Market Lett (2009) 20:227243 DOI 10.

1007/s11002-009-9075-4

Private labels and new product development

Published online: 6 May 2009 # Springer Science + Business Media, LLC 2009

Abstract This paper provides a positive theory of private labels in new product development when a non-integrated distribution channel is faced with demand uncertainty. We consider a regular marketing environment in which a manufacturer endowed with a branded product seeks to design a new product to resolve its retailers mis-targeting problem and to optimally screen consumers. Assuming that only linear pricing schemes are available and that the retailer learns the state of demand earlier than the manufacturer does, we show that the presence of a private label always improves channel efficiency. Moreover, a private label is more likely to prevail when the existing branded product is a premium item. Keywords Private label . New product development . Channel coordination . Product line design . Demand uncertainty

C.-M. Chen Department of Finance, National Taiwan University, No. 50, Lane 144, Section 4, Kee-Lung Road, Taipei, Taiwan, Republic of China e-mail: [email protected] S.-Y. Chou (*) Department and Graduate Institute of Business Administration, National Taiwan University, No. 50, Lane 144, Section 4, Kee-Lung Road, Taipei, Taiwan, Republic of China e-mail: [email protected] L. Hsiao Department of Business Administration, National Chung Hsing University, No. 250, Kuo-Kuang Road, Taichung, Taiwan, Republic of China e-mail: [email protected] I.-H. Wu Graduate Institute of Business Administration, National Taiwan University, 5F, No. 637, Bei-An Road, Taipei, Taiwan, Republic of China e-mail: [email protected]

228

Market Lett (2009) 20:227243

1 Introduction The literature of optimal product line design (Mussa and Rosen 1978; Wilson 1993) has typically assumed that a single manufacturer (a vertically integrated channel) designs a product line and uses it to optimally screen consumers with differential valuations. Villas-Boas (1998) extends this literature by incorporating a downstream dealer into the picture, and considers a manufacturer that must design a product line to deal with the dealers mis-targeting problem and to perform optimal screening on consumers at the same time. To illustrate the idea of Villas-Boas (1998), imagine that there are high-valuation consumers (the highs) and low-valuation consumers (the lows) facing a vertically integrated channel as in Mussa and Rosen (1978), and the channel must design two products (optimal product line policy) and sell to the consumers at optimally chosen retail prices (optimal targeting policy). Now consider an otherwise-identical channel with the manufacturer facing an independent retailer. In this non-integrated channel, the retailers unit cost (which is the manufacturers unit revenue) is typically greater than the unit cost of its vertically integrated counterpart (which is the manufacturers unit production cost), and hence the retailer tends to abandon the lows more often than its vertically integrated counterpart would. This is the well-known double-marginalization problem. To induce the retailer to carry out the optimal targeting policy from the channels perspective, the manufacturer must concede rent to the retailer, usually in the form of trade promotions, and the concession of rent inevitably distorts the manufacturers incentive to carry out the optimal product line policy. Hence, the equilibrium efficiency in a non-integrated channel tends to be lower than that in a vertically integrated channel. We argue that the product line design theory of Villas-Boas (1998) has ignored one important option, which is to delegate the design of some products to the retailer. The purpose of this article is to identify the marketing environments in which such a delegation leads to a Pareto improvement for the channel members, and to clarify the role of private labels in leading to such an improvement. Our theory shows that a private label prevails in equilibrium if and only if the manufacturers product line design and targeting decisions fail to attain full channel efficiency, and the latter tends to occur when the manufacturer is a premium-brand manufacturer rather than a second-tier brand manufacturer. The traditional view about private labels has been that they reflect the growing power of retailers relative to manufacturers, and that the power results from growing concentration among retailers, improved scanner information systems used by retailers, manufacturers ineffective pull promotions (Messenger and Narasimhan 1995)1, and the emergence of a competitive procurement market for private labels (Kumar and Steenkamp 2007). Our theory will provide a new perspective, which supplements the traditional view by clarifying when and why concentration of retailers and their innovative information systems can actually transform into private labels. While not denying that a competitive procurement market can obviously
1

Currently private labels account for 20%, 34%, and 45% of sales in the United States, Germany, and the United Kingdom, respectively (Kumar 2007). In both Europe and the United States, the success of private labels has induced retailers to actively engage in R&D activities (Dunne and Narasimhan 1999).

Market Lett (2009) 20:227243

229

promote the prevalence of private labels, our theory points out that private labels can arise solely from manufacturers failure to attain full channel efficiency. Our story goes as follows. Consider a non-integrated channel consisting of a manufacturer and a retailer. The manufacturer has an existing branded product at hand, and is considering developing a new product to optimally screen two segments of consumers, the highs and the lows. To fix ideas, suppose that a product can be either a high-end item or a low-end item. Developing a new product is costly, and only after the cost of new product development is spent will the manufacturer and the retailer know whether the new product falls in consumers latitude of acceptance. (If it does, we say that the demand for the new product is in the success state; otherwise, it is in the failure state.) Suppose furthermore that the manufacturer and the retailer can only use linear pricing schemes,2 and that the manufacturer gets to see the state of demand only after it chooses the wholesale prices. We focus on a regular marketing environment where an otherwise-identical integrated channel wishes to screen consumers with both the high- and low-end items in the success state, and in the failure state it would serve either all consumers with the existing low-end item or only the highs with the existing high-end item. Our first result shows that, in the regular marketing environment, the manufacturers ex-ante effort for new product development and ex-post targeting decision both coincide with the corresponding choices made by its vertically integrated counterpart, if the manufacturers existing branded product is a low-end item. To understand this result, it is useful to first consider the case where the manufacturer can make its wholesale prices contingent on the realized demand state. Recall from the product line design literature that the manufacturer must offer a trade promotion to the retailer in a demand state where the manufacturer wishes the retailer to serve the lows.3 This implies that the manufacturer has to provide trade promotions in both the success and failure states if the existing product is a low-end item: in the success state, the manufacturer wants to induce the retailer to screen consumers with two products instead of serving only the highs using the high-end item; and in the failure state, the manufacturer wants to induce the retailer to sell the low-end item to all consumers rather than to the highs only. From the manufacturers perspective, these two mis-targeting problems are actually identical in nature, because the wholesale price of the low-end item must be so chosen that the retailers loss from charging the highs less is equated to the retailers gain from serving the lows. Consequently, if the existing product is a low-end item, and if the manufacturer can make its wholesale prices contingent on the realized demand state, then the manufacturer will choose the same wholesale price for the low-end item in both demand states. This implies that the retailer makes the same profit in
2

We make this assumption for several reasons. First, previous research has pointed out that only manufacturers in a more powerful position than their retailers may use a two-part tariff (Jeuland and Shugan 1983), and a manufacturer's using quantity discounts may violate RobinsonPaman Acts prohibition of price discrimination against retailers. Second, in reality, most wholesale pricing contracts are extremely simple, renewed on quarterly basis, and specify only unit prices (Blattberg and Neslin 1990). Third, in both economics and marketing literature, it is standard to assume that manufacturer offers linear pricing schemes to its retailer (Gerstner and Hess 1991, 1995; Villas-Boas 1998). 3 If the manufacturer wants the retailer to serve only the highs, all it needs to do is to prevent the retailer from pricing low, which can be accomplished by choosing a high wholesale price. In this case, the manufacturer does not need to provide trade promotions.

230

Market Lett (2009) 20:227243

both demand states: the retailer obtains its failure-state profit by selling the low-end item to all consumers, and it obtains its success-state profit by selling respectively the high-end item to the highs and the low-end item to the lows, which, as a binding incentive compatibility constraint, must generate a profit equal to what the retailer would make by serving all consumers with the low-end item; see for example VillasBoas (1998). Hence, the retailers equilibrium profit is invariant over the two demand states, given that the manufacturer will choose the same wholesale price for the low-end item in both demand states. It follows that, if the existing product is a low-end item, and if the manufacturer can make its wholesale prices contingent on the realized demand state, the manufacturers equilibrium profit is equal to the total channel profit minus a constant. This explains why the manufacturers ex-ante investment in new product development coincides with the corresponding choice made by its vertically integrated counterpart. Since the ex-post targeting decisions and the ex-ante investment in new product development both coincide with the integrated channels choices, channel efficiency is fully attained in this case. Our second result strengthens the first result by showing that, when the manufacturers existing branded product is a low-end item, full channel efficiency can be attained even if the manufacturer must choose the wholesale prices before seeing the demand state. The idea is that the manufacturer can simply adopt the wholesale prices that it would choose in the success state when it were able to make its wholesale prices contingent on the realized demand state. These wholesale prices are of course optimal if subsequently the manufacturer observes the success state; and if the failure state is observed instead, only the wholesale price for the low-end item remains to matter, which, thanks to the fact that the manufacturers optimal wholesale price for the low-end item is identical in both demand states, is again optimal. Hence, lack of demand information does not hurt the manufacturer if the existing branded product is a low-end item. Our first two results lead to the conclusion that channel efficiency is fully attained if the existing branded product is a low-end item, whether or not the manufacturer can make its wholesale prices contingent on the demand state. An important consequence of this conclusion is that the manufacturer will never delegate new product development to the retailer when the existing branded product is a low-end item: since the total channel profit is already maximized under the manufacturers marketing strategies, there is no offer about new product (private label) development which the retailer can make to the manufacturer and will make both firms better off. Things are different when the manufacturers existing branded product is a highend item. Now if the manufacturer can make its wholesale prices contingent on the realized demand state, then it will offer a trade promotion to the retailer only in the success state. In other words, the manufacturers profit falls short of its vertically integrated counterparts profit in and only in the success state. Our third result shows that, since the ex-ante effort for new product development is costly, and since its effect is to promote the likelihood of the success state, the manufacturer will underinvest in new product development compared to its vertically integrated counterpart in this case. Our fourth result then shows that, in the regular marketing environment under study, things only get worse when the manufacturer cannot make its wholesale prices contingent on the realized demand state. These two results show that channel

Market Lett (2009) 20:227243

231

efficiency cannot be fully attained by the manufacturers marketing strategies, if the manufacturers existing branded product is a high-end item. This is where a private label may arise as a remedy for channel inefficiency. We assume that the retailer can learn the demand state before the manufacturer chooses wholesale prices.4 Imagine that the retailer can make an offer to the manufacturer asking for the right to develop the new product (the private label). The retailers offer states that if the manufacturer allows the retailer to develop the new product, and agrees to produce the new product for the retailer,5 then the retailer will order certain amounts of the existing and the new products at some specified wholesale prices, but the retailer will also be allowed to make a new offer to the manufacturer after it receives demand information. (For other optimal offers that the retailer can choose to make; see footnote 18 below.) The ordered amounts specified in the contract are chosen to be an integrated channels sales volumes in the success state, and the wholesale prices specified in the contract are so chosen that by accepting the retailers offer the manufacturer can make a profit which equals what the manufacturer is expected to make by turning down the retailers offer and developing the new product on its own. Since the manufacturer can turn down any new offer and stick to the old offer when subsequently the retailer makes a new offer, and since the manufacturer can make the same expected profit by accepting the retailers first offer, we can safely assume that the retailers first offer will be accepted by the manufacturer. It remains to investigate whether the retailer really benefits from making such an offer and why the presence of a private label improves channel efficiency. Observe that, given the first offer, the retailer will have to pay a fixed amount of money to the manufacturer in exchange for exactly the amounts of the two products that a vertically integrated channel would sell in the success state. If subsequently the retailer observes the success state, then its optimal strategy is to stick to the first offer and to mimic the vertically integrated channels retail pricing strategy in the success state. This is optimal for the retailer because, given the amounts of the two products for sale, the retail prices that generate the maximum channel revenue are those prices chosen by the vertically integrated channel. Can the retailer do better by making a new offer to the manufacturer? The answer is negative because any acceptable new offer must not lower the manufacturers profit, but the sum of the two firms profits is already maximized under the vertically integrated channels retail pricing strategies. On the other hand, in the failure state, the retailers
4

Plenty of evidence suggests that retailers may possess demand information that manufacturers do not have. There is a huge body of literature analyzing retailers' motivations for sharing their superior demand information with the upstream suppliers in a supply chain; see for example Lee et al. (2000). In studying what might determine the magnitude of slotting allowances that retailers charge manufacturers for new products, Rao and Mahi (2003) find in their survey data that most retailers and manufacturers pointed to the information advantage that a retailer may possess over the manufacturer about the likely success of a new product. 5 Manufacturers of premium brands have supplied a significant portion of private labels (Boyle 2003, Fortune). Since our purpose is to provide a new perspective regarding why private labels may arise efficiently in equilibrium, we choose to build our theory in an environment where a retailer can only seek a branded manufacturer's cooperation in the production of the private label. We show that even in this worst possible situation from the retailer's perspective, it may still be the private label rather than a branded new product that prevails in equilibrium.

232

Market Lett (2009) 20:227243

optimal strategy is to cancel its order for the new product, and to adjust the wholesale price for the existing product (or to simply pay a fine; see footnote 18 below) in such a way that the manufacturers profit is unaffected (which ensures that the new offer will be accepted by the manufacturer). Once again, what the retailer does is to mimic the vertically integrated channels targeting and pricing strategies in the failure state. This will generate the maximum channel profit in the failure state, and since the manufacturer will be given a fixed profit, this mimicking strategy generates the maximum profit that the retailer can possibly make in the failure state. Thus by making the first offer to the manufacturer exante, ex-post in each demand state the retailer will make a profit which equals the vertically integrated channels profit minus a constant. This explains why the retailer and its vertically integrated counterpart will ex-ante spend the same amount on new product development. The bottom line is that, when the manufacturers existing product is a high-end item, a private label will prevail in equilibrium, and when it does, channel efficiency is fully attained. Our theory is consistent with a fundamental principle in contract theory, which asserts that to attain efficient joint production the decision right should be allocated to the informed party (here, the retailer) rather than the uninformed party (the manufacturer); see section 7.2.1 of Bolton and Dewatripont (2005). In our model, the retailer has demand information that the manufacturer does not have. To optimally extract rent from the retailer, the manufacturer (the uninformed party) may distort its new product development or pricing strategies. By allowing the retailer (the informed party) to develop a private label, the channel restores efficiency in its ex-ante expenditure on new product development and ex-post targeting strategy. Our theory generates the following predictions. First, premium-brand manufacturers (whose brands are high-end items) do not resist private labels as much as second-tier brand manufacturers (whose brands are low-end items) do. This result is consistent with the findings in Pauwels and Srinivasan (2004). Second, private labels are accompanied by deep trade promotions from branded manufacturers.6 This result is consistent with the finding in Chintagunta et al. (2002) that, within the oats product category, the store brand introduction coincides with an increase in the retailers margins for the national brand. Third, retailers information advantage over manufacturers promotes the market share of private labels. The existing literature has emphasized retailers power relative to manufacturers in leading to the prevalence of private labels. Our theory points out that retailers information advantage about the likely success of new products can be a main source of their power. Fourth, across countries, product categories, and industries, those in which manufacturers are more sophisticated in acquiring demand information tend to have lower market shares for private labels.7 Thus, our theory provides an explanation to the variations in the market shares of private labels
In our model, a private label prevails in equilibrium if and only if the manufacturer accepts the retailers offer. In the absence of such an offer, the manufacturer would induce the retailer to serve only the highs, and would offer no trade promotion. 7 In our model, private labels arise because the manufacturers marketing strategies may fail to attain channel efficiency, which is partly due to the manufacturers lack of demand information. We thank an anonymous reviewer for an inspiring comment that allows us to discover this implication.
6

Market Lett (2009) 20:227243

233

observed in the US, Germany, and the UK. Fifth, consumers who purchase a private label tend to have a lower hedonic value for the category than their peers purchasing a national brand.8 The remainder of this article is organized as follows. In Section 2, we describe the game-theoretic model that depicts the interactions among a branded manufacturer, a retailer, and two segments of consumers. Main results and the underlying insights are reported in Section 3. Concluding remarks are given in Section 4.

2 The model Consider a non-integrated channel that consists of a branded manufacturer (M) and a retailer (R). M is endowed with a branded product E, and M and R can each spend a cost to develop a new product N, where E and N stand for existing and new products. Unlike in the literature of optimal product line design, where a product is solely described by its (vertical) quality level, here the new product must be described by its horizontal attributes as well as its vertical quality. More specifically, there are two classes of consumers, referred to as the highs and the lows, and their populations are respectively and (1). Each consumer may either purchase one unit of E, or one unit of N, or nothing. The highs and the lows have different ideal points for the new products horizontal attributes, and M and R do not have perfect information about the distribution of consumers tastes in this regard. By spending k2 /2 on market research, however, M and R can each design a new product that with probability may fall in the latitude of acceptance of all consumers (referred to as the success state, and denoted by G), but with probability (1) the new product may fall outside that latitude of acceptance (referred to as the failure state, and denoted by B), and we assume for simplicity that in the failure state no consumers are willing to buy the new product. For a product that falls in consumers latitude of acceptance (and we assume that the existing product always does), the vertical quality determines a consumers willingness to pay for it. For simplicity, we assume that only two quality levels are feasible, and hence a product can either be a high-end or a lowend item, and the unit production costs for the high- and the low-end items are respectively c2 and c1. Thus there are two possible cases to consider: (i) the case where the existing product is a high-end item and the firms must decide how much to spend for the development of a low-end item; and (ii) the case where the existing product is a low-end item and the firms must decide how much to spend for the development of a high-end item. The following table summarizes consumers valuations for a product with horizontal attributes that fall in their latitude of acceptance (Table 1).
8

In our model, the private label appears in equilibrium only if it is a low-end item and only if the demand is in the success state. In that situation, the existing and the new products will typically be used to screen consumers, with the private label and the national brand being respectively sold to the lows and the highs. The lows have a lower hedonic value than the highs for the category.

234 Table 1 Valuations of consumers Low-end product Lows (proportion 1) Highs (proportion ) 1 ' q1

Market Lett (2009) 20:227243

High-end product 1 2

Assumption 1 The parameters satisfy the following regularity conditions.9 ' (i) q2 > q 1 > q 1 > c2 > c1 > 0; ' (ii) q2 c2 > q 1 c1 ; n o iii Min bq ; bq' 1 bc b1 bq c > q > b2 bq '
2 1 1 2 2 1 1

1 b 2 c1 ; && h i2 '0h  i ' iv k > Max 1 b q 1 bq' 1 bc1 2b q 1 q1 ; b q2 c2 1 o  i q' c1 ; q1 bq' 1 bc1 ; 0 : 1 1

The interactions among M, R, and consumers are described in the following extensive game. Stage 1 R can make a take-it-or-leave-it offer (QHR, QLR, WHR, WLR) to M,10 stating that (i) R will order QHR units of the high-end product and QLR units of the low-end product from M at some wholesale prices WHR and WLR if M allows R to develop the new product and agrees to produce the new product for R; and (ii) R can make a new ' ' ' ' take-it-or-leave-it offer QHR ; QLR ; WHR ; WLR to M at Stage 3. If M accepts Rs offer, then R chooses ; or else, M chooses . We shall assume that R will not make an offer unless making the offer strictly raises Rs expected profit, and that M will do as R says as long as cooperating does not reduce Ms expected profit. Stage 2 M announces its wholesale prices WH and WL, if M has turned down Rs offer in Stage 1. (We shall also need to discuss the case where M can make its wholesale prices contingent on the realized demand state, and in that case Wij denotes Ms wholesale price for product item i in demand state j.) Stage 3 R privately sees the state of demand for the  product. Then R can make new  ' ' ' a new take-it-or-leave-it offer Q' ; QLR ; WHR ; WLR to M if M has accepted Rs HR
Condition (i) ensures that every targeting strategy generates a positive total surplus. Condition (ii) says that the total surplus from serving the highs is maximized by selling the highs the high-end item rather than the low-end item. Condition (iii) ensures that in the absence of the low-end item, the retailer always prefers to sell the high-end item to the highs only, that unit production costs for the two products do not differ by too much, and that whether the distribution channel is vertically integrated or not, the manufacturer endowed with two products that both fall in consumers latitude of acceptance always wishes to screen consumers with different products. Condition (iv) ensures that new product development is costly enough to become a relevant issue. We discuss these conditions in Section 4. 10 Like M, R is restricted to choose only a linear pricing scheme (WHR, WLR). No matter which one between M and R gets to choose (WHR, WLR), R is the one who will then specify the ordered quantities (QHR, QLR).
9

Market Lett (2009) 20:227243

235

first offer (QHR, QLR, WHR, WLR) in Stage 1.11 The manufacturer can choose to stick to the first offer or accept this new offer in this case. Stage 4 If M has accepted Rs offer in Stage 1, then there is a contract between M and R at the end of Stage 3, and R must carry out the transaction with M as that contract specified. Otherwise, given the wholesale prices that M chooses in Stage 2, R must decide how much to order for each product. Finally, R must choose its retail prices PHG, PHB, PLG, and PLB before consumers arrive, where note that retail prices can be made contingent on the realized demand state, and Pij is the price for product item i in demand state j. Stage 5 Consumers arrive, and given the retail prices they make purchase to maximize consumer surplus. We will show in Section 3 that the equilibrium of this game crucially depends on whether the existing branded product is a high-end or a low-end item. 3 Results 3.1 The case of low-end national brand As a benchmark, we first consider the vertically integrated channel where M and R are the same business entity. This implies that M is not facing Rs mis-targeting problem, and the retail prices can be made contingent on the realized demand state. 3.1.1 The integrated channel   ' Following Assumption 1, given , Ms optimal retail prices are PHG q2 q 1 q1 and PLG =PLB =1. Note that PHB is unspecified because consumers ignore the high-end new product  state B. Ms state-G and state-B profits are respectively in h  i ' b q2 q1 q 1 c2 1bq1 c1 and 1 c1. HenceinStage1,Mseekstomaximize n h o   i ' a b q 2 q 1 q 1 c2 1 b q 1 c1  1 aq1 c1 ka2 2 by choosing the optimal . The optimal solution for the integrated channel is h  i. a b q 2 c2 q ' c1 k 1 3.1.2 The non-integrated channel It is useful that we start with the sub-game where Rs offer in Stage 1 has been turned down by M. We shall move backward and analyze Stage 1 in Proposition 1
The assumed interactions between R and M try to capture the widely documented phenomena in distribution channels: concentration of retailers implies that they may be able to make a firm offer when negotiating with manufacturers, and innovation of information systems may explain why they may receive demand information before manufacturers do.
11

1 2

236

Market Lett (2009) 20:227243

after we complete the analysis for this sub-game. We shall first consider case where the manufacturer can see the demand state when choosing the wholesale prices. The manufacturers optimal strategy in this case is, under Assumption 1, to induce the retailer to screen consumers with two products in state G, and to serve all consumers with the low-end item in state B.12 Hence, the optimal wholesale prices WHG, WLG, and WLB must solve the following manufacturers profit-maximization problem; see Villas-Boas (1998): Max bWHG c2 1 bWLG c1 s:t: h   i ' b q2 q1 q 1 WHG 1 bq1 WLG ! bq2 WHG ; h   i ' b q2 q1 q1 WHG 1 bq 1 WLG ! q 1 WHG ; h   i   ' b q 2 q1 q1 WHG 1 bq 1 WLG ! b q ' WLG ; 1 h   i ' b q2 q1 q 1 WHG 1 bq1 WLG ! q1 WLG ; ' b q2 q1 q 1 WHG 1 bq1 WLG ! 0: h   i 3 3 4 5 6 7

Lemma 1 Suppose that the existing product is a low-end item, that Rs offer in Stage 1 has been rejected by M, and that Assumption 1 holds. If the manufacturer can make its wholesale prices contingent on the realized demand state, then the optimal state-G wholesale prices for the high- and low-end items are respectively WHG  .  . ' q ' q q2 q1 1 b and WLG q1 b q 1 1 b, and . optimal the 1 1  ' q state-B wholesale price for the low-end item is WLB q 1 b q1 1 b. 1 h  . ' q The manufacturers state-G and state-B profits are equal to b q2 q1 1  . h  . i ' 1 b c2 1 b q1 b q' q 1 1 b c1 and q 1 b q1 q 1 1 1 b c1 , respectively. Ex-ante, the manufacturer optimally chooses a aL b h  i. q 2 c2 q' c1 k , which is the same choice that a vertically integrated 1 channel would make.

In state B, the retailer would rather sell the low-end item to all consumers than to only the highs if and   only if q1 WLB ! b q' WLB , and hence the manufacturer is better off serving all consumers if and 1  .   1 b c1 ! b q ' c1 , which is true under Assumption 1. In state G, the only if q1 bq' 1 1 manufacturer may or may not want to serve the lows, but if it does, it never pays to serve the lows with the high-end item, and moreover, Assumption 1 (2 >1) implies that serving all consumers with the low-end item is dominated by screening consumers with different product items. Hence, the manufacturer is left with two un-dominated state-G strategies: to serve only 0the highs with the high-end item, or to screen consumers with two items. Because q 1 > b2 bq 1 1 b2 c1 , Assumption 1 implies that screening is the optimal strategy in state G.

12

Market Lett (2009) 20:227243

237

Lemma 1 shows that if the existing product is a low-end item and if the manufacturer can see the demand state when choosing the wholesale prices, then the non-integrated channel attains full channel efficiency: its ex-ante investment in new product development and ex-post targeting strategy in each demand state all coincide with the corresponding choices made by its integrated channel counterpart. To understand Lemma 1, note that in state B the manufacturer wishes the retailer to price the low-end item at 1, but the retailer is tempted to price it at q ' . To gain 1 cooperation from the retailer in state B, the optimal state-B wholesale price for the low-end item, W , mustequate the retailers loss from charging the highs less LB ' (which equals b q1 q1 ) to the retailers gain from serving the lows (which equals 1 b q 1 WLB ). Similarly, in state G the manufacturer wishes the retailer   ' to screen consumers by selling the two items at retail prices PHG q2 q1 PLG and PLG =1 respectively, but the retailer is tempted to sell only the high-end item at PHG =2 to the highs. In order to gain cooperation from the retailer, the state-G wholesale price for the low-end item, WLG , must again equate the io n h   retailers loss from   ' P ' charging the highs less (which equals b q2 q2 q1 b q1 q1 ) LG to the retailers gain from serving the lows (which equals 1 b q1 WLG ). Consequently, we have WLB WLG ; that is, the manufacturers optimal wholesale price for the low-end item is the same over the two demand states. Since in each demand state, the retailers equilibrium profit is equal to what it would make by selling only the low-end item to all consumers,13 and since the manufacturers optimal wholesale price for the low-end item is identical over the two demand states, we conclude that the retailer must receive the same profit in the two demand states. We have reached the conclusion that if the manufacturers existing product is a lowend item, and if the manufacturer can make its wholesale prices contingent on the realized demand state, then in equilibrium the manufacturers profit is equal to the total channel profit minus a constant. This explains why in Lemma 1 the manufacturer is willing to spend the same amount on new product development as its vertically integrated counterpart does. Now Lemma 2 is an immediate consequence of Lemma 1. Lemma 2 Suppose that Assumption 1 holds, that Rs offer in Stage 1 has been rejected by M, and that the existing product is a low-end item. Without being able to see the demand state when choosing the wholesale prices, the manufacturers  . ' optimal pricing strategy is WH WHG q 2 q 1 q1 1 b and WL  . WLG q1 b q ' q1 1 b; where WHG and WLG are defined in Lemma 1. 1 Moreover, the manufacturers state-G and state-B profits coincide respectively with its state-G and state-B profits stated in Lemma 1. Consequently, the manufacturer h  i. optimally chooses a aL b q2 c2 q ' c1 k ex-ante. 1 Lemma 2 can be understood as follows. When the manufacturer must choose wholesale prices before seeing the demand state, it can simply choose the wholesale
This is indeed the retailers state-B equilibrium profit. In state G, note that selling only the low-end item to all consumers is always an option to the retailer, and to deal with this mis-targeting problem, the manufacturers optimal wholesale prices WLG and WHG must equate the retailers equilibrium profit to what it would get by selling only the low-end item to all consumers.
13

238

Market Lett (2009) 20:227243

prices that it would choose in the success state when it could make its wholesale prices contingent on the realized demand state. These prices are optimal in the success state of course, and in the failure state only the low-end wholesale price will remain to matter, which, thanks to the fact that WLG WLB , is again optimal in the failure state. Lemmas 1 and 2 allow us to draw a conclusion about how likely private labels may prevail in the presence of a low-end national brand. Moving backward to Stage 1, we obtain the following result. Proposition 1 When the existing branded product is a low-end item, R will not make an offer to M in equilibrium, so that private labels cannot prevail in equilibrium.14

3.2 The case of high-end national brand We again start with the benchmark case of a vertically integrated channel. 3.2.1 The integrated channel In state G, the integrated channels optimal retail prices are identical as in Section 3.1. In state B, Assumption 1 implies that, unlike in Section 3.1, now the integrated channel prefers to sell the high-end item only to the highs.Hence, the integrated channels stateh  i G and state-B profits are respectively b q q' q c 1 bq c
2 1 1 2 1 1

and bq2 c2 . Ex-ante, the integrated channel seeks to maximize n h o   i ' a b q2 q1 q 1 c2 1 bq1 c1 1 abq2 c2  ka2 2 8

by choosing the optimal . In this case, the optimal solution for the integrated channel is15 h i. aH q 1 bq ' 1 bc1 k 9 1 3.2.2 The non-integrated channel We again start with sub-game where Rs offer in Stage 1 has been rejected by M. Also, we shall first consider the case where the manufacturer can see the demand state when choosing the wholesale prices. The manufacturers optimal strategy in
To see the idea, suppose instead that R does make an offer in equilibrium, which is accepted by M. This implies that by carrying out the transactions stated in the offer, Rs expected profit is strictly higher and Ms expected profit is not lower than without this offer. This contradicts the fact that without this offer the (expected) total channel profit has already been maximized. 15 Since the integrated channels state-G profit is the same as in Section 3.1, whether it spends more on new product development than in Section 3.1 depends on its state-B profits in the two cases. It spends more than in Section 3.1 if its state-B profit is lower than in Section 3.1.
14

Market Lett (2009) 20:227243

239

this case is, under Assumption 1, to induce the retailer to screen consumers with two products in state G, for exactly the reasons stated in Section 3.1, and to serve only the highs with the high-end item in state B. Hence, the same wholesale prices WHG, WLG solve the manufacturers profit-maximization problem in state G, and it is obvious that the optimal WHB q2 . Lemma 3 Suppose that the existing product is a high-end item, that Rs offer in Stage 1 has been rejected by M, and that Assumption 1 holds. If the manufacturer can make its wholesale prices contingent on the realized demand state, then the optimal state-G wholesale prices for the high- and low-end items are respectively WHG  .  . ' q ' q q2 q1 1 b and WLG q1 b q1 1 b, and the optimal 1 1
state-B wholesale price for the high-end item is WHB. q2 . The manufacturers stateh  i ' q G and state-B profits are equal to b q2 q1 1 b c2 1 b q 1 1  . ' Ex-ante, the manufacturer b q1 q1 1 b c1 and bq2 c2 respectively. i h ' 1 b2 c =k 1 b < a . optimally chooses a q1 b2 bq1 1 H

Lemma 3 shows that, although the targeting strategies in the two demand states coincide with the integrated channels choices, the manufacturer has an underinvestment problem in its ex-ante effort for new product development.16 The underinvestment stems from the fact that the retailers mis-targeting problem costs the manufacturer only in the success state. Thus, ex-ante, the manufacturer is less willing to promote the likelihood of the success state by spending on new product development. Hence, unlike in Section 3.1, channel efficiency is not fully attained in Lemma 3. Lemma 4 below shows that if the existing product is a high-end item, and if the manufacturer in the non-integrated channel must choose wholesale prices before seeing the demand state, then under Assumption 1 neither the ex-post targeting strategies nor the ex-ante new product development attains channel efficiency. Lemma 4 Suppose that the existing product is a high-end item, that Rs offer in Stage 1 has been rejected by M, and that Assumption 1 holds. If the manufacturer cannot make its wholesale prices contingent on the realized demand state, then its optimal strategy is to choose =0; i.e., to give up new product development entirely. Ex-post, the manufacturer always induces the retailer to sell the high-end item to only the highs. The optimal wholesale price is WH =2, and the manufacturer provides no trade promotion in equilibrium. The manufacturers equilibrium profit is (2 c2). Lemma 4 can be understood as follows. Since the manufacturer cannot vary the wholesale prices over the two demand states, and since choosing WH =2 is optimal in state B, choosing WH <2 in order to induce the retailer to serve the lows can never be optimal if it is accompanied by some >0 very close to zero. This happens because for close to zero, the manufacturer is almost sure to lose (2 WH) from
16

Note that Lemma 3 remains true even if we relax most conditions imposed in Assumption 1. What matters is that the retailers mis-targeting problem does not cost the manufacturer in state B, which is generally true when the existing branded product is a high-end item.

240

Market Lett (2009) 20:227243

serving the highs, and yet the expected revenue from serving the lows is close to zero. Thus, to induce the retailer to sell the low-end item to some or all consumers, must be chosen to be sufficiently large, which cannot be worthwhile if k>0 is sufficiently large, as stated in condition (iv) of Assumption 1. When the manufacturer can make its wholesale prices and targeting strategy contingent on the realized demand state, as in the cases depicted in Lemma 1 and Lemma 3, or when the channel is vertically integrated, the manufacturer can always choose its ex-post pricing strategy independently of its ex-ante decision about new product development. Things are different if the manufacturer cannot make its wholesale prices (and hence targeting strategy) contingent on the realized demand state. Note that the manufacturer is concerned about the waste in over-provision of trade promotion for the high-end item when the new product development fails in state B, and to minimize that waste the manufacturer must invest a lot in new product development ex-ante, so that the ex-ante and ex-post marketing decisions are intertwined in this case. Lemma 4 shows that, when new product development is sufficiently costly, the manufacturers optimal choice is to spare new product development and avoid the waste in over-provision of trade promotion for the highend item in state B. Hence, Lemma 4 highlights a channel inefficiency that arises from the manufacturers lack of demand information when making pricing decisions. Proposition 2 Suppose that the existing product is a high-end item and that Assumption 1 holds. If R does not make an offer to M in Stage 1, then in the subgame equilibrium M provides no trade promotion to R, and neither a low-end branded product nor a private label will be produced. Proposition 2 foretells the efficiency role of the retailers offer, which we now proceed to clarify. Consider the first stage of the game, where R is ready to make an offer to M. Consider the following offer that R makes to M in Stage 1: for some t satisfying q2 c2 > t > 0; QHR ; QLR ; WHR ; WLR b; 1 b; q2 t; c1 bt=1 b 10

Note that t>0 is the trade promotion that the manufacturer is required to provide the retailer for the high-end branded product. Recall from Lemma 4 that by turning down the retailers offer the manufacturer can make the profit (2 c2). By accepting the above offer, the manufacturers profit is at least bWHR c2 1 bWLR c1  q2 c2 , since the manufacturer can subsequently turn b  ' ' ' ' down any new offer QHR ; QLR ; WHR ; WLR that gives it a lower profit. Hence the offer defined in Eq. (10), once made, will be accepted by M. It remains to investigate whether the retailer has an incentive to make this offer in the first place. Note that after that offer is made and accepted, the retailer must pay the manufacturer the fixed amount WHR +(1)WLR in exchange for units of the high-end item and (1) units of the low-end item. To see what happens next, first suppose that the demand state is G. Upon seeing state G, the retailer has two options: it can stick to the offer that it made to the manufacturer, and try to sell the units of the high-end item and (1) units of the low-end item to consumers, or it can propose some

Market Lett (2009) 20:227243

241

  ' ' ' acceptable new offer Q' ; QLR ; WHR ; WLR to M, and see if it can do better. We claim HR that in this case Rs best strategy is to stick to the old offer, and to mimic the vertically integrated channels retail pricing strategy. If R chooses to stick to the old offer, then it must maximize the revenue from selling the units of the high-end item and (1) units of the low-end item. Essentially, the retailer becomes an integrated channel facing no variable costs. Following Assumption 1, the retailers optimal retail   ' prices are PHG q 2 q 1 q1 and PLG =1, which coincide with the retail prices that the vertically integrated channel would choose in state G, and hence generates the maximum state-G total channel profit for the retailer and the manufacturer as a whole. Since with the retailers offer, the retailers profit is equal to the total channel profit minus the constant (2 c2), there is no new offer that can be acceptable to the manufacturer and can raise the retailers profit at the same   time. Thus, the retailers state-G profit is equal to 1 bq 1 c1 b q ' q1 , 1 which is positive because it is the difference between the maximum total channel profit achieved by the vertically integrated channel and the channel profit generated by the manufacturers suboptimal strategy described in Lemma 4. What if the demand state is B? We claim in this case the retailer should optimally that  ' ' ' make the new offer Q' ; QLR ; WHR ; WLR b; 0; q 2 ; 0 to the manufacturer. (Note HR that what the retailer does is again to mimic the vertically integrated channels state-B targeting strategy.) Now, by agreeing to replace the first offer by this new offer, the manufacturer can make exactly the same profit, which is (2 c2), and hence the manufacturer should accept the new offer. With the new offer, the retailers state-B profit is zero, which is greater than what the retailer would make by sticking to the first offer, which is bq 2 bWHR 1 bWLR 1 bc1 < 0 11

Recall from Proposition 2 that the retailer gets zero profit if it makes no offers in Stage 1. Thus, we have shown that the retailer is better off making the first offer defined in Eq. (10) to the manufacturer in Stage 1.17 Since that offer attains channel efficiency without increasing the manufacturers profit, it is one best offer from the retailers perspective.18 Moreover, since with that offer the retailers equilibrium profit equals the total channel profit minus a constant, the retailers ex-ante expenditure on new product development also attains channel efficiency. Proposition 3 Suppose that the existing product is a high-end item and that Assumption 1 holds. Then in equilibrium the retailer will propose to develop the lowNote that the presence of the first offer avoids the un-desired outcome where both firms spend on new product development ex-ante, and the chance to make a new offer allows the channels retail pricing and production decisions to adapt to new demand information. 18 The above offer is merely one optimal choice. Several other alternatives can each attain full channel efficiency as well. For example, Rs first offer only asks for (i) Ms permission to develop the new product; (ii) Ms promise of producing the new product for R; and (iii) a right to offer some (QHR, QLR, WHR, WLR) after R learns the demand state. In this case R will offer (QHR, QLR, WHR, WLR)=(, 1, 2 t, c1 +t/(1)) in and only in state G. For another example, Rs first offer specifies (QHR, QLR, WHR, WLR)=(, 1, 2 t, c1 +t/(1)) and, instead of allowing R to make a new offer subsequently, it specifies a fine t on R if R chooses to cancel its order for the low-end item subsequently. In this case, R will cancel the order and pay the fine in and only in state B.
17

242

Market Lett (2009) 20:227243

end new product (a private label), which the manufacturer will approve. In equilibrium, channel efficiency is fully attained.

4 Conclusion We have developed a positive theory for private labels, showing that the presence of private labels is an indication of channel efficiency, and that allowing retailers to develop private labels raises the likelihood of new product introduction and its success rate. This view is consistent with the findings in Mills (1995) and Bontems et al. (1999). Our theory also implies that, among other things, (i) prevalence of private labels tends to be accompanied by deep trade promotions in national brands; and (ii) premium-brand manufacturers do not resist private labels as much as secondtier brand manufacturers do. These predictions are consistent with empirical findings. We have focused on a regular marketing environment where an otherwiseidentical integrated channel wishes to screen consumers with both the high- and lowend items in the success state, and in the failure state it would serve either all consumers with the existing low-end item or only the highs with the existing highend item. This has been the leading case that product line design theorists find most relevant and interesting. In environments not examined in this paper, it is possible that a private label may fail to prevail even with a high-end national brand. However, phenomena considered irregular from the perspective of the product line design literature may arise in those situations.19
Acknowledgments The authors would like to thank the anonymous reviewers and the editor for their helpful suggestions. Shan-Yu Chou also gratefully acknowledges the financial support (NSC 93-2416-H002-010-) from the National Science Council of Taiwan, Republic of China.

References
Blattberg, R. C., & Neslin, S. N. (1990). Sales promotion: concepts, methods, and strategies. Upper Saddle River: Prentice-Hall. Bolton, P., & Dewatripont, M. (2005). Contract theory. Cambridge: MIT Press. Bontems, P., Monier-Dilhan, S., & Requillart, V. (1999). Strategic effects of private labels. European Review of Agricultural Economics, 26(2), 147165. Boyle, M. (2003). Brand killers. Fortune, 148(3), 89101. Chintagunta, P. K., Bonfrer, A., & Song, I. (2002). Investigating the effects of store brand introduction on retailer demand and pricing behavior. Management Science, 48(10), 12421267. Dunne, D., & Narasimhan, C. (1999). The new appeal of private labels. Harvard Business Review, 77(3), 4152. Gerstner, E., & Hess, J. D. (1991). A theory of channel price promotions. American Economic Review, 81 (4), 872886. Gerstner, E., & Hess, J. D. (1995). Pull promotions and channel coordination. Marketing Science, 14(1), 4360.

19

For example, the manufacturer with a low-end existing product will always serve only the highs with its most valuable item (i.e., the new product in state G and the existing product in state B) if a set of conditions on the parameters hold.

Market Lett (2009) 20:227243

243

Jeuland, A. P., & Shugan, S. M. (1983). Managing channel profits. Marketing Science, 2(3), 239272. Kumar, N. (2007). The right way to fight for shelf domination. Advertising Age, 78(4), 1415. Kumar, N., & Steenkamp, J.-B. E. M. (2007). Private label strategy: how to meet the store brand challenge. Boston: Harvard Business School Publishing. Lee, H. L., So, K. C., & Tang, C. S. (2000). The value of information sharing in a two-level supply chain. Management Science, 46(5), 626643. Messenger, P. R., & Narasimhan, C. (1995). Has power shifted in the grocery channel? Marketing Science, 14(2), 189223. Mills, D. E. (1995). Why retailers sell private labels. Journal of Economics & Management Strategy, 4(3), 509528. Mussa, M., & Rosen, S. (1978). Monopoly and product quality. Journal of Economic Theory, 18(2), 301 317. Pauwels, K., & Srinivasan, S. (2004). Who benefits from store brand entry? Marketing Science, 23(3), 364390. Rao, A. R., & Mahi, H. (2003). The price of launching a new product: empirical evidence on factors affecting the relative magnitude of slotting allowances. Marketing Science, 22(2), 246268. Villas-Boas, J. M. (1998). Product line design for a distribution channel. Marketing Science, 7(2), 156 169. Wilson, R. B. (1993). Nonlinear pricing. New York: Oxford University Press.