavior A Jupiter report about on-line shoppers provides one possible explanation for the less-than-enthusiastic responses of the retail executives [29]. According to this study, whereas conventional wisdom holds that active multichannel shoppers are a lucrative group for on-line retailers to target, they tend to drain on-line customer service resources. The study says that although these consumers buy more frequently than others, they base their buying decisions on price rather than loyalty to any particular brand. For instance, of the 66 million people who research products and shop via the Internet, just 16 million favor the same brands off-line and on-line. These price-sensitive buyers stimulate on-line sales, but they do so at the expense of the e-tailer 9s profitability, because their lack of brand loyalty reduces the expected returns on the investment in channel coordination, CRM initiatives, and on-line customer-retention programs. Until now, the e-commerce market has believed that just the opposite is true, and thus that on-line buyers have the same level of brand loyalty as traditionally off-line consumers. Retailers based their multichannel strategies around this assumption. However, the Jupiter study, although acknowledging the value of multichannel shoppers, noted that they are generally not very loyal and that their loyalties are becoming even more divided. Only 19 percent of the $582 billion that on-line consumers are expected to spend by 2006 will come from shoppers loyal to certain brands. On the other hand, the proportion of those with no brand loyalty is expected to grow from 54 percent at present to about 64 percent in 2006 [29]. INTERNATIONAL JOURNAL OF ELECTRONIC COMMERCE 125 Technological Concerns Customers value, and increasingly expect, a seamless buying experience across on-line and off-line channels [14]. Data enter the enterprise from both on-line and off-line channels, and the most important issue is to assemble them in a timely, coherent, and accurate manner. No retailer wants to over- look a piece of the customer profile, and no customer wants to be asked for the same information more than once. Achieving true coordination between on-line and off-line channels, however, requires seamless integration and coordination at the technology level (e.g., back-end database systems). On the technical side, legacy systems pose a difficult hurdle. Old systems are hard to replace or upgrade, especially when the economic situation is diffi- cult. Even when a legacy system is not performing up to the mark, the IT department will rarely choose to displace it and build a new one in a shaky economic climate. Analysts unanimously recommend a careful, thoughtful, strategically planned, incremental approach to integrating on-line and off- line channels [14]. The goal should be to identify the type of customers whose multichannel experience should be improved and then to determine exactly how to do it. For example, since dynamic inventory management is diffi- cult, crossover returns to off-line channels are a relatively simple matter, whereas it is harder to implement ordering on-line, and taking delivery at the off-line channel. A retail enterprise might first implement in-person re- turns of products ordered on-line, and then eventually move toward more complex cross-channel integration [14]. Operational Concerns The Shop.org study raises the question of whether traditional off-line sellers (e.g., retailers) are free to choose the optimal channel strategy involving Web- based channels [19]. The decision is comparatively easy for direct sellers, that is, enterprises that traditionally sell through mail-order catalogs [1], because they already have the infrastructure to support Web-based direct trade. All they have to do is switch their contact medium with their customers from paper-based catalogs to Web-based catalogs. However, the decision to change or modify the traditional channel strategy is not as easy for retailers, the other major group of off-line sellers. Their heavy investment in off-line transaction- specific assets makes it difficult for them to switch completely to on-line sell- ing. Even when retailers are ready to incorporate the Web-based channel in their overall channel strategy, the challenge for most of them comes less from the technology than from the complexities of managing a well-coordinated dual-channel structure [30]. The problem is more serious for firms that want to employ a hybrid model in which goods are showcased in one channel but sold on the other channel, or goods are sold in one channel but can be returned or exchanged at another channel. To separate the purchase of goods from fulfillment in that way, organizational and inter-firm incentives will have to be significantly altered, because the incentives for sales effort are usually tied to the measure of actual 126 KING, SEN, AND XIA sales. The most important issue is ensuring that incentives are aligned across multiple channels [19, 30]. Another concern is related to measuring the performance of on-line and off-line sales channels. For instance, a price- conscious consumer, after obtaining in-store sales advice or simply checking out the product in the off-line store, can discover the lowest price on the Internet and then buy from the channel that gives the lower price. This type of consumer behavior would make it difficult for the seller to measure the performance of its on-line and off-line channels. A final concern for firms following a coordinated dual-channel strategy is to improve the match between what consumers want and what they can get, thus reducing the compromise consumers have to make [19, 28]. In this regard, firms are in an awkward transition. Their cold d distributed inventory system for off-line stores competes with their cnew d centralized inventory-management system for on-line stores, although some hybrid systems try to integrate the two and, in turn, further increase the cost of coordination between the on-line and off-line channels. All these factors may explain why all sellers would adopt a coordinated dual- channel strategy. Managerial Implications What, then, are the implications of these findings for the traditional off-line seller? Before deciding on their optimal channel strategy, sellers need to be aware of the attitude of prospective customers and the size of the segment of buyers with low or high intentions to buy on-line. With time, on-line channels are going to become more advanced and reliable. As this lowers discomfort costs, a larger percentage of buyers will have high intentions to buy on-line. Most sellers will be forced into a dual-channel strategy (whether coordinated or uncoordinated) by the competition in their respective markets, as is already evident in certain industries. These sellers, however, will have to address the concerns raised by multichannel management if they are to implement successfully a coordinated dual-channel strategy. If they find it difficult to do so, they would be better off adopting an uncoordinated dual- channel strategy. Many leading retailers are already addressing the basic aspects of operating in a multichannel environment. Few, if any, however, have truly exploited the potential of integrated multichannel retailing to increase customer loyalty and economic returns. A report by IBM argues that consumers will become increasingly fragmented in the channels they use [27]. Therefore, rather than a short-term focus on maximizing the next transaction, the emphasis should be on optimizing customer lifetime value across the on-line and off-line channels. The report further suggests that the retailer 9s approach to multichannel integration should be flexible enough to adjust to changes in the market, the availability of technology, and customer expectations. At the same time, the

less