NEW PRODUCT LAUNCHES AND MANUFACTURERS’ RETURNS: ADDRESSING THE CHALLENGES OF LAUNCHING NEW PRODUCTS THROUGH LARGE RETAILERS By Tereza Dean A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILSOPHY Marketing 2012 ABSTRACT NEW PRODUCT LAUNCHES AND MANUFACTURERS’ RETURNS: ADDRESSING THE CHALLENGES OF LAUNCHING NEW PRODUCTS THROUGH LARGE RETAILERS By Tereza Dean Manufacturers are increasingly turning to innovation and the development of new products in the hopes of generating increased sales and profits (Abetti 2000, Chandy and Tellis 2000). However, the value that innovation efforts generate for firms is limited when new products are launched through large retailers, because manufacturers must share a portion of the value with these retailers. This is a challenge, since manufacturers and retailers are rivals for value extraction. For example, manufacturers often complain that retailers are creative in finding unpredictable methods of extracting additional revenues (Iyer and Villas-Boas 2003) and that retailers gain higher shares of profits at their expense (Dukes et al. 2006). Additionally, manufacturers are increasingly dependent on their retailers. Since the concentration of retailing across several sectors limits the distribution channels available to manufacturers (Hultink 1999), retailers have begun to play an increasingly important role in the successful launch of new products. While the growing power of retailers creates challenges for manufacturers, research on new product launches is currently limited. It has not yet addressed distribution-related issues and the difficulties that manufacturers face when launching new products. Understanding and addressing these challenges is important, because the development of strong distribution channels and the establishment of appropriate channel activities play a particularly critical role in any new product launch (DiBenedetto 1999). To address these issues, this two-essay dissertation will investigate the challenges that manufacturers face when launching new products through retailers. The first essay draws on Governance Value Analysis and examines two key research questions: how manufacturers govern the introduction of new products and how this governance impacts retailers’ behavior after new products are launched. The focus of this essay is on managing a single new product introduction. The second essay draws on the literature on reciprocity and examines value creation and value sharing problems between manufacturers and retailers, specifically, how manufacturers’ new product launch decisions impact manufacturers’ returns. The focus of this essay is on decision-making with respect to multiple new product introductions. Overall, the research presented here provides insights for marketing academics as well as marketing managers on how manufacturers should structure their relationships with retailers in order to enhance value creation and value extraction from new product launches. Copyright by Tereza Dean 2012 ACKNOWLEDGEMENTS The PhD program marks one of the most challenging and difficult time periods in my life full of struggles and doubts. At the same time, it also marks a wonderful period in which I have learned to lean on help and kindness of others who were there for me throughout the entire process and without whom I would never achieve this goal. First, a great amount of thanks goes to my mentor and a committee chair Dr. Roger J. Calantone. He has been my advisor for the entire duration of the PhD program and guided me through all of the necessary steps. He has generously shared his experience whenever I needed it and has been incredibly patient with me. I especially appreciate his willingness to allow me to uncover and conduct research on topics that I truly love (the dissertation being one of them). Without his invaluable support, guidance, and encouragement this dissertation would not have been possible. I would also like to thank my dissertation committee: Dr. David A. Griffith, Dr. Cornelia L. Droge and Dr. Ranjani Krishnan. Specifically, Dr. David Griffith deserves recognition for voluntarily spending countless hours on my dissertation and for continuously providing me with detailed feedback without which it would not be possible to complete this dissertation. Throughout the duration of the PhD program, he also patiently mentored me and taught me to become a more responsible researcher. On numerous occasions it was his encouragement that gave me the strength to continue. I will be forever indebted to him. Dr. Cornelia Droge also provided me with invaluable insight. It was her attention to detail and desire for precision that taught me so much about what it really means to be a scholar and a researcher, and how difficult this may be. My dissertation would not be what it is today without her. Additionally, I would like to thank Dr. Ranjani Krishnan for her support and guidance v throughout the entire dissertation process. Dr. Clay Voorhees and Dr. Berna Devezer also deserve a special thanks and recognition. Although they were not members of my committee, they were both willing to help and to provide me with guidance in their respective areas of expertise. I admire that and for their help I am truly grateful. I also consider myself blessed to find great support among other doctoral students who were there for me, and with me, every step of the way. Special thanks belongs to Praneet Randhawa who provided invaluable help in the most important and difficult times. I also want to thank Hugo DeCampos for his kindness, support, and feedback, and Brian Baldus and Hannah Lee for their willingness to engage in long methodological discussions. Further, I extend my thanks to my friends Matthew Castel, Robert Wiedmer, David Dreyfus, and Yanhui Zhao for always encouraging me, strengthening me, and for bringing a smile to my face even in the most difficult times. I also wish to thank Wyatt Schrock and Chris Fleming. It has truly been an honor to be in the PhD program with these wonderful people and I will always remember and cherish their friendships. Special recognition goes to my family. My husband, David Dean, has done everything he could to support me. He has made tremendous sacrifices so that I can complete the program and has been there through the good and the bad. Without his support, kindness, and patience, completion of the PhD program would not have been at all possible. My parents, Pavel Šedlbauer and Jiřina Šedlbauerová also deserve special thanks. They have both done everything humanly possible to help me through this process. My father has supported me, consulted me, encouraged me, and frequently calmed me down. My mother, although located in Europe, has come many times to support me, to watch my son, to nourish me, and to do everything a good mother does. I am forever grateful to them. vi Most importantly, none of this would have been possible without my God, Lord Jesus Christ who truly deserves all the glory and honor. vii TABLE OF CONTENTS LIST OF TABLES………………………………………………………………………... x LIST OF FIGURES...……………………………...……………………………………... xi INTRODUCTION...…………..………………………………………………………….. 1 ESSAY 1 NEW PRODUCT LAUNCHES AND MANUFACTURERS` RETURNS: UNDERSTANDING GOVERNANCE DECISIONS BETWEEN MANUFACTURERS AND RETAILERS FOR NEW PRODUCTS…………………………………………….. Theoretical Background…….…………………………………………………………... Hypotheses………………….…………………………………………………………... New Product Innovativeness..………………………………………….................. Market Uncertainty………………………………………….................................. Performance Ambiguity……..……………………………………...……….......... Frequency of New Product Introduction.…………..……………………….......... Contract Specificity..……………………………………….……………...……… New Product Success……...…………...…………….………….………...……… Methodology……………….………………………………………………………........ Overview………………...…………………………………………….………….. Study 1: Field Survey…………………………………………………………………… Research Context and Data Collection………………………………………........ Measures…………………...…………………………………….……………….. Control Variables ..……………………………………...............................……… Analysis and Results..………….………………………………………...……...……… Measurement Model Analysis..………………….………………………...……… Hypotheses testing..……………...………………………………………...……… Discussion of the results for Study 1..……………………………………...…...… Study 2: Experimental Design..……….………………………………………...……… Sampling and Data Collection ..…………………………………………...……… Stimuli and Measures ..……………………………..……………………...……… Control Variables ..………………………………………………………...……… Hypotheses testing ..…………………………..…………………………...……… Discussion of the results of Study 2..…….....……………………………...……… General Discussion..….………………………...……………………………….……… Theoretical Contributions..……..………………………..………………....……… Managerial Implications.………………………..……….………………...….…… Limitations and Future Research..……………………………………….....……… 5 8 12 12 13 15 17 19 22 24 24 25 25 27 30 31 31 33 40 43 43 44 48 48 54 57 58 60 62 viii ESSAY 2 NEW PRODUCT LAUNCHES AND MANUFACTURERS` RETURNS: UNDERSTANDING THE IMPACT OF NEW PRODUCT LAUNCH DECISION ON VALUE CLAIMED AND CREATED FROM NEW PRODUCTS……………………... Theoretical Background………………………………………………………………… The Proposed Model……….…………………………………………………………… Hypotheses….…………………………………………………………………………... Prior New Product Success..………………………………..………….................. The Innovativeness of Manufacturers` Products…………….................................. Frequency of New Product Introduction.…………..……………………….......... Degree of Selectivity…………..…………………………………………….......... Methodology……….………………………………………………………………........ Overview………………...…………………………………………….………….. Study 1: Field Survey…………………………………………………………………… Research Context and Data Collection………………………………………........ Measures…………………...…………………………………….….…………….. Control Variables ..………………...…………………...............................……… Analysis and Results..…………………………………….……………...……...……… Measurement Model Analysis..…………..……….…...…………………...……… Hypotheses testing..………………...…………….………………………...……… Discussion of the results for Study 1..……………………………………...……… Study 2: Experimental Design..…………..……………………………………...……… Sampling and Data Collection ..….………………………………………...……… Stimuli and Measures ..……………………………..……………………...……… Measurement Model Analysis……………………………………………...……… Hypotheses testing ..………………….………..…………………………...……… Discussion of the results of Study 2..…….....……………………………...……… General Discussion..…………………………….……………………………….……… Theoretical Contributions..……..………..………………..………………...……… Managerial Implications.…………….…….……..……….………………...……… Limitations and Future Research..………..…………….…………………...……… 65 67 71 74 75 78 80 83 85 85 86 86 87 91 93 93 95 101 104 105 106 108 109 118 120 121 123 125 APPENDICES.…………………………………...………………………………………. 127 REFERENCES…………………………………...……………………………………….. 142 ix LIST OF TABLES Table 1.1: Variance Covariance Matrix for Study 1………………………………..……. 27 Table 1.2: Results of the Interaction Effect Model for Study 1…………………...……... 36 Table 1.3: Means, Standard Deviations and Correlations for Study 2…………………… 47 Table 1.4: Results of Study 2……………………………………………………………... 49 Table 1.5: Post-hoc Analyses for Relational Behaviors……………………………….…. 52 Table 2.1: Variance Covariance Matrix for Study 1………………………………...…… 89 Table 2.2: Results of the Interaction Effect Model for Study 1………………………..… 97 Table 2.3: Means, Standard Deviations and Correlations for Study 2…………………… 107 Table 2.4: Results of Study 2…………………………………………….……………..… 109 Table 2.5: Post-hoc Analyses for Value Claimed and Value Created………….………… 113 Table 2.6: Results of Study 2…………………………………………….……………….. 117 x LIST OF FIGURES Figure 1.1A: Proposed Theoretical Model….…………………………..….……………. 14 Figure 1.1B: Proposed Theoretical Model….…………………………..….……………. 20 Figure 1.2: Graphical Interpretation of the Moderation Effects of the Frequency of New Product Introductions (Study 1)……..………….... 38 Figure 1.3: Graphical Interpretation of the Impact of New Product Success on Relational Behaviors Over Time (Study 2) ……………………………… 53 Figure 2.1: Proposed Theoretical Model ………………………………………………… 72 Figure 2.2: Graphical Interpretation of the Moderation Effects of the Innovativeness of Manufacturer`s Products (Study 1) ..……………… 99 Figure 2.3: Graphical Interpretation of the Moderation Effects of the Frequency of New Product Introductions (Study 1)……….…………. 100 Figure 2.4: Graphical Interpretation of the Impact of Prior New Product Success on Value Claimed and Created Over time (Study 2) ……………….......…… 114 xi INTRODUCTION In their continuing search for competitive advantage, manufacturers are increasingly turning to innovation and the development of new products. They do this in the hopes of generating increased sales and profits (Abetti 2000, Chandy and Tellis 2000). However, the value that innovation efforts generate for firms is limited when new products are launched through large retailers, because manufacturers must share a portion of the value with these retailers. This is a challenge, since manufacturers and retailers are rivals for value extraction. Manufacturers often complain that retailers are creative in finding unpredictable methods of extracting additional revenues (Iyer and Villas-Boas 2003) and that retailers gain higher shares of profits at their expense (Dukes et al. 2006). For example, “vendors complain – usually off the record – of an unceasing barrage of demands from powerful retailers that want everything from payment of fines for shipment errors and product labeling errors to a large number of free samples” (Iyer and Villas-Boas 2003). Additionally, manufacturers are increasingly dependent on their retailers. Since the concentration of retailing across several sectors limits the distribution channels available to manufacturers (Hultink 1999), retailers have begun to play an increasingly important role in the successful launch of new products. For example, the refusal of dominant retailers to carry certain products may block national distribution and negatively impact new product performance (Luo et al. 2007). Further, such refusal may result in limited access for consumers, who often shop first at large retail chains when considering a purchase. The increasing power and importance of retailers is demonstrated by the following statistics: 1 “The largest discount retailers are responsible for 80 per cent of the daily groceries sales in the United States – 10 years ago, it was only 30 per cent” (Thomassen et al. 2006) “In the United States, retail is the second largest industry in terms of both establishments and number of employees, generating approximately $3.8 trillion in sales (Thomassen et al. (2006) While the growing power of retailers creates challenges for manufacturers, research on innovation and new product launches is currently limited. It has not yet addressed distributionrelated issues and the difficulties that manufacturers face when launching new products. Understanding and addressing these challenges is important, because successfully launching new products is critical to manufacturers for maintaining market leadership. The development of strong distribution channels and the establishment of appropriate channel activities play a particularly critical role in any new product launch (DiBenedetto 1999). To address these issues, this two-essay dissertation will investigate the challenges that manufacturers face when launching new products through retailers. The first essay draws on Governance Value Analysis and examines two key research questions: how manufacturers govern the introduction of new products and how this governance impacts retailers’ behavior after new products are launched. The focus of this essay is on managing a single new product introduction. The second essay draws on the literature on reciprocity and examines value creation and value sharing problems between manufacturers and retailers, specifically, how manufacturers’ new product launch decisions impact manufacturers’ returns. The focus of this essay is on decision-making with respect to multiple new product introductions. Overall, the research presented here provides insights for marketing academics as well as marketing managers on how manufacturers should structure their relationships with retailers in order to 2 enhance value creation and value extraction from new product launches. The following is a more detailed overview of each of the two essays. Essay 1 investigates two key research questions: 1. How do manufacturers govern the introduction of a new product in a way that minimizes transaction costs and maximizes value? 2. How does this governance impact retailers’ relational behaviors toward the new product after it is launched? By utilizing Governance Value Analysis (Ghosh and John 1999, 2005) as the theoretical framework, this essay investigates the impact of the interaction between new product innovativeness and exchange attributes (market uncertainty, performance ambiguity, and frequency of new product introduction) on contract specificity and the implications of contract specificity on behaviors toward the new product after it is launched. Two studies are designed to test each research question. Study 1 is a field survey administered to manufacturers who launch their products through large retailers. Study 2 employs a longitudinal experiment administered to retailers in order to capture the retailers’ behaviors toward the newly launch product. Overall, this essay finds empirical support for the propositions put forth by Governance Value Analysis, both for the design of contracts and the predictions about the behavior of the parties in an exchange over time (Ghosh and John 1999). Essay 2 investigates how new product launch decisions and prior new product success impact manufacturers’ returns from subsequent new product introductions. By drawing on the literature on reciprocity as the causal mechanism explaining relationships among the constructs, this essay investigates how prior new product success interacts with product launch decisions (the innovativeness of manufacturer`s products, frequency of new product introduction, and degree of selectivity) in determining value claimed by the manufacturer and value created from new products. Of specific interest is the examination of how new product decisions and 3 performance impact retailers’ willingness to reciprocate via value claimed and value created. Two studies are employed to test the proposed hypotheses. Study 1 is a field survey administered to retail managers in the food and health and beauty product industries. Study 2 is a longitudinal experiment, whose goal is to replicate the findings of Study 1 and extend them by testing whether reciprocity is the underlying mechanism that explains hypothesized relationships. Overall, this essay empirically demonstrates that both manufacturers’ new product launch decisions and prior new product success are important in determining manufacturers’ returns. In addition, this essay demonstrates that the extent to which retailers reciprocate varies based on manufacturers’ new product launch decisions. 4 ESSAY 1 NEW PRODUCT LAUNCHES AND MANUFACTURERS` RETURNS: UNDERSTANDING GOVERNANCE DECISIONS BETWEEN MANUFACTURERS AND RETAILERS FOR NEW PRODUCTS Effective product launch is a critical driver of new product performance, but it is often the “most expensive, most risky and least well managed” factor (Calantone and Montoya-Weiss 1993). Today`s competitive marketplace makes the successful launch of a product increasingly important, but also increasingly difficult. This is particularly true when new products are launched through dominant retailers. On the one hand, firms must continuously innovate to avoid obsolescence of their product lines (Montgomery 1975). On the other hand, new product launches can be very costly. Apart from commercialization costs, manufacturers also incur additional costs associated with slotting fees (Desiraju 2001), difficulty in forecasting, pricing, and monitoring of new products due to limited access to consumers. In addition, manufacturers are removed from the actual point of sale and have either limited or delayed information about the new product`s performance. This increases their reliance on the retailer, slows response times, and creates substantial requirements for the coordination of various tasks, joint decision making, and information sharing. Therefore, establishment of appropriate distribution channel activities becomes of utmost importance. Proper governance mechanisms for the introduction of new products must be carefully devised to allow manufacturers to minimize their costs, maximize their value (Ghosh and John 1999), and ensure the necessary cooperation and information sharing between manufacturers and retailers on behalf of the new product after it is launched. An important characteristic of a distribution channel is contract design and negotiation of the terms of trade. Manufacturers must carefully devise their contractual terms to facilitate 5 desired outcomes. This includes governing the activities performed by manufacturers and retailers and dividing the generated value between the exchange partners. Unfortunately, our understanding of how firms should govern the introduction of new products is currently limited because the topic is under-researched. For example, the literature on innovation focuses on factors that impact new product profitability (Atuahene-Gima 1995, Wuyts et al. 2004), but this research is limited in addressing distribution related issues. Moreover, while both Transaction Cost Analysis and the literature on contracting offer considerable insights into designing contracts (Lusch and Brown 1996, Mooi and Ghosh 2010), they do not study contractual design for new products that face unique challenges (e.g., high risk, high uncertainty, the need for closer cooperation etc.). Alternatively, the economics literature empirically measures optimal coordination of marketing channels to maximize profit creation and profit sharing (Iyer and Villas-Boas 2003, Luo et al. 2007), but it does not address contractual issues or issues related to new products. Therefore, the goal of this essay is to close these gaps and empirically address two key research questions: 1. How do manufacturers govern the introduction of new products in a way that minimizes transaction costs and maximizes value? 2. How does this governance impact retailers’ relational behaviors toward the new product after it is launched? To answer these questions, this essay draws on Governance Value Analysis, according to which a three-way fit among firm resources, exchange attributes and governance determines success in creating and claiming value (Ghosh and John 1999). In fact, it is the trade-offs among these factors that are “the core insight offered by the model” (Ghosh and John 1999, p.131). What makes Governance Value Analysis particularly fitting for the study of new product launches is the addition of firm-specific considerations to the standard Transaction Cost Analysis. Each firm differs in its efforts to innovate, both in terms of how innovative the new 6 products are, and in terms of how frequently they are introduced. This provides manufacturers with a wide array of strategic options by which to govern their relationships with retailers. This essay will therefore examine the interaction of exchange attributes (i.e., market uncertainty, performance ambiguity, frequency of new product introduction) with a firm`s innovation efforts (e.g., degree of new product innovativeness) in determining governance decisions (e.g., contract specificity) in order to answer the first research question of how manufacturers govern the introduction of new products. The impact of the proper alignment of these three constructs to a theoretical prediction of Governance Value Analysis should, according to the theory, maximize the value claimed by the manufacturer and the value created from the new products (Ghosh and John 1999). Ghosh and John (1999) and the literature on contracting (Lusch and Brown 1996) also propose that governance impacts the behavior of the parties in an exchange. For example, if value claiming problems are not properly managed, activities associated with value creation may be affected negatively (Ghosh and John 1999). This essay therefore also examines the impact of contract specificity on retailers’ relational behaviors toward the new product in order to answer the second research question of how governance impacts retailers’ behaviors toward a new product after it is launched. Since the behavior of the retailers during the contractual period cannot be divorced from the way the new product performs, the impact of new product success on the development of relational behaviors is also examined. The object of this essay is to contribute to the literature in three ways. The first contribution is to Governance Value Analysis. Although advances have been made in the development and testing of Governance Value Analysis, its empirical support remains limited. This essay extends the theory by incorporating a firm`s innovation effort (as a firm-specific 7 resource) and testing its interaction with exchange attributes in impacting contract specificity. The second contribution is to provide guidance to managers on what strategic options should be pursued for new products and how new product introduction should be governed to maximize the manufacturer`s value. The third contribution is to advance to our understanding of how governance mechanisms impact relational behaviors after new products are launched. Since new products require close coordination and cooperation, joint decision making, and information sharing between manufacturers and retailers, it is important to understand how governance impacts the engagement of retailers in product-related relational behaviors during the contractually specified period. THEORETICAL BACKGROUND When manufacturers introduce new products through major retailers, they need to set up governance mechanisms in ways that support the new product launch. An important governance mechanism is contract design, which governs activities performed between manufacturers and retailers and facilitates exchange (Willimason 1979). To understand the influence that new products have on contract design, this essay draws on Governance Value Analysis, an extension of Transaction Cost Analysis. Transaction Cost Analysis provides a theoretical lens for understanding how interfirm relationships should be organized. Although the original framework focused on governance forms that examine discrete choices between market and hierarchical governances (Rindfleisch and Heide 1997, Williamson 1965), more recent developments have lead to the inclusion of a variety of hybrid mechanisms. While research on governance forms is used to help manage problems that may occur in long-term relationships (Rindfleisch and Heide 1997), research on governance mechanisms addresses more short-term relationship management issues. Since the 8 focus of this essay is on governance mechanisms, a more detailed overview of the literature on governance mechanisms will now be provided. A variety of mechanisms have previously been indentified. Heide and John (1988) propose offsetting investments, Anderson and Weitz (1992) suggest crafting of incentive structures through the use of pledges, and Stump and Heide (1996) focus on monitoring. A large body of Transaction Cost Analysis research examines contracting issues that are important in investigating bilateral relationships. For example, in one of the early studies, Joskow (1987) examines the relationship between the duration of coal contracts and relationship-specific investments. He finds that as relationship-specific investments become more important, the parties rely on longer-term contracts that specify the terms of trade. In another influential study, Lusch and Brown (1996) examine the antecedents and performance consequences of explicit and normative contracts. They find that the performance of marketing activities can be coordinated through such contracts. According to Transaction Cost Analysis, contracts should allow for the organization of transactions in a way that economizes production expenses and transaction costs (Willimason 1965). Transaction costs can be divided into three categories: safeguarding, measurement and adaptation (Rindfleisch and Heide 1997). Safeguarding costs occur as a result of partner`s opportunism, after specific investments in the relationship have been made (John and Weitz 1988). Measurement or performance evaluation costs occur as a result of performance ambiguity and represent expenses associated with assessing contractual compliance. In other words, it is difficult for firms to determine adherence to contractual agreements (John and Weitz 1988). Adaptation costs result from environmental uncertainty and represent the difficulty in modifying 9 agreements due to changing circumstances. When circumstances change, firms need to adjust and renegotiate contracts, which can be costly (Grover and Malhotra 2003). These exchange attributes influence the way contracts are designed and alter both ex ante transaction costs of negotiating and crafting contracts and ex post transaction costs of enforcing and monitoring agreements (Mooi and Ghosh 2010). Ex post costs are costs that arise in the execution and implementation stages and they are associated with keeping contract terms open. Leaving terms of trade open increases the danger of misunderstanding due to a lack of clearly defined roles and it increases the risk of opportunistic renegotiations (Wathne and Heide 2000). Alternatively, ex ante costs are associated with keeping contractual terms specific. They include managerial time and effort in projecting future scenario and the costs associated with the search for information and the negotiation of mutually acceptable solutions (Mooi and Ghosh 2010). While less specific contracts permit greater flexibility and opportunity to adapt, specific contracts are more difficult to renegotiate because positions are stated more clearly (Ghosh and John 2005). Therefore, the exchange attributes influence these costs in the following way: Performance ambiguity increases the ex post costs of monitoring and enforcing of agreements (Williamson 1996) because it is difficult to assess contractual compliance. Environmental uncertainty raises the ex ante costs of drafting and negotiating contracts because foreseeing future contingencies becomes more difficult. Governance Value Analysis (Ghosh and John 1999) is a recent extension of Transaction Cost Analysis. It has been proposed to address some of the weaknesses in Transaction Cost Analysis, such as its limited application to strategic marketing choices. Ghosh and John (1999) expand Transaction Cost Analysis to address marketing strategy decisions more closely. By doing so, the authors emphasize both value maximization and cost minimization and incorporate 10 firm heterogeneity into the model by adding firm-specific considerations to the standard Transaction Cost Analysis. The basic structure of Governance Value Analysis consists of four core constructs (Ghosh and John 1999). Two are from Transaction Cost Analysis and comprise the attributes of exchange (e.g. asset specificity, market uncertainty, and performance ambiguity) and governance forms (e.g. market, hierarchies and relational). Two additional constructs specific to each firm are added. These include positioning and resources. Ghosh and John (1999) define positioning as “the particular bundle of benefits selected by the firm to be created and delivered to the target customer.” Resources are defined as scarce and imperfectly mobile skills, or assets that are owned by the firm. The authors categorize these resources into end-customer (e.g. brand equity, customer loyalty, switching costs etc), supply chain (channel), and technological (unique equipment, processes and patents). Technological resources, particularly a firm`s innovation efforts, are at the focus of this essay. Technological resources provide firms with value through the uniqueness and scarcity of the product. The core thesis of Governance Value Analysis is that these four basic constructs influence one another. That is, positioning must be matched with resources, exchange attributes and governance forms. When this is done in accordance with the theory, this leads to minimization of transaction costs and therefore to value maximization. Since the constructs operate at varying levels within the firm, Governance Value Analysis is argued to be a mixed-level model. Building on Ghosh and John (1999), Ghosh and John (2005) empirically test the Governance Value Analysis by examining a three-way fit among firm resources, investments, and governance. The authors demonstrate that firms` resources matter in the way governance is established. Kim et al. (2011) also apply Governance Value Analysis in partially integrated 11 channels and examine influence of three extra-dyadic effects (brand reputation, sales force feedback, and retailer feedback) on ongoing governance decisions. Even though some studies have empirically tested the Governance Value Analysis, further empirical research is needed. Governance Value Analysis also makes predictions about the behavior of the parties in an exchange in response to established governance. Specifically, Ghosh and John (1999) highlight that “unless governance is devised to manage the value claiming problem, value creation is affected negatively” (p.133). For example, the authors state that the partners in an exchange will minimize their ex post disadvantage by “(1) scaling back investment, (2) adapting less, and (3) foregoing activities that are hazardous from a measurement standpoint” (Ghosh and John 1999, p. 133). In this sense, the authors make predictions about how the behaviors between two partners in an exchange will differ based on specific governances. The impact of governance on behaviors was similarly tested by Lusch and Brown (1996), who argued that channel contracting (the way contracts are set up) impacts relational behaviors in an exchange relationship. The authors, however, found no relationship between explicit contracts and relational behaviors. HYPOTHESES Research Question 1: How do manufacturers govern the introduction of new products in a way that minimizes transaction costs and maximizes value? New product Innovativeness. Figure 1.1A depicts the proposed model that addresses the first research question. The interplay between ex ante and ex post transaction costs plays a key role in the choice of the most efficient level of contract specificity to govern the introduction of new products. Therefore, contract specificity is hypothesized to be contingent on new product innovativeness and market uncertainty. When new product innovativeness is low, the 12 uncertainty associated with new product introduction is also low. In addition, the potential for financial returns is limited and often short-lived. As a result, ex ante as well as ex post contractual costs to manufacturers are low. As new product innovativeness increases, both ex ante and ex post costs to manufacturers increase. Low familiarity with technologies and/or markets, difficulty in specifying circumstances of exchange beforehand, and high opportunity costs of maladaptation (Abetti 2000) all raise ex ante costs and favor low contract specificity. However, as product innovativeness increases, ex post hazards also increase. Order revision and maintenance of flexibility, for example, can be difficult and costly for highly innovative products that often require new production processes or new materials. Additionally, the threat of opportunism from the retailer over the generated margin stream can be substantial if the product succeeds. This is because highly innovative products have a great potential for financial returns and profitability (Sorescu 2003, Wuyts et al. 2004), creating a greater margin stream over which the retailer can bargain opportunistically. In summary, since an increase in new product innovativeness is associated with increases in both ex ate and ex post costs, manufacturers must balance the additional benefits of crafting less specific contracts with expected ex post costs (Ghosh and John 2005). Market Uncertainty. When market uncertainty is high, it becomes difficult to predict changing customer needs and preferences accurately (Wathne and Heide 2004). Complex, littleknown, turbulent circumstances make forecasting and predictions about the future difficult and costly (Anderson 1985). “This forces firms to change previously planned courses of action and decisions involving existing assets and/or abandon previous investments in favor of striking out in new directions” (Ghosh and John 1999, p. 134). The inability to predict contingencies and the 13 need for greater adaptation creates problems in writing contracts (Williamson 1996) and raises ex ante costs. A manufacturer introducing a new product in an uncertain market needs to take the above-mentioned factors into consideration when negotiating contracts. Figure 1.1A: Proposed Theoretical Model Market Uncertainty Contract Specificity New Product Innovativeness Performance Ambiguity Frequency of New Product Introduction When the new product innovativeness is low, the opportunity cost of maladaptation is also low because the benefits are shorter-lived and because such products are less differentiated than highly innovative products. Therefore, a manufacturer introducing a product with a low degree of innovativeness in an uncertain market should negotiate specific contracts to minimize costs associated with making adjustments and to protect the generated margin stream from exploitation by the retailer. As the degree of product innovativeness increases, the opportunity costs of maladaptation also increase and manufacturers should craft increasingly less specific contracts in order to be able to adapt to changing markets. Keeping contractual terms open minimizes costs associated with inaccurate forecasting and contract renegotiation. This provides manufacturers with the necessary flexibility to respond to actual product performance by quickly 14 changing orders, pricing, inventory, or product characteristics. This approach is consistent with what the literature terms lean launch, where manufacturers have a limited commitment of inventory during the introductory stages and a flexible logistic system. This allows them to respond rapidly to the new product`s performance (Bowersox et al. 1999). When market uncertainty is low, manufacturers better understand customers` needs and preferences and so the accuracy of forecasting is much higher than when uncertainty is high (Wathne and Heide 2004). This decreases ex ante costs. As a result, as new product innovativeness increases under low market uncertainty, manufacturers should negotiate more specific contracts and engage in what the literature terms an anticipatory launch (Bowersox et al. 1999). The goal of an anticipatory launch is to decrease uncertainties associated with the introduction of new products through research that includes demand forecasting, preliminary market testing, development of experimental test markets, and engagement of focus groups. This allows manufacturers to negotiate specific contracts and eases demands on production and inventory planning. It also protects the margin stream generated by the introduction of a highly innovative product. Therefore: H1: All else being equal, when market uncertainty is high, the negative impact of new product`s innovativeness on contract specificity is greater than when market uncertainty is low (absolute value). Performance Ambiguity. The relationship between new product innovativeness and contract specificity is also hypothesized to be contingent on performance ambiguity. Under conditions of high performance ambiguity, it is difficult to assess the retail performance of new products launched through a specific retailer. As a result, the degree of the retailer`s contribution to profit generated from the sales of a specific product is not easily verifiable ex post (Ghosh and John 1999). Under these conditions, ex post hazards increase, since the value generated from 15 new products becomes open to exploitation. Specifically, increasing the new product innovativeness under conditions of high performance ambiguity has the following effects. First, as the degree of product innovativeness increases, the negative consequences potentially associated with high performance ambiguity become more severe. For example, proper retailer support becomes more important as new product innovativeness increases. Consumers are not yet familiar with the product and they need to be made aware of its existence and its benefits. Additionally, since the manufacturer`s knowledge and understanding of the market and of consumers is limited for highly innovative products, it becomes increasingly important for them to be able to observe the new product’s retail performance. When the retail price or marketing mix activity cannot be observed, the manufacturer cannot determine what exact retail price and marketing efforts produced the realized demand (Iyer and Villas-Boas 2003). Therefore, if the retailer does not provide the expected level of implementation and support for the new product, or if the retailer`s behavior cannot be easily observed, the costs to manufacturers may be severe. Second, should the new product succeed, the generated margin stream may be substantial for highly innovative new products because they have a great potential for financial returns (Sorescu 2003) and a greater effect on profitability than less innovative new products (Wuyts et al. 2004). This creates a greater margin stream, which tends to attract opportunistic bargaining on the part of the retailer. For these reasons, as new product innovativeness increases under conditions of high performance ambiguity, manufacturers should craft more specific contracts that explicitly state how various future situations will be handled (Lusch and Brown 1996). The establishment of formal rules and procedures to govern such relationships reduces behavioral 16 uncertainty and discourages opportunism, since relationships in formal contracting regimes have been found to be less vulnerable to ambiguity (Carson et al. 2006, Jap and Ganesan 2000). Alternatively, when performance ambiguity is low, it is likely that the value generated from the exchange relationship will be equitably split, because assessment of the retail performance of new products is directly observable. This decreases ex post hazards (Ghosh and John 1999). Low performance ambiguity can be achieved with retailers who use various computer programs, such as Electronic Data Interchange, that enable easy transmission of data between manufacturers and retailers. This allows manufacturers to monitor new product performance more accurately. Therefore, when performance ambiguity is low, manufacturers should negotiate less specific contracts as new product innovativeness increases, in order to minimize ex ante costs and engage in ex post value enhancing adjustments (Mooi and Ghosh 2010). H2: All else being equal, when performance ambiguity is low, the negative impact of a new product`s innovativeness on contract specificity is greater than when performance ambiguity is high (absolute value). Frequency of new product introduction. The frequency of new product introduction is also expected to alter the ex ante and ex post contracting costs and is hypothesized to interact with new product innovativeness. When the frequency of new product introduction is high, manufacturers maintain a large variety of products on the shelves, continuously calibrating their product offerings to consumer needs. As a result, the complexity of the relationship between the manufacturer and the retailer increases, creating a greater need for the coordination of various tasks and for joint decision making. This, in turn, increases the transaction costs associated with frequent and costly contract adaptations, difficulty in reaching agreements (Mooi and Ghosh 17 2010), and difficulty in foreseeing future events, as the amount of information that needs to be processed increases (Grover and Malhotra 2003). Additionally, the introduction of each new product generates new value, but it also alters the value of existing products on the market. When new products are introduced frequently, increasing new product innovativeness has several effects. First, the introduction of highly innovative new products is a high-risk strategy (since they frequently fail), but it also has potentially high returns if they should succeed (Abetti 2000, Sorescu 2003). Since highly innovative new products address new markets, new consumer segments, or previously unrecognized needs, manufacturers need to respond quickly to consumers and modify or adjust new products appropriately. If necessary, manufacturers should also be able to expand their product line quickly to capitalize on the emerging market opportunity. Frequent new product introductions allow manufacturers to better calibrate their product offerings to markets and consumers, enhancing the performance of the highly innovative product and generating greater value. This increases the margin stream and raises ex post hazards. To decrease ex post hazards, manufacturers should craft specific contracts when new product innovativeness is high. When new product innovativeness is low, the risks associated with new product launch are also low. Additionally, the benefits of less innovative products are shorter-lived and their margin streams are more limited. When new products are introduced frequently, with the introduction of each new product either (a) the value of existing products decreases, should both an existing and new product be on the shelves; or (b) the margin stream from existing products is forgone, should the manufacturer be forced to replace the existing product with the new one. Therefore, a high frequency of new product introduction further limits the margin stream that can be generated from less innovative new products, limiting ex post hazards. Since the ex post 18 hazards are limited, manufacturers should craft non-specific contracts to maintain flexibility and to minimize the costs of contract renegotiations. Therefore, for the reasons just given, under the condition of high frequency of new product introduction, as the degree of innovativeness increases, contract specificity should also increase. The relationship between new product innovativeness and contract specificity is hypothesized to be the opposite when new products are introduced less frequently. Under those conditions, when new product innovativeness is low, both ex ante and ex post costs are also low, because market demand is relatively predictable and the revenue stream from products with a low degree of innovativeness is limited (Sorescu 2003). This favors high contract specificity, since specific contracts are less costly to craft and the need for contract renegotiation is low (Mooi and Ghosh 2010). As new product innovativeness increases, familiarity with markets and/or technologies decreases (Abetti 2000) and the circumstances of exchange cannot be easily specified beforehand. Since the frequency of new product introduction is also low, the value generated from the exchange is more likely to be limited, thus favoring low contract specificity. Therefore: H3: All else being equal, the greater the new product`s innovativeness, the greater (lower) the contract specificity when frequency of innovation is high (low) Research Question 2: How does governance impact a retailer`s behavior towards a new product after it is launched? Contract Specificity. Figure 1.1B depicts the proposed model addressing the second research question of how governance impacts retailers’ relational behaviors. As identified in prior literature, the relational behaviors include solidarity, flexibility, and information exchange (Hoppner and Griffith 2011, Lusch and Brown 1996). Solidarity is defined as the extent to which the retailer works jointly with the manufacturer for the benefit of the new product. Flexibility is 19 defined as the extent to which the retailer works around the contract for the benefit of the new product. Information exchange is defined as the extent to which the retailer proactively provides useful information about the new product to the manufacturer. Figure 1.1B: Proposed Theoretical Model Contract Specificity Relational Behaviors New Product Success Prior research has yielded inconclusive results regarding the relationship between contractual design and relational behaviors. Some studies find that the negotiation of specific contracts allows the exchange partners to state explicitly how they would handle various situations that might occur in the future (Mooi and Ghosh 2001) and to define appropriate and inappropriate behaviors in the relationship (Lusch and Brown 1996). Therefore, specific contracts should positively impact relational behaviors to the extent that they are included in the contractual agreement (Lusch and Brown 1996), and promote more cooperative, long-term, trusting exchange relationships (Poppo and Zenger 2002). Other studies, however, find that specific contracts produce more conflict, undermine trust, and encourage opportunistic behaviors, thus negatively impacting engagement in relational behaviors (Young and Wilkinson 1989). Since the prior literature is inconclusive, this essay draws on the propositions of Ghosh and John (1999) to explain the hypothesized effects. 20 These authors argue that governance impacts the behavior of actors in an exchange relationship. They argue that when value claiming is not properly managed, the exchange partners will try to minimize their ex post disadvantage by “(1) scaling back investment, (2) adapting less, and (3) forgoing activities that are hazardous from a measurement standpoint” (Ghosh and John 1999, p. 133). Since engagement in product-related relational behaviors (solidarity, flexibility, information exchange) requires retailers to adapt more and to invest in the new product (activities that are both hazardous from a measurement standpoint), the extent to which contractual terms are kept open or made specific is hypothesized to determine the extent of ex post disadvantage and therefore to influence retailers’ engagement in relational behaviors. The reasoning is as follows. Contractual designs influence the cost of running the system. Ex ante and ex post costs of monitoring and enforcement change with varying levels of contract specificity in the following way. When contractual terms are kept open, ex post costs in the execution and implementation stages rise (Mooi and Ghosh 2010), but when contractual terms are made specific, ex ante costs rise while ex post costs are lower due to more clearly defined roles and responsibilities. This decreases the danger of opportunistic renegotiations (Wathne and Heide 2000) and protects the generated margin stream (Ghosh and John 2005). When specific contracts are established and ex post costs are minimized, retailers are hypothesized to be more likely to engage in productrelated relational behaviors. This is because specific contracts shield the exchange partners from risk (Poppo and Zenger 2002) and protect the generated margin stream from opportunistic renegotiations (Wathne and Heide 2000). This encourages retailers to engage in value-creating activities on the behalf of the new product and so increases their engagement in relational behaviors. 21 Alternatively, when the contractual terms are kept open, ex post costs increase because the exchange partners can bargain opportunistically over the generated margin stream. According to Governance Value Analysis (Ghosh and John 1999), since the ability to claim value is uncertain, retailers will scale back activities that are hazardous from a measurement standpoint, including engagement in relational behaviors. Therefore: H4: When contract specificity is high, engagement in relational behaviors will be higher than when contract specificity is low. New Product Success. Since the behavior of retailers during the contractual period cannot be divorced from the way the new product performs, the impact of new product success on the development of relational behaviors over time is also examined. Prior literature argues that relational behaviors develop as a result of prolonged dependence and cooperative planning (Dwyer et al. 1987) and that relational behaviors transpire over time. Each transaction must therefore be treated in terms of its history and anticipated future (Lusch and Brown 1996). When the new product is launched, its future performance is uncertain and the data available on past performance is either nonexistent or significantly limited. As a result, the impact of new product success on product-related relational behaviors is expected to change over time, as the uncertainty regarding the new product`s performance is resolved. When the success of the new product is over the contractual period is high, engagement in relational behaviors is hypothesized to increase over the time. The reasoning is as follows. First, since investment in relational behaviors is costly in terms of time and resource allocation (Larson 1992), retailers will increase their engagement in relational behaviors only after the uncertainty regarding a new product`s future performance decreases. Second, as the expectation of potential future pay-offs appears more likely to be fulfilled over time, the level of cooperation 22 and the degree of engagement in relational behaviors should increase (Poppo and Zenger 2002, Lusch and Brown 1996). Therefore: H5a: In the condition of a high new product success, engagement in product related relational behaviors will increase over the duration of the contractual agreement When the success of the new product over the contractual period is low, the retailer must decide whether to engage in product-related relational behaviors to try to improve the new product`s performance. It is hypothesized that retailers will initially engage in relational behaviors but that this engagement will decrease over time. This is expected to occur for two reasons. First, new product launch is costly (Ogawa and Piller 2006). If the new product fails, the retailer may lose money. This motivates the retailer to engage in product-related relational behaviors and to cooperate closely with the manufacturer in an effort to minimize losses and improve the new product`s performance. This behavior is based on a calculative process of minimizing costs and maximizing returns (Bercovitz et al. 2006). Second, the future performance of the new product is initially uncertain. Even though there may be a period of poor performance, many new products may eventually succeed (Bass 1969). Over time, however, this uncertainty is resolved. Since relational behaviors develop over time (Lusch and Brown 1996), if a new product continues to perform poorly and the expectation of future returns is limited, retailers are expected to decrease their engagement in relational behaviors. H5b: In the condition of a low new product success, engagement in product related relational behaviors will decrease over the duration of the contractual agreement When the success of a new product is low over a sustained period but it finally improves over time, engagement in product-related relational behaviors is expected first to decrease and then to increase. As argued above, the retailer will be initially motivated to engage in relational 23 behaviors in response to a poor product performance and closely cooperate with the manufacturer in an effort to minimize losses and improve the new product`s performance. However, if the new product continues to perform poorly, engagement in product-related relational behaviors will decrease because relational behaviors develop in the context of past history and anticipated future (Lusch and Brown 1996). But if the new product begins to perform well after a period of poor performance, the retailers will again increase their engagement in relational behaviors because of potential future returns. This is because the expectation of future returns increases the level of cooperation in the present (Poppo and Zenger 202). Therefore: H5c: In the condition when initial low new product success is eventually followed by high success, engagement in product-related relational behaviors will first decrease and then increase over the duration of the contractual agreement METHODOLOGY Overview Two studies are conducted to test the proposed hypotheses. The goal of Study 1 is to address the first research question of how manufacturers govern the introduction of new products. This study tests the moderating role of market uncertainty, performance ambiguity, and frequency of new product introduction on the relationship between new product innovativeness and contract specificity. It is a field survey administered to manufacturers who launch new products through large retailers. To analyze the results of Study 1, structural equation modeling (SEM) is used to examine the series of simultaneous relationships among the key constructs. The goal of Study 2 is to address the second research question of how governance impacts retailers’ behavior toward new products after they are launched. This study evaluates the impact of contract specificity and new product success on product-related relational behaviors 24 over the duration of the contract. This study is an experiment administered to retailers. Since the goal of Study 2 is to test how product-related relational behaviors develop and change over time, longitudinal experimental design is applied. The use of an experimental design increases the internal validity of the findings, allows for the isolation of hypothesized effects, and provides evidence of internal validity and causality of the hypotheses. To analyze the results of Study 2, a repeated measures ANCOVA model is used. Study 1: Field Survey Research Context and Data Collection To test the hypotheses, Study 1 examines how the manufacturers of consumer packaged goods (CPGs) launch new products through large retailers. CPGs were selected for several reasons. First, they represent a substantial portion of the U.S. economy, but they are largely underrepresented in empirical research (Sorescu and Spanjol 2008). Second, CPG manufacturers engage in frequent innovations, offering a large array of diversified products. This provides an appropriate context for studying new product launches since the diversity associated with the new products helps minimize any category-specific effects. Finally, CPG manufacturers have a limited ability to vertically integrate even though the transaction costs are substantial. This is because maintaining a high degree of control over product-related processes requires substantial direct out-of –pocket expenses. Proper management of governance mechanisms therefore becomes crucial. A market research company was used to administer online survey questionnaires to respondents who were part of its proprietary online panel. A random sample of 1290 qualified respondents was selected from this panel of potential respondents. To enhance the response rate, the respondents were compensated by the market research firm for participation in this study. 25 Follow- up emails containing a second survey were sent to non-respondents. In total, 217 questionnaires were received back, a response rate of 16.8%. After careful examination of the returns, 82 responses were excluded due to poor quality of responses or a large amount of missing data on key variables. The final sample consisted of 135 completed and usable questionnaires. The respondents held various positions within their organizations. They included account and category managers, directors, sales managers, and brand managers. They also represented various functions including sales (39.1%), marketing (17.8%), and innovation (8.6%). To ensure the appropriateness of the respondents, the potential participants were screened based on their involvement in the process of getting new products into selected retail stores and based on whether they were knowledgeable about the contractual terms negotiated for new products between the manufacturer and the retailer. Participants who fit all of the screening criteria were allowed to proceed to the survey, where they were asked to think of a new product that was recently launched and a major retailer through which this new product was launched. They were also asked to identify a new product and a retailer with which they were personally involved. Nonresponse bias was assessed using Armstrong and Overton`s (1977) procedure by comparing early and late respondents in terms of demographic variables and key study constructs. The results indicate that nonresponse bias is minimal because no significant differences were found on any of the items used in the study. The final sample represents manufacturers from multiple CPG industries. The top product categories represented in the sample include food (13.3%), health and beauty (12.6%), and sporting goods (5.9%). The median sales of the manufacturers are $50 million; 83.7% of firms generated sales over $1 million. The median number of employees is 300. The respondents 26 selected retailers with whom they have been doing business for an average of 16 years; on average, 23% of the selected category business went to these retailers. The selected new products had been on the market for an average of 11 months. Measures The key constructs in Study 1 are operationalized using multi-item reflective scales. Appendices 1 and 2 report the scales for the constructs and control variables. Table 1.1 reports the Variance Covariance Matrix. Table 1.1: Variance Covariance Matrix for Study 1 Construct 1. New Product Innovative ness 2. Market Uncertainty 3. Performance Ambiguity 4. Frequency of New Product Introduction (ln) 5. Contract Specificity 6. Advertising 7. Transaction Size (ln) 8. Relationship Length (ln) 9. Contract Duration 1 0.44 2 3 4 6 7 -0.10 0.69 -0.12 0.21 0.83 0.11 -0.12 -0.00 3.24 0.11 -0.09 -0.19 0.29 0.46 0.12 0.36 0.00 0.02 -0.02 -0.29 0.18 0.43 0.18 0.30 1.01 0.10 9.15 0.06 -0.06 -0.10 0.31 0.08 0.16 0.56 1.07 1.35 -2.20 -1.80 -2.22 2.22 -0.96 -0.08 0.23 27 5 8 9 111.8 New product innovativeness is defined as the extent to which the new product differs from competing alternatives in a way that is meaningful to customers (Fang 2008). New products are defined as products that, when introduced to the retailer, require a new stock-keeping unit. New products have varying degrees of innovativeness that can range from incremental to breakthrough. Incremental innovations involve minor changes in technology and offer minor improvements over existing products (Chandy and Tellis 1998); they include simple product improvements and alterations (Zhou et al. 2005). Breakthrough innovations include new products that are distinct from competitor`s products and offer new technologies, unique features (Calantone et al. 2006), and distinct benefits to consumers (Atuahene-Gima 1995). The measure of new product innovativeness uses a seven-item, five-point semantic differential scale, adapted from Fang (2008). The items ask respondents whether the new product is novel and offers new ideas relative to other products in the same category. Frequency of new product introduction is defined as the number of new products introduced by the manufacturer through a specific retailer in a related product category. It is important to note that frequency of new product introduction refers only to products introduced through a specific retailer. This characteristic is important because manufacturers may frequently introduce new products, but may choose to introduce only a few through certain retailers. Since different contracts are crafted with different retailers, it is the frequency of new product introduction specific to the retailer that will impact contract specificity. Additionally, frequency of new product introduction refer to new products launched in a related product category. Whenever a new product is introduced in a specific category, the sales, price, or marketing of existing products in that same category is likely to be altered. This introduces uncertainties that farsighted manufacturers consider when negotiating contracts. Three open-ended items were 28 developed to operationalize the frequency of new product introduction. These include the number of new SKUs created, the total number of new products launched yearly, and the average number of new products introduced yearly by the manufacturer through the specific retailer. Uncertainty. In accordance with Transaction Cost Analysis, manufacturers who sell both existing and new products through retailers face two types of uncertainties: market uncertainty and performance ambiguity. Market uncertainty is defined as the difficulty of making accurate predictions about the market for the new product (Celly and Frazier 1996). This creates unpredictable sales environments (Anderson 1985). In uncertain markets, the circumstances of exchange cannot be easily specified ex ante, and this gives rise to adaptation problems. Manufacturers who face uncertain markets experience difficulties in terms of product design and volume requirements, creating an ongoing need for flexibility and revision of coordinated action (Wathne and Heide 2004). While this increases the transaction costs associated with renegotiation, failure to adapt may result in a lost opportunity (Rindfleisch and Heide 1997). Market uncertainty is operationalized by asking respondents to describe their expectations about the market for the new product in terms of effectiveness and accuracy of selling efforts, sales forecasts, and marketing actions. This measure, adapted from Celly and Frazier (1996), uses a three-item, fivepoint semantic differential scale. Performance ambiguity is defined as a difficulty in assessing the retail performance of new products launched through a specific retailer. If the retailer`s true level of performance cannot be ascertained (Stump and Heide 1996), the manufacturer`s ability to measure the benefits and costs of the retailer`s contributions is limited (Ghosh and John 2005). The ability to assess retail performance is particularly germane in the context of new products, which provide 29 end-product enhancements. These are far less measurable than cost-reduction efforts (Ghosh and John 1999). The potential inability to distribute margin streams equitably increases the danger of opportunistic behavior on the part of the retailer. Building on prior literature (Kim et al. 2011), a new measure of performance ambiguity is developed in this study using a three-item, five-point semantic differential scale. Contract specificity. Contract specificity is a governance mechanism that encompasses “the initiation, termination, and ongoing relationship maintenance between a set of parties” (Heide 1994). In other words, it is a mode of organizing transactions. Contract specificity is defined as the degree to which contract terms are specified in detail ex ante (Mooi and Ghosh 2010). Low contract specificity means that explicit, formal terms are left open for possible modification through subsequent negotiations. Less specific contracts permit greater opportunity for ex post appropriation, while more specific contracts are more difficult to renegotiate because positions are stated more clearly (Ghosh and John 2005). The measure for contract specificity uses a three-item, five-point Likert scale. Control Variables Numerous factors apart from the model may influence contract specificity. As a result, four control variables are included: transaction size, the length of the relationship between the manufacturer and retailer, contract duration, and advertising. (The measures for control variables are included in Appendix 1.2) Transaction size is measured as the initial monthly purchase for a new product. As the size of transactions increases, manufacturers draft more specific contracts, because the hazards they face increase with increasing transaction size (Heide 1994). 30 The length of the relationship influences contract specificity because it may lead to a development of trust between the parties that decreases the need for safeguarding (Mooi and Ghosh 2010), resulting in lower contract specificity. Contract duration refers to the length of time for which two parties agree ex ante to abide by the terms of the contract (Joskow 1987). Contracts that have shorter duration are more likely to be more specific than contracts crafted for longer time periods, because the shorter time frame covered by the contract decreases the need for adaptation. Finally, advertising support for new products is included because it plays a very important role in new product launch and success. A manufacturer who provides substantial advertising support for its new products is more likely to craft specific contracts for two reasons. First, advertising has been shown to allow products to be sold at a higher prices and lower retail margins (Ailawadi and Harlam 2004). Therefore, manufacturers are more likely to craft specific contracts to protect themselves from exploitation in the value claiming stage. Second, advertising can play a significant role in the success of the new product, giving manufacturers leverage over retailers. Again, to protect the margin stream from ex post exploitation, manufacturers are more likely to craft specific contracts. Advertising is measured using a two-item, five-point Likert scale. ANALYSIS AND RESULTS Measurement Model Analysis The measurement model was estimated using confirmatory factor analysis with EQS 6.1. The measurement model consisted of the reflective multi-item latent constructs of new product innovativeness, market uncertainty, performance ambiguity, frequency of new product 31 introduction, and contract specificity. Since frequency of new product introduction was nonnormally distributed with a long right tail, the natural logarithm was taken and used in further analysis. Appendix 1.1 represents the results of the measurement model analysis, together with item loadings, composite reliabilities and average variance extracted (AVE). The overall chi-square goodness-of-fit index for the model is 222.149 based on 144 degrees of freedom. The measurement fit indexes for the confirmatory measurement models all meet the critical values for a model of good fit (Hu and Bentler 1999): the comparative fit index (CFI) was 0.977, the root mean square error of approximation was (RMSEA) 0.064, and the standardized root mean square residual (SRMR) was 0.055. In support of convergent validity, all factor loadings are large (ranging from 0.59 to 0.99) and significant (t-value > 2.00). To test discriminant validity, interconstruct correlations, which should significantly depart from 1.0 (Bagozzi et al. 1991), were examined. All correlations are significantly smaller than 1.0. Additionally, for all variables, the AVEs are larger than the squared correlations, therefore adequately confirming discriminant validity. Finally, the composite reliabilities (reported in Appendix 1.1) of all constructs range from 0.80 to 0.93, indicating acceptable levels of reliability for each construct. Since both independent and dependent measures were obtained from the same source, they are susceptible to common method bias. Three separate tests were conducted to determine the presence of common method bias. First, Harmon`s one-factor test (McFarlin and Sweeney 1992, Sanchez and Brock 1996) was conducted. This test entails entering all of the items of latent variables into a single factor using confirmatory factor analysis. The fit statistics for this model were quite poor (chi-square goodness-of-fit index of 876.35 with 104 degrees of freedom; CFI 0.383, RMSEA 0.235, and 32 SRMR 0.188), indicating that common method bias is minimal. Second, Lindell and Whitney`s (2001) marker variable assessment technique was employed. A variable (i.e., marker) was identified beyond the scope of the study, assessing its smallest correlation coefficient with theoretical predictors. The marker variable selected was the number of years that the respondents had been working in their current positions. Next, this variable`s coefficient was partialled out from the bivariate correlations. The partialled results were then compared against unadjusted correlations. After partialling out the number of years that respondents had been working in their current positions, all of the significant bivariate correlations among key predictors and outcomes maintained their statistical significance. Third, correlations between endogenous and exogenous errors were also examined. Collectively, the results suggest that the risk of common method bias is minimal. Hypotheses testing Structural equation modeling was used to test the conceptual model, which estimated contract specificity as a function of new product innovativeness; market uncertainty; performance ambiguity; frequency of new product introduction; interactions between new product innovativeness and (a) market uncertainty, (b) performance ambiguity, and (c) frequency of new product introduction; and the control variables. Since two control variables (transaction size and relationship length) were non-normally distributed with a long right tail, the natural logarithm was taken and used in further analysis. Additionally, the dataset contained missing values on transaction size (14.8%) and relationship length (2%). The missing values for transaction size were imputed using an EM method because of the high level of missing data, a small sample size that excludes the possibility of listwise deletion, and a nonrandom missing data pattern. The missing values for relationship length were mean imputed. 33 Following Ping (1995, 2007), the latent variable interactions were estimated using a single indicant technique. The following steps were taken in accordance with Ping (1995): (1) verifying of indicator normality, (2) assuming the latent variables are independent of the error terms and of each other, (3) unidimensionalizing each latent variable, (4) centering the observed variables at zero by subtracting the mean, (5) estimating loadings and error variances for the linear independent variable indicators using a measurement model, (6) using these estimates to calculate the estimates of the loadings and error variances for the interaction latent variable indicators and (7) specifying these estimates as fixed values in a structural model, then estimating that model. Table 1.2 presents the results of the interaction effects model. The structural model was estimated simultaneously with the measurement model using 1 raw data as an input. The overall chi-square goodness-of-fit index of 460.405 with 318 degrees of freedom, the CFI (0.970), the RMSEA (0.058), and the SRMR (0.106) all indicate a good 2 model fit. No support was found for H1 that tests the moderating impact of market uncertainty on the relationship between new product innovativeness and contract specificity. The results indicate that the moderating effect of market uncertainty is not significant (β1= -0.103, p > 0.05). The results also show that market uncertainty does not directly impact contract specificity (β= 0.048, p > 0.05). This finding is surprising and contrary to the prior literature that suggests that due to increased ex ante costs associated with market uncertainty, firms will write less specific contracts (Williamson 1996). One potential explanation is that new product introduction 1 Due to improper solutions, the factor loadings between two items for frequency of new product introduction were constrained to be equal. The LM test revealed that this constraint was valid. In addition, the error terms between the items for frequency of new product introduction and the interaction term (new product innovativeness* frequency of new product introduction) were allowed to covary. 2 Although SRMR is above the accepted cut off value, when combined with other fit indices, the results indicates that the model fits well. 34 creates a unique context in which uncertainty about the markets does not impact the way that contracts are crafted. This could occur because new product introduction is inherently risky and firms need to be flexible in their actions and decisions regarding new products, regardless of the level of market uncertainty. In support of H2, the results suggest that the interaction between new product innovativeness and performance ambiguity on contract specificity is positive and significant (β2= 0.202, p < 0.01). These results support the hypothesis that performance ambiguity positively moderates the relationship between new product innovativeness and contract specificity, but that (a) under the condition of low performance ambiguity, new product innovativeness has a negative impact on contract specificity, and (b) under the condition of high performance ambiguity, new product innovativeness has a positive impact on contract specificity. This is consistent with prior literature that has found that in ambiguous environments, contractual terms are specific, limiting opportunism (Carson et al. 2006). Given these results, it is surprising to find that performance ambiguity has a significant but negative main effect on contract specificity (β= -0.258, p < 0.01). One potential explanation for this finding is that given bounded rationality, greater performance ambiguity creates difficulties in assessing contractual compliance (Heide and John 1990), decreasing the effectiveness of specific contractual agreements and resulting in less specific contracts. 35 Table 1.2: Results of the Interaction Effect Model for Study 1 Constructs Contract Specificity Standardized Coefficient (t statistic) New Product Innovativeness 0.066 n.s. (0.841) Market Uncertainty 0.048 n.s. (0.589) Performance Ambiguity -0.258** (-3.316) Frequency of New Product Introduction 0.211** (2.699) New Product Innovativeness x Market Uncertainty -0.103 n.s. (-1.148) New Product Innovativeness x Performance Ambiguity 0.202** (2.460) New Product Innovativeness x Frequency of New Product Introduction 0.194** (2.416) Transaction Size 0.080* (1.890) Relationship Length -0.038 n.s. (-1.430) Agreement Duration 0.341** (5.627) Advertising 0.299** (3.581) 2 Notes: χ = 460.405, d.f. = 318; CFI = .970; RMSEA = .058, SRMR = 0.106, *p < .05. **p < .01. 36 The results also show that the interaction between new product innovativeness and frequency of new product introduction on contract specificity is positive and significant (β3 = 0.194, p < 0.01), in support of H3. These results indicate that when frequency of new product introduction is high, increasing new product innovativeness results in greater contract specificity. Alternatively, when frequency of new product introduction is low, increasing new product innovativeness results in lower contract specificity. Additionally, although not explicitly hypothesized, frequency of new product introduction was found to have a strongly significant and positive main effect on contract specificity (β = 0.211 p < 0.01), suggesting that manufacturers want to protect the margin stream generated by frequent new product introductions. Finally, three out of four control variables were statistically significant. Transaction size (β = 0.080 p < 0.05), agreement duration (β = 0.341 p < 0.01), and advertising (β = 0.299 p < 0.01) were all found to positively and significantly impact contract specificity. As the size of the transaction, contract duration, and advertising increase, manufacturers draft more specific contracts because the ex post hazards for them increase (Heide 1994, Joskow 1987). The impact of the relationship length on contract specificity was found to be negative and not significant (β = -0.038 p > 0.05). To improve the understanding of the moderating effects of performance ambiguity and frequency of new product introduction, post-hoc graphical analyses were performed. A plot of the interaction effects is presented in Figure 1.2. This plot was created by adapting the procedure described in Aiken and West (1991), using standardized path coefficients (Cortina et al. 2001). Standardized coefficients were used because the intercept for the unstandardized equation can 37 only be generated from the use of mean structures which are not provided when using full information maximum likelihood estimation. Figure 1.2: Graphical Interpretation of the Moderation Effects of the Frequency of New Product Introduction (Study 1) For interpretation of the references to color in this and all other figures, the reader is referred to the electronic version of this dissertation. A: The Effect of the Performance Ambiguity on Contract Specificity Contract Specificity 0,6 0,4 0,2 0 ‐0,2 -0,4 -0,6 low New Product Innovativeness high _ 0 Low Performance Ambiguity High Performance Ambiguity 38 Figure 1.2 (cont`d) Contract Specificity B: The Effect of the Frequency of New Product Introduction on Contract Specificity 0,6 0,5 0,4 0,3 0,2 0,1 0 -0,1 -0,2 -0,3 ‐0,4 low New Product Innovativeness _ high 0 Low Frequency High Frequency In Figure 1.2, panel A shows the moderating effect of performance ambiguity and panel B shows the moderating effect of frequency of new product introduction. As the graphs indicate, the impact of new product innovativeness on contract specificity differs with the levels of performance ambiguity and frequency of new product introduction. Specifically, when performance ambiguity is low, increasing new product innovativeness decreases contract specificity. When performance ambiguity is high, increasing new product innovativeness increases contract specificity. A similar effect can be observed for the frequency of new product introduction. Increasing new product innovativeness under the condition of a high frequency of new product introduction increases contract specificity. Increasing new product innovativeness under the condition of a low frequency of new product introduction decreases contract specificity. 39 Discussion of the results for Study 1 The literature on contracting offers considerable insights into designing contracts (Lusch and Brown 1996, Mooi and Ghosh 2010). The literature on innovation studies how new product launch impacts new product profitability (Atuahene-Gima 1995, Wuyts et al. 2004). Research in each of these two streams, however, focuses either on contractual design or on new product launch, addressing each topic in isolation. This study deviates from this research by examining contractual issues for new product launches that face unique challenges. These challenges include high risk, high uncertainty, and the necessity for manufacturers (a) to closely cooperate with retailers to ensure new product success and (b) to obtain accurate, complete, and timely information about the product`s retail performance. Since these challenges influence the way contracts are designed, manufacturers need to understand how the conditions surrounding new product launches impact the way contracts should be crafted. The main purpose of this study is to better understand how manufacturers manage new product introductions and how their governance differs under various conditions surrounding the exchange. Overall, the study shows that when designing contracts, the type of the new product that the manufacturer launches does not have a direct impact on contract specificity. However, this does play an important role in contracting under varying conditions of performance ambiguity and frequency of new product introduction. Therefore, this study shows the importance of the contingent alignment and finds support for the propositions put forth by Governance Value Analysis. The results also suggest that manufacturers should pursue different strategies when launching new products under different conditions. Specifically, the results demonstrate that market uncertainty does not directly impact contract specificity and does not moderate the relationship between new product innovativeness 40 and contract specificity. While this finding is surprising and contrary to the prior literature (Williamson 1996), it is possible that the impact of market uncertainty on contractual designs in the context of new product introductions is not significant. This may be because new product introduction is inherently risky: there is always a level of uncertainty about the markets, requiring firms to be flexible in their actions and decisions regarding new products regardless of the extent of market uncertainty. The main as well as moderating effects of performance ambiguity and frequency of new product introduction were found to be significant. Specifically, the results demonstrate that when performance ambiguity is high and frequency of new product introduction is also high, manufacturers will craft more specific contracts. This is consistent with the predictions of Governance Value Analysis (Ghosh and John 1999), which states that as transaction costs increase, manufacturers will craft more specific contracts to safeguard against ex post hazards. The finding that when performance ambiguity is low, increasing new product innovativeness results in decreasing contract specificity is in accordance with the theoretical prediction. In the presence of low performance ambiguity, when it is easy to assess the retail performance of new products, it is likely that the value generated from the exchange relationship will be equitably split and that the proper product support will be implemented. Therefore, manufacturers launching highly innovative products will craft less specific contracts, affording them the necessary flexibility without incurring large ex post costs. However, the finding that when performance ambiguity is high, increasing new product innovativeness results in increasing contract specificity was surprising. Although it was hypothesized that when performance ambiguity is high, the impact of new product innovativeness on contract specificity will be less negative than when performance ambiguity is low, the finding of a positive relationship was 41 unexpected. This result suggests that manufacturers are more concerned with protecting themselves against opportunism, ex post exploitation, and retailer`s non-compliance (all associated with the presence of high performance ambiguity) than with leaving contractual terms open to permit flexibility in adjusting to changing circumstances. Frequency of new product introduction is another context under which new products are launched that was found to influence contract specificity. Frequently launching new products increases the complexity of the relationship between manufacturers and retailers and raises both ex ante and ex post costs. For example, frequently changing products on the shelves and continuously calibrating product offerings to better fit the market requires flexibility and adaptation on the part of both manufacturers and retailers, raising ex ante costs. Frequent new product introductions, however, also create greater value through product enhancements that must be protected from exploitation by the retailer, creating ex post hazards. This study shows that when frequency of new product introduction is high, as new product innovativeness increases, manufacturers will craft increasingly specific contracts. Alternatively, when frequency of new product introduction is low, these contracts will become less specific with increasing new product innovativeness. This result is in accordance with the theoretical prediction. When frequency of new product introduction is high, manufacturers are better able to respond to changing consumer needs and calibrate their product offerings to provide consumers with greater variety and choice. Therefore, as new product innovativeness increases, the benefits associated with frequent new product introductions become enhanced, resulting in negotiation of increasingly specific contracts. In this way, manufacturers protect their investments and manage value-claiming so that value creation is not affected negatively (Ghosh and John 1999). 42 Alternatively, when the frequency of new product introduction is low, increasing new product innovativeness was found to result in decreasing contract specificity. This result suggests that since predicting the circumstances of exchange for highly innovative products is difficult, and since ex post hazards to highly innovative products are lower than when frequency of new product introduction is high, manufacturers craft less specific contracts. This gives manufacturers the ability to adapt to changing circumstances without the fear of having large margin streams vulnerable to ex post exploitation. Study 2: Experimental Design Sampling and Data Collection To test the hypotheses addressing how governance impacts retailers’ relational behaviors toward the new product after it is launched, the impact of contract specificity and new product success on product-related relational behaviors over the duration of the contract is examined. Since the focus is on understanding retailers’ behaviors, the respondents were sampled from a list of managers working for retail firms. To ensure the appropriateness of the respondents, the participants were screened based on two key criteria: they had to be informed about decisions made regarding new products and they had to be knowledgeable about the contractual terms negotiated for new products between the retail organization and manufacturers. Participants who fit all of the screening criteria were allowed to proceed to the experiment. A random sample of managers was selected from a proprietary online panel of a market research company. In total, 224 responses were received from approximately 812 potential participants. After careful examination, 26 responses were excluded due to large missing data on key variables or poor quality. The final sample consisted of 198 completed and usable responses, 43 a 24% response rate. This approximates the response rates of comparable studies administered online. Nonresponse bias was assessed using the Amrstrong and Overton`s (1977) procedure by comparing early and late respondents on the study constructs. The results indicate that nonresponse bias is minimal because no significant differences were found on the key constructs used in the study. The respondents in the final sample have an average of 16 years of experience in the industry and work for retailers with median sales of $1 million and median number of employees of 78. The top product categories represented in the sample include food (23%), health and beauty (12%), and apparel (6%), supplied by manufacturers who represent on average 35% of the retailer`s category business. Stimuli and Measures Following Ganesan’s (2010) approach of anchoring on a real supplier before proceeding to the experiment, Study 2 was organized into two parts. In Part A the respondents were first asked to think of a branded new product (any product needing a new SKU) that (a) one of their suppliers just launched through them, (b) whose performance was not yet known and (c) for 3 which a 12 month contract had been signed . The respondents were then asked to provide background information about the manufacturer, the new product, and the initial contractual terms. The following variables were measured in Part A: contract specificity, new product innovativeness, relationship length, and transaction size. All measures for these constructs are 3 The duration of 12 months was selected because Study 1 revealed that the most common contractual length for new products is 12 months. Additionally, it was necessary to control for contract duration because the length of the contract can impact the behaviors and investments of the partners in an exchange (Brickley et al. 2006). When contracts are signed for a short time frame, the retailers are less likely to engage in product-related relational behaviors because the contracts may soon expire. As contractual length increases, presence of relational behaviors also increases. 44 identical to measures used in Study 1, using a three-item, five point Likert scale for each construct (Appendix 1.5). In Part B, the respondents were directed to a longitudinal experiment. In order to understand and predict actors` behaviors over time, repeated measures were used that required the participants to report on the extent of their engagement in product-related relational behaviors at three different points during the 12-month contractual time period. The experiment manipulated new product success between subjects (high success vs. low success) over three different time periods (within subjects). To test the hypotheses, three between-subject treatment conditions were examined. In the first condition, the success of the new product was manipulated to be high across all three time periods (henceforth referred to as the high product success group). In the second condition, the success of the new product was manipulated to be low across all three time periods (henceforth referred to as the low product success group). In the third condition, the product success was low in the first two time periods and high in the third time period (henceforth referred to as the low/high product success group). The respondents were randomly assigned to one of the three experimental conditions in which they were given three hypothetical scenarios over the imagined duration of the contract (one in each time period). The respondents were asked to imagine that the new product had been on the market for 3 months (for stage 1 manipulation), 6 months (for stage 2 manipulation), and 9 months (for stage 3 manipulation) of the 12 month contract and that its sales were either far above or far below the category average, depending on the treatment condition. Appendix 1.3 provides the scenario descriptions. At the end of each quarter (3 month period), the respondents were asked to report on their product-related relational behaviors. Consistent with prior literature, relational behaviors are 45 operationalized as a second-order latent construct consisting of three first-order dimensions: solidarity, flexibility, and information exchange (Hoppner and Griffith 2011, Lusch and Brown 1996). Solidarity captures the desire to work for mutual concerns (Dwyer et al. 1987) and is defined as the extent to which the retailer works jointly with the manufacturer for the benefit of the new product. Flexibility captures the willingness to make adaptations as circumstances change (Heide and John 1992) and is defined as the extent to which the retailer works around the contract for the benefit of the new product. Information exchange enhances communication between two parties and is defined as the extent to which the retailer proactively provides useful information about the new product to the manufacturer. The measure for solidarity uses a threeitem, five-point Likert scale; the measures for flexibility and information exchange use two-item, five point Likert scales. All three measures are adapted from Hoppner and Griffith (2011) and Lusch and Brown (1996) and are reported in Appendix 1.4. To examine the psychometric properties of latent constructs (contract specificity and product-related relational behaviors), a hierarchical confirmatory factor analysis was conducted since prior research operationalizes relational behaviors as a second order construct (Hoppner and Griffith 2011, Lusch and Brown 1996). Appendices 4 and 5 report the scales, factor loadings, and reliabilities for these constructs. Table 1.3 reports the means, standard deviations, and correlations between contract specificity, product-related relational behaviors and control variables. 46 Table 1.3: Means, Standard Deviations and Correlations for Study 2 Construct 1. Conract Specificity 2. Relational Behaviors 3. Relationship Length 4. Transaction Size 5. NP Innovativeness Mean 3.68 SD 0.92 1 1 2 3 3.63 0.68 0.39 1 2.28 1.03 -0.01 0.01 1 8.25 3.54 3.06 0.79 0.09 0.13 0.08 0.34 4 0.23 0.15 5 1 0.07 1 The fit of the measurement model meets the critical values for a model of a good fit (Hu and Bentler 1999): chi-square goodness-of-fit index 59.68 with 32 degrees of freedom, comparative fit index (CFI) 0.74, root mean square error of approximation (RMSEA) 0.066 and standardized root mean-square residual (SRMR) 0.049. All items load highly on their first-order 4 factors, and the first-order factors load highly on the second-order factor. The final measure of product-related relational behaviors consists of the mean of the first-order dimensions. The results of the measurement model, the factor loadings, and the reliabilities are presented in Appendix 1.4. Finally, the manipulation checks for new product success across the three time periods were included at the end of the experiment (the measures for the manipulation checks are included in the Appendix 1.5). Significant mean differences in the correct directions are found for low versus high new product success groups for stage 1 manipulation (3.47 vs. 3.75, t = 2.54, p <0.05), stage 2 manipulation (3.42 vs. 3.91, t = -4.04, p <0.01)) and stage 3 manipulation (3.18 vs. 3.83, t = -4.81, p <0.01). 4 To avoid an improper solution, a constraint was imposed on the error term of solidarity on relational behavior. The LM test revealed that this constraint was valid. 47 Control Variables Because the respondents were asked to anchor on a real manufacturer (Ganesan 2010), three control variables were included in the model: relationship length, new product innovativeness, and transaction size. The length of the relationship between the manufacturer and the retailer was included because the longer the exchange partners conduct business with each other, the more likely they are to engage in relational behaviors and accept short-term disadvantages (Lusch and Brown 1996). Transaction size was included because as the size of the transaction increases, hazards to retailers also increase (Heide 1994). This motivates the retailers to engage in product-related relational behaviors. Finally, when new products are highly innovative, they take longer to succeed (Bass 1969), but they also have a greater potential for financial returns and profitability (Sorescu 2003). Both of these things increase motivation to engage in relational behaviors. The measures for all of the control variables are identical to Study 1 and are reported in Appendix 1.5. Since the final dataset contains missing values on transaction size (21%), these values were imputed using EM method. This was done because of a high level of missing data, because of a small sample size that excludes the possibility of listwise deletion, and because of a nonrandom missing data pattern. Hypotheses testing A repeated-measures ANCOVA model was used to test the main effect of contract specificity on relational behaviors and the impact of new product success on development of relational behaviors over time. Since contract specificity was not manipulated but measured prior to the manipulations as a continuous variable, it was dichotomized around the mean (3.67) into low contract specificity and high contract specificity. Length of the relationship, transaction size, 48 and new product innovativeness were included as covariates. Since transaction size and relationship length were non-normally distributed with a long right tail, the natural logarithm of these variables was taken and used in further analysis. The results are summarized in Table 1.4. Table 1.4: Results of Study 2 df F-values Relational Behaviors Covariates Relationship Length Transaction Size NP Innovativeness Time x Relationship Length Time x Transaction Size Time x NP Innovativeness 1 1 1 2 2 2 0.45 n.s. 3.38 n.s. 11.93** 0.42 n.s. 2.00 n.s. 0.67 n.s. Within subject effects: Time Time x NP Success Time x Contract Specif. 2 4 2 0.90 n.s. 5.87** 0.48 n.s. 2 1 1 7.64** 16.69** 0.84 n.s. Source Between subject effects: NP Success Contract Specificity NP Success x Contract Specificity *p < .05. **p < .01. The multivariate results show no significant main effect of time (Wilks`s lambda = 0.99, F = 1.13, p >0.05), and no significant interaction between contract specificity and time (Wilks`s lambda = 0.10, F = 0.32, p >0.05). No significant effect is also found for the interaction between time and (a) new product innovativeness (Wilks`s lambda = 0.99, F = 0.70, p >0.05), (b) relationship length (Wilks`s lambda = 0.99, F = 0.85, p >0.05), and (c) transaction size (Wilks`s 49 lambda = 0.99, F = 1.45, p >0.05). The results do, however, show an interaction between time and new product success (Wilks`s lambda = 0.91, F = 4.55, p <0.01). The test of between-subject effects reveals a significant main effect of contract specificity on relational behaviors (F = 16.69, p < 0.01), supporting H4. The marginal means indicate that at low levels of contract specificity, engagement in relational behaviors is significantly smaller (MRelBeh = 3.46) than when contract specificity is high (MRelBeh = 3.81). This suggests that high contract specificity results in a greater engagement in relational behaviors. Additionally, the main effect of new product success on relational behaviors is also significant (F = 7.64, p < 0.01). Specifically, the high product success group was found to have the highest engagement in relational behaviors (MRelBeh = 3.83), followed by the low/high success group (MRelBeh= 3.65). The lowest engagement in relational behaviors was observed in the low success group (MRelBeh = 3.42). Post-hoc tests using Fisher`s Least Significant Distance (reported in Table 1.5A) reveal that all means are significantly different from one another, except for the mean between the high and high/low product success groups. Finally, the interaction between new product success and contract specificity was not significant (F = 0.84, p > 0.05), in line with the theoretical argumentation. The test of within-subject effects shows no significant impact of time (F = 0.90, p > 0.05) or significant interaction between time and contract specificity (F = 0.48, p > 0.05). A significant interaction, however, is found between time and new product success (F = 5.87, p < 0.01). Supporting H5, the results show that new product success impacts relational behaviors over time. To test hypotheses H5a-c, the marginal means for each new product success group over time were estimated and a series of post-hoc procedures conducted using Fisher`s Least Significant Distance (see Tables 1.5B and 1.5C). Additionally, plots representing the marginal means of 50 relational behaviors across different new product success groups in each time period were graphed (Figure 1.3) to aid interpretation. Comparison of marginal means in the high product success group over time reveals that engagement in relational behaviors increases over time (Table 1.5C), from MRelBeh_1 = 3.73 in quarter 2 to MRelBeh_2 = 3.89 in quarter 3 and MRelBeh_3 = 3.88 in quarter 4. The marginal means between quarter 1 and quarter 2 are significantly different (p <0.01), supporting hypothesis H5a that when new product success is high, engagement in relational behaviors increases over time. It is, however, interesting to note that there is no significant difference between the means in quarter 2 and quarter 3. Comparison of marginal means in the low product success group over time reveals that engagement in relational behaviors decreases over time from MRelBeh_1 = 3.53 in quarter 2 to MRelBeh_2 = 3.50 in quarter 3 and MRelBeh_3 = 3.24 in quarter 4, supporting hypothesis H5b. There are no differences between the marginal means in quarters 2 and 3. However, there is a significant difference between the mean in quarter 4 and quarters 2 and 3 (p < 0.01). This suggests that retailers engage in relational behaviors even though the new product is not successful for a period of time before ceasing their engagement. Finally, in the low/high product success group, the marginal means over time do not significantly differ. In quarter 2 the marginal mean is MRelBeh_1 = 3.64, in quarter 3 it is MRelBeh_2 = 3.64, and in quarter 4 it is MRelBeh_3 = 3.66. This is contrary to H5c, which hypothesized that relational behaviors over the duration of the contract would first decrease and then increase. 51 Table 1.5: Post-hoc Analyses for Relational Behaviors A. Separate Analyses for High Success, Low Success and High/Low Success groups New Product Success High vs. Low High vs. Low/High Low vs. Low/High Mean Difference Relational Behaviors 0.41 0.19 -0.22 Sig. 0.000 0.079 0.035 B. Separate Analyses for Time Period 1, Time Period 2, and Time Period 3 New Product Success Quarter 2 Quarter 3 Quarter 4 High vs. Low High vs. Low/High Low vs. Low/High High vs. Low High vs. Low/High Low vs. Low/High High vs. Low High vs. Low/High Low vs. Low/High Mean Difference Relational Behaviors 0.21 0.09 -0.11 0.39 0.25 -0.14 0.63 0.22 -0.42 Sig. 0.056 0.943 0.290 0.001 0.036 0.253 0.000 0.100 0.001 C. Separate Analyses for Low Product Success, High Product Success and Low/High Product Success New Product Success High new product success Low new product success Low/ High new product success Mean Difference in Relational Behaviors -0.16 -0.15 0.01 0.03 0.28 0.26 0.00 -0.02 -0.03 Quarter 2 vs.3 Quarter 2 vs.4 Quarter 3 vs.4 Quarter 2 vs.3 Quarter 2 vs.4 Quarter 3 vs.4 Quarter 2 vs.3 Quarter 2 vs.4 Quarter 3 vs.4 52 Sig. 0.005 0.091 0.858 0.632 0.001 0.000 0.942 0.795 0.680 Relational Behaviors Figure 1.3: Graphical Interpretation of the Impact of New Product Success on Relational Behaviors Over Time (Study 2) 4 3,9 3,8 3,7 3,6 3,5 3,4 3,3 3,2 3,1 3 _ Q2 Q3 Q4 Performance H/L success L success 0 H success When comparing the means of each product success group in a specific time period, one can see interesting findings. The results reported in Table 1.5B show no significant difference among the means of the product success groups in quarter 2. This suggests that the new product performance in the first quarter has no impact on the level of engagement in relational behaviors. In quarter 3, the difference in marginal means between the low product success and low/high product success groups continues to be statistically insignificant, while the mean of the high product success group is statistically different (p <0.05). This suggests that retailers respond to the success of the new product by increasing their engagement in relational behaviors, while the level of engagement in the low product success groups remains unchanged. 53 The greatest statistical differences across all means (p <0.001) are found in quarter 4, suggesting that retailers ceased to engage in relational behaviors in the low product success group while continuing their engagement in the high and low/high product success groups. Discussion of the results of Study 2 The goal of Study 2 is to address the second research question of how contractual governance established at the time of product launch impacts development of relational behaviors over the duration of the contract. Although prior literature has examined the relationship between contractual designs and relational behaviors (Ferguson et al. 2005, Zheng et al. 2008), it has primarily focused on the management of buyer-supplier relationships. Examination of the development of relational behaviors over time has been limited. Therefore, this study advances on prior literature by studying the engagement in, and development of, product-related relational behaviors for new products over time. The main challenge when launching new products is that they entail large risks and uncertainties, which on the one hand encourage involvement in product-related relational behaviors (Poppo and Zenger 2002, Zeng et al. 2008) to ensure the product`s success, but on the other hand discourage them due to the uncertainty of future performance and revenues. The experimental design in Study 2 confirms that contract specificity positively impacts relational behaviors. Therefore, when contract specificity is high, engagement in relational behaviors is higher than when contract specificity is low. This is consistent with recent research that finds that specific contracts promote relational behaviors (Ferguson et al. 2005, Zheng et al. 2008) It also supports the propositions put forth by Ghosh and John (1999, p. 133) that when exchange partners are exposed to ex post hazards they try to minimize their disadvantage “(1) by scaling back investment, (2) adapting less and (3) forgoing activities hazardous from a 54 measurement standpoint.” These results suggest that specific contracts decrease the severity of risk, encourage cooperation (Poppo and Zenger 2002), and increase retailer`s willingness to exchange information, be flexible, and work for mutual benefit. The results of how relational behaviors develop over time given the performance of the new product are interesting. They show that after the new product has been on the market for three months, new product success has no impact on the level of engagement in relational behaviors. One explanation is that the future performance of the new product at this stage remains uncertain. The expectation of future returns is therefore also uncertain, causing retailers to wait until this uncertainty is resolved. After the new product has been on the market for 6 months of the 12-month contract, the results show that retailers significantly increase their engagement in product-related relational behaviors in the high product success group. This increase is significant relative to the past, as well as relative to the other groups, where the product success to this point has been low. The finding is interesting because engagement in relational behaviors can be costly in terms of time and resource allocation (Larson 1992) and these investments need to be warranted. Continuous high product success appears to resolve the uncertainties regarding the new product`s future performance and the expectation of high future returns drives engagement in relational behaviors in the present. This is consistent with prior research that expectation of positive future returns extends the expectation of continuity, which positively impacts engagement in relational behaviors (Lusch and Brown 1996, Heide and Miner 1992). Alternatively, in this time period, no differences in the level of engagement in productrelated relational behaviors relative to the past are observed when product success is low. In other words, poor product performance does not cause a decrease in product-related relational 55 behaviors. There could be two explanations for this finding. First, the expectation of the new product`s future performance, and therefore of future payoffs, remains uncertain. As a result, retailers are willing to keep incurring short-term costs on behalf of the product, in hopes of improving its future performance. Second, retailers are locked into a 12-month contract on a new product that generates limited returns, and this motivates them to engage in product-related relational behaviors in an effort to minimize losses. After the new product has been on the market for 9 months of the 12-month contract, new product success has the greatest impact on the level of engagement in relational behaviors across all groups, since all means significantly differ from one another. In the high product success group, retailers continue to engage in product-related relational behaviors, but there is no increase from the prior period. The expectation of future returns and the expectation of extending (or renewing) the 12-month contract prompt the retailer to continue engaging in product-related relational behaviors. This is consistent with prior findings that expectation of future exchanges will encourage cooperation (Lusch and Brown 1996, Poppo and Zenger 2002). In the low product success group, the level of relational behaviors significantly decreases in the last quarter, relative to the past. The mean level of engagement in relational behaviors is also significantly below the mean of the other groups. This finding suggests that retailers cease to make further investments into the new product and therefore decrease their level of engagement in relational behaviors. Since relational behaviors develop over time, and each transaction must be viewed in the context of its history and anticipated future (Lusch and Brown 1996), this finding is not surprising. Given the product`s poor past performance and limited expectation of future exchanges and returns, retailers are not motivated to incur costs on behalf of a failing product. 56 Finally, it is interesting to note that the level of relational behaviors during the contractual period in the low/high product success group did not change over time. Although an increase in the level of product-related relational behaviors in last quarter was hypothesized, it is possible that retailers are hesitant to act on a behalf of a product when its past performance has been poor. This suggests that the future performance of the new product remains uncertain, and as a result, retailers do not increase nor decrease their engagement in relational behaviors. GENERAL DISCUSSION Drawing on Governance Value Analysis as a theoretical foundation, this essay investigates how manufacturers govern the introduction of new products and how this governance impacts retailers’ behaviors after the new product is launched. Specifically, this essay investigates (1) the impact of the interaction between new product innovativeness and exchange attributes on contract specificity (2) and the implications of contract specificity and product success on behaviors toward the new product after it is launched. Since new product introductions are inherently risky (Abetti 2000), establishment of appropriate distribution channel activities is a key factor in the launch of new products (DiBenedetto 1999). Manufacturers launching new products through retailers are removed from the actual point of sale and they have limited or delayed information about the new product`s performance. This increases their reliance on the retailer, slows their response times, and creates substantial requirements for coordination, joint decision making, and information sharing. Therefore, proper governance mechanisms must be devised to allow manufacturers to minimize their costs and maximize their value (Ghosh and John 1999) and to ensure the necessary cooperation and information sharing on behalf of the new product after it is launched. 57 Overall, this essay empirically demonstrates the importance of using a contingent alignment framework. Its conclusions support the propositions put forth by Governance Value Analysis, both for designing contracts and for predicting the behavior of the parties in an exchange over time (Ghosh and John 1999). The following discussion of theoretical and managerial implications reflects the three focal objectives of this essay: (1) to examine how manufacturers govern the introduction of new products in a way that minimizes transaction costs and maximizes value, (2) to test how this governance impacts retailers’ relational behaviors toward the new product after it is launched, and (3) to study how relational behaviors develop over time, given the performance of a new product during the contractual period. Theoretical Contributions This essay makes several important theoretical contributions to the marketing literature and Governance Value Analysis. First, it extends the theory by incorporating firms’ innovation efforts as a firm-specific resource and tests their impact on governance given exchange attributes. While existing literature focuses on problems associated with either contractual designs or new product launches, each topic is addressed and studied in isolation. For example, while the literature on contracting offers considerable insight into designing contracts (Lusch and Brown 1996, Mooi and Ghosh 2010), it does not study contractual designs for new products that face unique challenges such as high risk, high uncertainty, and necessity for closer cooperation between the partners. Additionally, the literature on innovation focuses on factors that impact new product profitability (Atuahene-Gima 1995, Wuyts et al. 2004), but research is limited in addressing distribution-related issues. This essay fills that gap by studying contracts for new 58 products that face unique challenges that influence the balance of ex ante and ex post transaction costs and thus alter the way governance mechanisms should be established. Second, although some advances have been made in the development and testing of Governance Value Analysis, its empirical support remains limited (Ghosh and John 2005). This essay contributes to the theory by empirically testing its propositions in the context of new product introductions. Overall, the results show support for the propositions of Governance Value Analysis. Specifically, Study 1 shows that the type of the new product that the manufacturer launches does not have a direct impact on contract specificity, but that this plays an important role in contracting under varying conditions of performance ambiguity and frequency of new product introduction. This suggests that manufacturers should pursue a wide array of strategic options when launching new products under different conditions. Study 2 then shows support for the propositions about the behavior of the parties in an exchange over the duration of the contract. Ghosh and John (1999) argue that governance will impact the behavior of the parties in an exchange over time. Specifically, they argue that when value claiming is not properly managed, the exchange partners will try to minimize their ex post disadvantage by scaling back investments or by adapting less. Specific contracts thereby allow the exchange partners to engage in relational behaviors because formal contracts help ensure that the early (and more vulnerable) stages of exchange are successful, that the severity of the risk to which the exchange partners are exposed is narrowed (Poppo and Zenger 2002), and the generated margin stream is protected (Mooi and Ghosh 2010). Therefore, this essay confirms the findings of recent research (Zheng et al. 2008) that contract specificity positively impacts engagement in relational behaviors. 59 Third, although prior literature has examined the relationship between contractual design and the development of cooperative relationships (Ferguson et al. 2005, Zheng et al. 2008), the examination of the development of relational behaviors over time has been limited. Particularly in the context of new product introductions, where retailers lock themselves into contracts (usually for a period of 12 months) for products whose performance is uncertain, understanding how relational behaviors develop over time is important. The findings of this essay are that new product success plays an increasingly important role over time. When the new product is launched, its level of performance does not impact the extent to which retailers engage in product-related relational behaviors. However, over time, the product`s performance has an increasing impact on engagement in product-related relational behaviors. This suggests that as uncertainty associated with new products decreases and the observability of the product`s performance increases over time, the expectation of future returns becomes more certain, influencing the level of engagement in product-related relational behaviors. This is consistent with prior literature that argues that the level of engagement in cooperative norms is a result of a calculative process (Bercovitz et al. 2006) and that the expectation of future returns alters the level of cooperation in the present (Poppo and Zenger 2002). Managerial Implications The findings of this essay have important implications for manufacturers as well. Manufacturers who sell their products through retail chains face numerous challenges. First, they must continuously innovate to avoid obsolescence of their product lines (Montgomery 1975), but new product launches can be very costly (Ogawa and Piller 2006) and it can be challenging to negotiate favorable terms of trade (Iyer and Villas-Boas 2003). Second, manufacturers must be able to respond quickly and adapt to any changes after new products are launched, but they are 60 removed from the actual point of sale and so they have limited or delayed information about a new product`s performance. This increases their reliance on the retailer, slows their response times, and creates substantial requirements for the coordination of various tasks, joint decision making, and information sharing. Therefore, design of proper governance mechanisms for the introduction of new products must be carefully devised (a) to allow manufacturers to maximize their value, (b) to protect the generated margin stream (Ghosh and John 1999), (c) to minimize risks, and (d) to ensure the necessary cooperation and support for new products from retailers. Given that manufacturers have a wide array of strategic options for managing new product introductions, establishing the most efficient governance mechanisms can be a complex task. This essay provides guidance to manufacturers on how to manage their new product introductions to maximize and protect returns and to ensure proper support after new products are launched. The key findings of this essay are that manufacturers should negotiate specific contracts when (a) the potential for future returns is high, such in the case of highly innovative new products (Sorescu 2003) and frequent introductions, or (b) when there is uncertainty over the way returns may be divided for highly innovative products, such in the case of a high performance ambiguity (Ghosh and John 2005). This finding is interesting because low familiarity with technologies and/or markets for innovative new products (Abetti 2000) makes it difficult to specify circumstances of exchange beforehand, favoring less specific contracts. The findings, however, suggest that specific contracts decrease the severity of risks associated with new product launches, protect manufacturers from ex post exploitation, and thus protect the generated margin stream. This is consistent with some literature that suggests that formal contracts help ensure that early, more vulnerable, stages of exchange are successful (Poppo and Zenger 2002). 61 The findings also suggest that manufacturers should negotiate less specific contracts only under specific circumstances: (1) when the expected returns from new products are limited and (2) when the retail performance of the new product is directly observable, allowing the value generated from the exchange relationship to be equitably split (Ghosh and John 2005). In addition, the finding that when contracts are not specific, involvement in product-related relational behaviors is lower than when contract are specific suggests that less specific contracts should be negotiated when the need for joint cooperation, extensive information exchange, and flexibility after the new product is launched is relatively small. Finally, while the performance of the new product after it is launched initially has little impact on the level of engagement in product-related relational behaviors, contract specificity plays an important role across the entire duration of the contract. Therefore, it is important for manufacturers to design their contracts in a way that supports the new product in its early, more vulnerable stages. Over time, should the new product be successful, retailers will increase their engagement in product-related relational behaviors and act on the behalf of the new product. Limitations and Future Research While this essay provides insight into how manufacturers govern the introduction of new products, this essay has several limitations that future research could address. First, although using a longitudinal experiment enhances the causal inferences, the cross-sectional nature of the survey instrument limits the determination of the direction of causality. In the survey, it was conceptualized that exchange attributes would influence the relationship between new product innovativeness and contract specificity. It could be argued, however, that previous contractual designs could influence the number of new products that the manufacturer decides to launch and 62 the innovativeness of these products. Future research should work to determine the direction of causality more clearly. Second, the sample is restricted to U.S. manufacturers and retailers. The way that new products are governed could differ across countries, and this would threaten the generalizability of the results. Additionally, although this essay collected data from manufacturers as well as retailers, the implications regarding the way governance is viewed and treated by manufacturers as opposed to retailers is limited. Future research could examine the differences in governing new products between manufacturers and retailers. Third, this essay focuses on the introduction of a single new product by a manufacturer through a retailer. New products, however, are launched in the context of past new product introductions, competitive new product introductions, established relationships, and expectations of future exchanges. Since these contextual factors are beyond the scope of this essay, how these factors impact governance of new products could be addressed by future research. Fourth, the way that contract specificity is operationalized is abstract and does not fully capture the specific contractual terms negotiated between manufacturers and retailers. For example, the specificity of the terms may vary for price, profit sharing conditions, payment terms, shelf space allocation, advertising support, promotional schedules, purchase quantities, merchandising efforts, and so forth. Inclusion of more specific contractual measures would enhance our understanding of how manufacturers govern the introduction of new products and how this governance impacts the behavior of actors during the contractual time period. Finally, future research could examine the impact of various contextual factors on the way new product introductions are governed. This includes not only the impact of past new product introductions and the expectation of future exchanges, but also the impact of existing 63 relationships between manufacturers and retailers and the impact of relationships with other exchange partners on the way new products are governed. 64 ESSAY 2 NEW PRODUCT LAUNCHES AND MANUFACTURERS` RETURNS: UNDERSTANDING THE IMPACT OF NEW PRODUCT LAUNCH DECISION ON VALUE CLAIMED AND CREATED FROM NEW PRODUCTS In the continuing search for competitive advantage, manufacturers increasingly turn to innovation and the development of new products. They do this to generate increased sales and profits (Abetti 2000, Chandy and Tellis 2000). However, the value that innovations generate is limited when new products are launched through large retailers, because manufacturers must share a portion of the value with their retailers. This is a challenge, since manufacturers and retailers are rivals for value extraction. Manufacturers often complain that retailers creatively find unpredictable ways to extract additional revenues (Iyer and Villas-Boas 2003) and that retailers gain additional profits at their expense (Dukes et al. 2006). Manufacturers are also increasingly dependent on retailers for the success of new products because the concentration of retailing across several sectors limits the distribution channels available to manufacturers (Hultink et al. 1998). For example, dominant retailers have become the gatekeepers for numerous new products; refusal by such retailers to carry certain products may block national distribution and negatively impact new product performance (Luo et al. 2007). Since the negotiating power of manufacturers is limited, it is important to understand how new product launch decisions influence retailers’ willingness to share a greater portion of the returns from successful innovation efforts. Specifically, this essay examines how manufacturers’ past new product launch decisions stimulate or inhibit retailers’ willingness to share a greater portion of the value from new product launches with manufacturers. . Unfortunately, our understanding of the impact that new product launch decisions have on returns from the retail channel is limited at present. Prior research on new product launches 65 has examined the benefits of innovation to manufacturers (Calantone et al. 2010, Chandy and Tellis 2000) and the importance of distribution strategies and distribution channel decisions in new product launches (Hultink et al. 1998, Luo et al. 2007, Montgomery 1975). Limited attention, however, has been paid to the conflict that can arise between exchange partners who must share returns. Similarly, prior channel literature offers considerable insight into channel relationships and the distribution of returns (Leventhal et al. 1969, Samaha et al. 2011), but studies addressing value sharing and extraction for new products are scarce. A deeper understanding is needed of how manufacturers and retailers share returns from new products and what factors determine how returns are divided. Such an understanding would allow manufacturers to launch new products more profitably and help managers make better decisions regarding new product launches. Therefore, the goal of this essay is to address these gaps in prior research and examine the question of how manufacturers’ past new product launch decisions interact with prior new product success in determining manufacturers’ returns on current new product introductions. To answer this question, this essay draws on the literature of reciprocity. Reciprocity has been defined as a universal social norm, where an action performed by one party requires a compensating movement by the other party (Gouldner 1960). The notion of reciprocity is appropriate for this study because it is at the core of marketing relationships and plays a complementary role to the self-interest that also occurs in economic exchanges (Bagozzi 1995). Reciprocity decreases the incidence of exchange partners shirking their responsibilities and reaping rewards without giving back (Cook and Rice 2006). It therefore plays an important role in value creation and value distribution between two exchange partners. As a result, two outcome variables are included in the model: value claimed by the manufacturer and value created from 66 new products. Additionally, since reciprocity “evokes obligation toward others on the basis of their past behavior” (Gouldner 1960, p. 170), this essay examines when prior new product success stimulates reciprocity in retailers, allowing manufacturers to claim and create greater value from new products. Manufacturers’ new product launch decisions (i.e., the innovativeness of the manufacturer`s products, frequency of new product introduction, and degree of selectivity) are then hypothesized to moderate the relationship between prior new product success and manufacturers’ returns. These product launch decisions are theorized to alter the feeling of indebtedness and impact the magnitude of the retailer`s reciprocal response, leading to increased value created from new products and claimed by the manufacturer. In summary, the contribution of this essay is threefold. The first contribution is to extend the research on reciprocity and test whether a retailer`s reciprocal behavior is manifested in performance variables and whether there are circumstances that stimulate rather than inhibit the magnitude of a reciprocal response. The second contribution is to extend the literature on innovation by enhancing our understanding of how manufacturers’ new product launch decisions impact value sharing and value creation with a retailer. The third contribution is to provide guidance to managers on how past new product launch decisions impact returns from new product launches. THEORETICAL BACKGROUND The notion of reciprocity has been widely applied in the literature across multiple disciplines (Axelrod 1981, Bosse et al. 2009, Falk and Fischbacher 2006, Umphress et al. 2010), as well as marketing (Anderson and Weitz 1992, Bagozzi 1975, Dwyer et al. 1987, Hoppner and Griffith 2011). Reciprocity is defined as a social norm whereby an action performed by one party 67 requires a compensating movement by the other party (Houston and Gassenheimer 1987). Several authors highlight the importance of reciprocity by stating that it is a universal norm (Gouldner 1960) which is at “the core of marketing relationships” (Bagozzi 1995, p. 275) and important for the development and maintenance of exchange relationships (Dwyer et al. 1987, Axelrod 1981). In economic exchanges, reciprocity plays a role complementary to self-interest by providing self-regulatory control over one`s actions (Bagozzi 1995). Laying a foundation for the norm of reciprocity, Gouldner (1960) identifies it as a key variable in stabilizing relationships whereby a person should give benefits in return for receiving them. In other words, when one partner in an exchange receives a benefit from the other, this recipient becomes indebted to the donor and remains so until the debt is repaid. This repayment, however, may not be immediate (immediacy reciprocity) nor equal (equivalence reciprocity) to what has been received. Gouldner (1960) therefore argues that it is possible that benefits supplied to an exchange partner will be repaid at a later time, or will be only partially repaid. This proposition was recently tested empirically by Hoppner and Griffith (2011), who found that equivalence and immediacy in reciprocity prescribe which relational behaviors are appropriate for firms to perform and when these behaviors should be performed. Additionally, Gouldner (1960) argues that reciprocity is a quantifiable variable. The extent of reciprocity in exchange relationships may vary; it may be completely absent in some relationships (i.e. an exchange partner gives nothing in return for benefits received). This proposition is interesting because our understanding of the conditions under which a reciprocal response may vary remains largely limited. Whether reciprocity is present in marketing exchange relationships, and the extent to which it is present, is crucially important for the development and maintenance of those relationships (Dwyer et al. 1987, Axelrod 1981). First, in order for 68 relationships to develop, the positive actions of one party must be reciprocated by the other party (Dwyer et al. 1987). Should the actions not be reciprocated, exploitation may occur, resulting in an unequal exchange (Goudner 1960) and the breakdown of the relationship. By contrast, acts of reciprocity nurture social relationships through a cycle of giving and countergiving. This can be demonstrated by the commitments that partners make to the relationship, whereby each channel member`s commitment is dependent on the perception of the other party`s commitment to the relationship (Anderson and Weitz 1992). In other words, the behavior of one party is contingent on the probability that the other party will reciprocate one`s actions (Axelrod 1981). The main risk that an exchange partner faces is that of making an investment or incurring a cost on behalf of the relationship that will not be reciprocated (Palmatier et al. 2009). Expectation of receiving back what one contributes to the relationship suggests that reciprocity plays an important role in the way returns are shared and distributed between exchange partners. Multiple studies support this proposition. First, reciprocity was found to provide an additional motivation to develop and maintain relationships over and above economic incentives (Pervan 2009), and thus to decrease the incidence of exchange partners shirking their responsibilities and reaping rewards without giving back (Cook and Rice 2006). Second, reciprocity limits self-interested behavior since it is rooted in self-regulation and control over one`s actions (Bagozzi 1995), thus it balances relationships. This is consistent with the view that reciprocity is a moral norm, under which partners should give benefits in return to those who give them benefits, because of an obligation to repay which transcends self-interested behavior (Gouldner 1960). Third, reciprocity allows partners in an exchange to reward and punish each other`s moves, disciplining one another (Rokkan et al. 2003). In ongoing relationships, this is possible due to repeated interactions characterized by either positive or negative reciprocal 69 behaviors, that is, tit for tat strategies (Axelrod 1981). Thus, “through such expectations of reciprocity…the future casts a shadow back upon the present, affecting current behavior patterns” (Parkhe 1993, p. 799). Relationships characterized by high power asymmetry illustrate how reciprocity impacts exchange relationships and limits self-interested behaviors. When power asymmetry is high, that is, when one partner is dependent on another for valued resources (Dwyer et al. 1987), the use of coercive influence strategies and opportunism by the less dependent partner increases (Frazier 1986). This positive impact of power asymmetry on opportunism should, however, be mitigated by reciprocity, because it increases the motives to repay and share benefits, even when power differences may favor exploitation (Gouldner 1960). The norm of reciprocity should therefore safeguard powerful parties against the temptations of their own power and inhibit the emergence of exploitative relations (Gouldner 1960). A commonly held definition of reciprocity is that it is a norm driven by a feeling of indebtedness, which results in a moral obligation to repay (Goudner 1960). Several performance outcomes and antecedents were found to influence and to be influenced by the feeling of indebtedness leading to reciprocal response. For example, Palmatier et al. (2009) examine the impact of gratitude and gratitude-based reciprocal behaviors on performance. Although this study was conducted in the context of relationship marketing, it highlights the important role that reciprocity may play in obtaining returns and enhancing one`s performance. Gratitude, as argued, influences how people perceive and repay benefits gained from the exchange relationship, impacting performance outcomes (Palmatier et al. 2009). While past research examined the importance of reciprocity in exchange relationships, our understanding is largely limited of what circumstances stimulate rather than inhibit 70 reciprocal responses and whether there are circumstances and situations in which reciprocity may not develop at all. This is particularly true in the context of the manufacturer-retailer dyad, which is characterized by a high competitiveness that disfavors reciprocity. For example, manufacturers frequently make new product launch decisions (such as what types of new product to launch, how frequently to launch new products and through what distributors) without carefully considering how these may impact the retailer`s behavior and either stimulate or inhibit their motivation to reciprocate. The purpose of this essay is to address this gap. THE PROPOSED MODEL Building on the literature on reciprocity that provides an explanatory causal mechanism for the suggested relationships among variables, the proposed model tests the relationship between prior new product success and manufacturers’ returns from new product launches (specifically, value claimed by the manufacturer and value created from new products), moderated by manufacturer`s product launch decisions. (Figure 2.1 depicts the proposed model.) Prior new product success is defined as the commercial performance of a manufacturer`s new products over the past three years, relative to industry average (Gatignon and Xuereb 1997). The level of a manufacturer`s returns in response to prior new product success is hypothesized to be a direct result of actions by the retailer to repay the manufacturer and to reciprocate the benefits received in response to a moral obligation to repay (Gouldner 1960). Two outcome variables, which capture manufacturers’ returns, are included in the model. (1) Value claimed by the manufacturer (henceforth referred to as value claimed) is defined as the portion of the value that the manufacturer claims on new product launches relative to average manufacturers. (2) Value created from new products (henceforth referred to as value created) is defined as the size of total 71 outcomes that the manufacturer can generate from new product launches, relative to average manufacturers. Value claimed and value created represent, respectively, the portion of the pie that the manufacturer claims and the size of the pie that can be generated from new products. Figure 2.1: Proposed Theoretical Model Innovativeness of Manufacturer`s Products Value Claimed by the Manufacturer Prior New Product Success Value Created From New Products Degree of Selectivity Frequency of New Product Introduction Manufacturers’ past product launch decisions are hypothesized to moderate the relationship between prior new product success and value claimed and value created. These past product launch decisions include the innovativeness of new products launched by the manufacturer (innovativeness of manufacturers’ products), the number of new products launched (frequency of new product introduction), and the extent to which the manufacturer has been selective in introducing new products through the retailer (degree of selectivity). Innovativeness of manufacturers’ products refers to how innovative manufacturers’ products were in the past, on average, compared with the new products of other manufacturers (Fang 2008). The innovativeness of manufacturer`s products was selected because of its demonstrated importance in the innovation literature and because of its questionable link to 72 performance outcomes. While some prior research shows that innovative and novel products provide firms with higher performance (Calantone et al. 2010), other research shows that less innovative products are the ones that provide firms with higher performance due to lower uncertainty and higher synergy with firm`s resources (Song and Parry 1996). The difficulty of determining the relationship between new product innovativeness and performance is associated with the fact that highly innovative products represent a high-return but also a high-risk strategy (Abetti 2000, Sorescu 2003). It is therefore important to understand how retailers respond to introductions of highly innovative products, specifically how these introductions impact retailers’ reciprocal behavior and thus manufacturers’ returns. This is particularly significant since retailers are playing an increasingly important role in the success of new products (Luo et al. 2007). Frequency of new product introduction refers to the number of new products introduced in the past by the manufacturer. This moderator was selected because manufacturers seek to improve their performance not only by increasing the innovativeness of their products, but also by increasing how frequently they introduce new products. As a result, prior literature has focused on studying factors that lead to a greater number of new products introduced by a firm (Katila and Ahuja 2002). Unfortunately, our understanding is limited of how this strategy impacts the retailer`s behavior and therefore the manufacturer`s returns. Do retailers prefer frequent new product launches to keep their product lines fresh, or do they prefer less frequent new product launches to achieve lower costs, higher supply-chain efficiency, and lower complexity? Degree of selectivity refers to the extent to which the manufacturer refrained from launching new products through competing retailers in the past (Fein and Anderson 1997). This 73 moderator was included because prior research suggests that selectivity can be viewed as a pledge, or a credible commitment, strengthening interorganizational relationships (Fein and Anderson 1997). This has important implications for new product launches, since manufacturers must continuously decide on how many sales outlets should be established in a particular geographical area and through which retailers new products should be launched. While low selectivity increases new product availability and exposure to consumers (Frazier and Lassar 1996), high selectivity, while limiting manufacturer`s sales (Fein and Anderson 1997), may result in greater margins. It is therefore important to understand whether manufacturers launching new products benefit from increased selectivity in the form of increased returns from new product launches. HYPOTHESES According to the norm of reciprocity, an action performed by one party requires a compensating movement by the other party (Houston and Gassenheimer 1987). When one partner to an exchange receives a benefit from another, the recipient becomes indebted to the donor and remains so until the debt is repaid. Some authors argue that the mere recognition of a benefit generates an obligation to repay (Becker 1986), and this motivates the partner to “increase compliance with any subsequent requests” (Palmatier et al. 2009, p.4). In the context of the manufacturer-retailer dyad, as the retailer receives benefits from a manufacturer`s successful past new product introductions, a sense of indebtedness should develop, motivating the retailer to give to the manufacturer in return. It is hypothesized that this will be manifested in behaviors that allow the manufacturer to claim or create greater value from current new product launches. A more detailed description of this process follows. 74 The development and launch of new products is inherently risky. Research shows that newly launched products suffer from high failure rates, often over 50% (Ogawa and Piller 2006). Retailers understand this problem. As a result, they set up safeguards to protect themselves and to minimize losses from failed new products. For example, retailers make manufacturers bear all the risks and costs associated with new product introduction by mechanisms such as slotting fees (Rao and Mahi 2003, Sullivan 1997) or contractual requirements in which manufacturers guarantee certain minimal sales levels in order to gain distribution (Iyer and Villas-Boas 2003). Retailers may also demand that manufacturers bear the costs of advertising and promotion for new products. All of the above-mentioned mechanisms increase manufacturers’ costs and significantly limit the value that they can claim from new product launches. Additionally, retailers may provide minimal support for new products or allocate limited shelf space, which inhibits the value that can be created from new product introductions. As a result, it is difficult for manufacturers to launch new products successfully and profitably, because the majority of the benefits generated by new product introductions are claimed by retailers. Prior New product success. While retailers’ behavior to protect themselves against risks is understandable given high new product failure rates, when a manufacturer has demonstrated the ability to innovate by introducing successful new products over time (i.e., having high prior new product success), this behavior should change. According to the norm of reciprocity, the safeguards that retailers set up should decrease and retailers should allow manufacturers to receive more favorable terms of trade from current new product launches or to provide better support for the manufacturer`s products. High prior new product success is hypothesized to stimulate retailers’ sense of indebtedness and motivation to reciprocate for two reasons. 75 First, the retailer receives benefits at the expense of the manufacturer, because the costs and risks associated with new product launch are mostly carried by the manufacturer, but the benefits generated from successful new products are primarily claimed by the retailer. According to the norm of reciprocity, the recognition of a benefit received should generate an obligation to repay (Becker 1986) and stimulate reciprocal response (Pervan 2009). Second, prior new product success is expected to increase the motivation to reciprocate because manufacturers incur costs on behalf of the relationship without receiving anything in return. Specifically, when the manufacturer is for a time willing to accept less favorable terms of trade in hopes that the retailer will reciprocate at a later time, demonstrates a long-term orientation toward the relationship and shows a willingness to make short-term sacrifices in order to obtain long-term benefits (Dwyer et al. 1987). This way, manufacturers make intentional investments that are costly and that entail some risks, signaling their commitment to the relationship, which should stimulate reciprocity (Tesser et al. 1968, Leventhal 1969). Therefore, it is hypothesized that the greater the prior new product success, the greater the obligation to repay, and the greater the retailer`s motivation to reciprocate, via increasing manufacturers’ returns. In a manufacturer-retailer dyad, when reciprocity is stimulated, the retailer is expected to reward the manufacturer and provide the manufacturer with additional benefits. This includes increasing the manufacturer`s returns from new products. This act of reciprocity as a behavioral response to a manufacturer`s prior actions can occur in two ways. First, the retailer can repay the manufacturer by sharing a greater portion of the profit margins on current new product launches, increasing the portion of the value that the manufacturer can claim. This can be done, for example, by decreasing safeguards that protect retailers from losses associated with failed new 76 product introductions. It can also be done by lowering slotting fees (Rao and Mahi 2003) or by reformulating contractual terms to provide manufacturers with more favorable terms of trade and thus increasing their value claimed. Second, the retailer can also positively reciprocate by helping the manufacturer increase the value created from new product launches. For example, the retailer may increase shelf space allocation, provide a better location in the store, or improve product support. Through these behaviors, the retailer increases its investments to support the manufacturer`s new products and signals good faith toward the manufacturer through a willingness to repay (Frazier and Lassar 1996). Therefore: H1a: Prior new product success will positively impact value claimed by the manufacturer H1b: Prior new product success will positively impact value created from new products It is hypothesized that some past new product launch decisions made by manufacturers will inhibit a retailer`s motivation to reciprocate, while other decisions will enhance this motivation, thus altering the magnitude of change in the value claimed and value created as a result of prior new product success. This is consistent with Gouldner`s (1960) view that reciprocity may occur in exchange relationships to varying degrees. For example, while some relationships may be balanced with strict reciprocity, others may be characterized by unequal exchanges where the extent to which each party reciprocates differs. Still other relationships may be characterized by an absence of reciprocity: the exchange partner gives nothing in return for benefits received. It is hypothesized that the magnitude of the reciprocal response to prior new product success will differ with varying product launch decisions because these alter either the perception of the benefits that the retailer receives or the actual value of the benefits received from prior new product successes. Specifically, manufacturers’ new product launch decisions are 77 hypothesized to alter the magnitude of change in the value claimed and created in response to prior new product success. These product launch decisions include the innovativeness of new products launched by the manufacturer (i.e., the innovativeness of manufacturer`s products), the number of new products launched (i.e., frequency of new product introduction), and the number of retailers through which the new products are launched (i.e., degree of selectivity). The Innovativeness of Manufacturers’ Products. The relationship between prior new product success and value claimed and value created is hypothesized to be contingent on the innovativeness of manufacturers’ products. When new products are highly innovative, they are distinct from existing products in the product category in terms of product design, the ability to satisfy new needs or wants, and the degree to which they embody new technologies or unique features (Calantone et al. 2006, Calantone et al. 2010, Chandy and Tellis 1998). Increasing the innovativeness of the manufacturer`s products when prior new product success is high is hypothesized to generate a greater reciprocal response by the retailer and thus alter the magnitude of change in the value claimed and value created. The reasoning is as follows. Prior research suggests that increasing the amount of risk and cost that one party incurs on behalf of the relationship should increase the motivation to reciprocate (Palmatier et al. 2009, Tesser et al. 1968). In the context of this study, when the innovativeness of a manufacturer`s products is high, the manufacturer incurs greater investment costs than when the innovativeness of these products is low. This is because retailers, in an effort to protect themselves against the losses often associated with new products, set up higher safeguards for more innovative products than less innovative products. This is because innovative new products have a higher likelihood of a failure due to larger uncertainties and risks (Abetti 2000) than less innovative products. As a result, retailers charge higher slotting fees (Sullivan 1997), demand contractual safeguards that 78 may call for guarantees of profits, or require a buyback of unsold products (Iyer and Villas-Boas 2003). These factors negatively impact manufacturers’ returns. Additionally, highly innovative products require greater promotion and advertising than less innovative products (a cost also borne by manufacturers), further decreasing manufacturers’ returns. Since manufacturers are required to make substantial investments when launching innovative products, their returns from new product launches are greatly limited. However, while innovative products generate substantial costs, successful innovative products generate substantial returns (Sorescu 2003). Thus, when prior new product success is high and the innovativeness of manufacturer`s products is also high, (a) the manufacturer incurs greater costs while (b) the retailer gains greater benefits than when the innovativeness of products is low. Therefore, increasing the innovativeness of manufacturers’ products should enhance the reciprocal response to prior new product success and result in a greater value claimed by the manufacturer and greater value created from new products. Moreover, enhanced reciprocal response is also expected because as the innovativeness of manufacturers’ products increases, it becomes more difficult to launch new products successfully. Since retailers are aware of this fact, they are more likely to recognize and acknowledge this manufacturer`s new product launches as more valuable, increasing the magnitude of the reciprocal response. Finally, successful and highly innovative products provide more value to retailers than less innovative products because they are distinct from existing products in the category and have far greater potential to generate value. Since prior literature suggests that reciprocity increases with increasing value of the benefit to the recipient (Tesser et al. 1968), this further supports the arguments made above. 79 Alternatively, when the innovativeness of manufacturers’ products is low, the impact of high prior new product success on value claimed and created is hypothesized to be smaller than when innovativeness is high. This is because non-innovative products have, on average, relatively low failure rates and low launch costs. As a result, the safeguards that retailers set are low, and manufacturers are able to negotiate more favorable terms of trade, decreasing the retailer`s motivation to reciprocate. Additionally, the value of the benefits that less innovative and successful products generate to retailers is smaller than benefits generated by more innovative and successful products, also resulting in a smaller motivation to reciprocate positively. These arguments are consistent with prior research that the smaller the benefit received, the smaller the feelings of indebtedness, and the smaller the reciprocal response (Haisley and Loewenstein 2011). For the above reasons, the impact of prior new product success on the retailer`s motivation to reciprocate, when coupled with low innovativeness of manufacturer`s products, is hypothesized to be smaller than when the innovativeness of manufacturer`s products is high. This is hypothesized to result in a smaller magnitude of change in value claimed and value created. Therefore: H2a: All else being equal, when the innovativeness of manufacturer`s products is high, the positive impact of prior new product success on value claimed by the manufacturer is greater than when the innovativeness of manufacturer`s products is low H2b: All else being equal, when the innovativeness of manufacturer`s products is high, the positive impact of prior new product success on value created from new products is greater than when the innovativeness of manufacturer`s products is low Frequency of New Product Introduction. Considering the volume of new products introduced today and the availability of large product offerings, the practice of product proliferation by many manufacturers is evident. Product proliferation is a common marketing strategy of leading food manufacturers (Connor 1981) that is characterized by a large number of 80 new product introductions, wide product variety, and long product lines. For example, Crest and Colgate had at one point more than 35 different types of toothpaste (Quelch and Kenny 1994), while in the beverage category, almost two thousand new SKUs are added each year (Khermouch 1995). While some manufacturers undertake the strategy of product proliferation, other manufacturers, choose to limit their new product offering and concentrate on their most popular innovations (Quelch and Kenny 1994). SmithKline Beecham`s Aquafresh toothpaste provides an example of this approach. While prior literature addresses the benefits (e.g., increase in the overall demand, better satisfaction of consumer`s needs) and costs (e.g., increase in production costs, lower channel efficiency, difficult trial purchase and evaluation) of product proliferation, our understanding of how this strategy may change retailers' behavior and thus value claimed and value created from new products is limited. Increasing the frequency of new product introduction when the prior success of new products is high has several effects. First, as the frequency of new product introduction increases, the benefits that the retailer obtains from each successful new product introduction decrease, resulting in a decrease in the retailer`s motivation to reciprocate. This is because a high frequency of new product introduction reduces the retailer`s average turnover rate and profit per SKU (Quelch and Kenny 1994). For these reasons, manufacturers with a lower frequency of new product introduction allow retailers to gain higher direct product profits on each new product launch than manufacturers with a high frequency of new product introduction. For example suppose that two different manufacturers generate the same returns on their new products. One does so through frequent new product introduction while the other does so through fewer but more impactful new product introductions. The retailer is expected to be less likely to feel indebted to repay the manufacturer with a high number of product introductions because each 81 new product launch is not as impactful. This is hypothesized to manifest itself in manufacturers’ returns. Further, prior research shows that the magnitude of retailers’ reciprocal responses can also change with the retailers` attributions of the motive and intention behind manufacturers` actions (Leventhal 1969). The strategy of high frequency of new product introduction is sometimes adopted by manufacturers to ward off competition and raise the admission price to the category for new, smaller brands or private label competitors (Quelch and Kenny 1994). Since this strategy may hurt the retailer`s product category, decrease efficiency, and increase costs associated with managing the category, the retailer is hypothesized to be less likely to reciprocate and increase value claimed and value created by the manufacturer. Alternatively, a low frequency of new product introduction is hypothesized to increase the impact of high prior new product success on the retailer`s response and thus value claimed and value created. This is because relationships between manufacturers and retailers, when the frequency of new product introduction is low, are less complex and it becomes easier to keep an account of balanced returns (Pervan 2009). Since high product variety and change in product offering raises costs, reduces efficiency, confuses consumers, and leads to shortages of popular products (Berman 2010), low frequency of new product introduction keeps retailers’ administrative costs minimal while increasing their average turnover rate and profits per SKU (Quelch and Kenny 1994). Since the retailer receives greater benefits, the retailer should feel indebted and be more likely to increase the value claimed by the manufacturer or to increase the value created from new products. Therefore: H3a: All else being equal, when the frequency of new product introduction is high, the positive impact of prior new product success on value claimed by the manufacturer is lower than when the frequency of new product introduction is low 82 H3b: All else being equal, when the frequency of new product introduction is high, the positive impact of prior new product success on value created from new products is lower than when the frequency of new product introduction is low Degree of Selectivity. The last proposed moderator that is hypothesized to impact the relationship between prior new product success and value claimed and value created is degree of selectivity. This refers to the extent to which the manufacturer has refrained from launching new products through competing retailers in the past (Fein and Anderson 1997). When the degree of selectivity is low, manufacturers are selective in their choice of retail outlets, that is, they limit the number of retailers who are allowed to carry certain products. Although prior literature discusses selectivity in regard to the brand, this study examines selectivity for new products, since it is not uncommon to offer exclusive distribution for a single new product (e.g., a specific package size) under a brand that is carried by multiple retailers. Overall, a high degree of selectivity is hypothesized to leverage the positive impact of prior new product success on value and value created, for several reasons. First, manufacturers are willing to limit their distribution not only to enhance their image, but also to promote retailer support of their products (Frazier and Lassar 1996). When doing so, manufacturers incur costs associated with lost sales opportunities due to more limited market coverage. They also run the risks associated with the possibility that the retailer will not reciprocate but will rather exploit this favorable position (Fein and Anderson 1997). For these reasons, increasing one`s selectivity has been viewed in the literature as a credible commitment that signals goodwill and invites reciprocal action (Anderson and Weitz 1992). This view is consistent with other research that demonstrates that when an exchange partner incurs costs (Tesser et al. 1968) or risks (Palmatier et al. 2009) on behalf of the relationship, the sense of indebtedness and motivation to repay increases. Therefore, when prior new product success is 83 high, increasing the distribution selectivity should leverage the positive impact of prior new product success on value claimed and value created. Second, willingly increasing one`s degree of selectivity is a manufacturer`s intentional strategy, whereby the manufacturer acts in accordance with its own free will in selecting only certain retailers through which to launch its new products. Since prior research shows that actions done intentionally increase the motivation to reciprocate (Leventhal 1969), increasing the degree of selectivity is proposed to enhance the relationship between prior new product success and value claimed and value created. In addition, when degree of selectivity is high and prior new product success is also high, manufacturers’ products generate greater competitive advantage and benefits for the retailer, since there are no other distributors in the geographical area who are allowed to carry the same products. As a result, the greater the benefits received by the retailer, the greater the feelings of indebtedness, and the greater the magnitude of the reciprocal response (Haisley and Loewenstein 2011). Alternatively, when degree of selectivity is low, the impact of prior new product success on manufacturer`s returns is hypothesized to be not as pronounced as when the degree of selectivity is high. This is because when a new product has a low degree of selectivity, the number of distributors within a given market is not restricted (Fein and Anderson 1997) and the product is available in other retail outlets as well. This generates little or no competitive advantage to individual retailers. Therefore, the benefits generated from successful prior new product introductions are not as pronounced as when the distribution is limited. Additionally, a manufacturer who does not restrict its distribution does not incur any risks or costs on behalf of the relationship, and this decreases the retailer’s sense of indebtedness and motivation to reciprocate. For all of the above-mentioned reasons, it is hypothesized that value claimed by the 84 manufacturer and value created from new products will be lower when degree of selectivity is low than when it is high. Therefore: H4a: All else being equal, when the degree of selectivity is high, the positive impact of prior new product success on value claimed by the manufacturer is greater than when the degree of selectivity is low H4b: All else being equal, when the degree of selectivity is high, the positive impact of prior new product success on value created from new products is greater than when the degree of selectivity is low METHODOLOGY Overview Two studies are conducted to test the proposed hypotheses. The goal of Study 1 is to test the proposed model in order to explore the question of how new product launch decisions impact manufacturers’ returns from new product introductions (see Figure 2.1). This study examines how the innovativeness of manufacturer`s products, the frequency of new product introduction, and the degree of selectivity impact the relationship between prior new product success and (a) value claimed and (b) value created (see Figure 2.1). Study 1 is a field survey administered to retail managers in the food and health-and-beauty product categories. To analyze the results of Study 1, structural equation modeling (SEM) is used to examine the series of simultaneous relationships among the key constructs. The goal of Study 2 is threefold. First, Study 2 attempts to replicate the findings of Study 1 in an experimental design while expanding the focus from food and health-and-beauty product categories to consumer packaged goods (CPG). This minimizes any product-specific category effects. Second, Study 2 extends the findings of study 1 by testing whether reciprocity is the underlying mechanism that explains the hypothesized relationships. The internal validity of 85 experimental designs and the possibility of isolating hypothesized effects allows one to test the underlying causal mechanisms that explain the hypothesized relationships. The third goal is to explore the effects of time on the development of reciprocity in new relationships. Study 2 employs a longitudinal experimental design administered to retailers in the CPG category. To analyze the results of Study 2, repeated measures ANOVA and ANCOVA models are used. Study 1: Field Survey Research Context and Data Collection Study 1 is a field study examining dyadic relationships between manufacturers and retailers in the food and health-and-beauty product categories. These two industries were selected for two key reasons. First, manufacturers in both industries engage in frequent innovations, thus providing an appropriate context for studying new product launches. Second, the food and health-and-beauty product categories offer a large array of diversified products that differ vastly in terms of the magnitude of profit margins that both manufacturers and retailers obtain. The diversity associated with this sample helps minimize any product-specific category effects. Online survey questionnaires were administered to respondents by a market research company that used its proprietary online panel to contact potential participants. A random sample of 974 qualified respondents was selected from this panel. To enhance the response rate, the respondents were compensated by the market research firm for participating in this study, and follow- up emails with a second survey were sent to nonrespondents. In total, 201 completed and usable questionnaires were received for a response rate of 20.6%. In order to ensure the appropriateness of the respondents, the participants were screened based upon their job titles (i.e., buyer, category manager, store manager), product categories (i.e., 86 food, health-and-beauty), and involvement in decision making regarding new products. Participants who fit all of the screening criteria were allowed to proceed to the survey. They were directed to complete the questionnaire pertaining to a single manufacturer who had supplied the retailer with branded products in the food and/or health-and-beauty product categories, and with whom the retailer had been doing business for at least three years. Nonresponse bias was assessed using Armstrong and Overton`s (1977) procedure by comparing early and late respondents with regard to key demographic variables and study constructs. The results indicate that nonresponse bias is minimal because no significant differences were found on any of the items used in the study. The final sample represented a balance between the food (54.2%) and health-and-beauty (45.8%) product categories. The median sales of the retailers in the sample were $10 million and 86% of the firms generated sales of over $1 million. The median number of employees was 300. The manufacturers selected in the sample had been engaged with these retailers for an average of 15 years, and on average, 39% of the selected category business went to these manufacturers. Lastly, the respondents had on average 15 years of industry experience and they were either retail buyers (17.4%), category managers (9.9%), or store managers (52.7%) who were working for U.S. retailers and were responsible for purchasing decisions in the food or health-and-beauty product categories. Measures The key constructs in this study are operationalized using multi-item reflective scales. Appendices 1 and 2 report the scales for the key constructs and control variables. Table 2.1 reports the Variance Covariance Matrix. Prior new product success is defined as the commercial performance of the manufacturer`s new products over the past three years, relative to industry average. New 87 products are defined as products that, when introduced through the retailer, require a new stockkeeping unit. A time frame of the past three years was selected for several reasons. First, selecting a time period greater than one year allows for a more accurate assessment of the successes and failures of new products, since some new products may require time to take off, while others may be carried for a while and then be discontinued. Second, selecting a three-year time period allows for a more accurate assessment of the manufacturer`s innovation abilities, since the success of new products may fluctuate from year to year. Lastly, a three-year time period is contextually relevant to retail buyers. The success of new products launched four or five years ago will most likely have little impact on the retailer`s current decision making. The measure for prior new product success builds on Kabadayi et al. (2007) and is operationalized in terms of (a) contribution to sales and (b) contribution to profit, using a fivepoint scale from “far below the industry average” to “far above the industry average”. Four items ask respondents about the extent to which the manufacturer`s new products introduced over the past three years contributed to sales, and three items ask about the extent to which these new products contributed to profits. This measure was selected because it captures both the total size of the pie (i.e., sales) generated from the manufacturer`s new products and the portion of the pie that the retailer receives from the new products (i.e., retail margins). Prior literature demonstrates that both of these factors are important in determining distribution success (Frazier and Lassar 1996, Hoch and Shumeet 1993). 88 Table 2.1: Variance Covariance Matrix for Study 1 Construct 1. Prior new product success 2. Frequency of new product introduction 3. The Innovativeness of manufacturers` products 4. Degree of selectivity 5. Value claimed 6. Value created 7. Advertising 8. Relationship length (ln) 9. Interdependence 10. Asymmetry 1 0.45 2 3 0.32 0.57 0.30 0.27 0.81 0.21 0.12 0.13 0.22 0.30 -0.07 -0.01 0.23 0.08 0.13 0.19 0.03 -0.10 0.02 0.27 0.14 0.16 0.35 0.02 -0.12 -0.00 4 5 6 7 8 9 10 0.96 0.28 0.29 0.15 -0.05 -0.41 0.03 0.39 0.41 0.17 -0.02 -0.28 0.05 0.40 0.21 -0.02 -0.31 0.02 0.74 0.05 -0.21 -0.08 0.16 0.09 -0.02 1.90 -0.03 0.54 89 Product launch decisions. Three key aspects of the manufacturer`s innovation efforts are considered. The first is the innovativeness of the manufacturer`s products, which is defined as the extent to which these products differed, on average, from new products launched by other manufacturers over the past three years. New products can have varying degrees of innovativeness that can range from incremental to breakthrough. Incremental innovations involve minor changes in technology, they offer minor improvements over existing products on the market (Chandy and Tellis 1998), and they include simple product improvements and alterations (Zhou et al. 2005). Breakthrough innovations include new products that are distinct from competitors’ products and offer new technologies, unique features (Calantone et al. 2006), and distinct benefits to consumers (Atuahene-Gima 1995). The measure for the innovativeness of manufacturers’ products uses a five-items, seven-point semantic differential scale and is adapted from Fang (2008). The second aspect considered is the frequency of new product introduction. This is defined as the number of new products introduced over the past three years by the manufacturer through a specific retailer, relative to industry average. It is important to note that the frequency of new product introduction refers only to products introduced through a specific retailer. This characteristic is important because manufacturers may frequently introduce new products to the market, but may choose to introduce only a few through certain retailers. Four items, using a five-point scale from “far below the industry average” to “far above the industry average,” were developed to operationalize the frequency of new product introduction by the manufacturer. The third aspect of manufacturers’ innovation efforts is degree of selectivity. This is defined as the extent to which, over the past three years, the manufacturer refrained from launching new products through competing retailers. When the degree of selectivity is high, 90 manufacturers are selective in their choice of retail outlets, putting limits on the number of retailers who are allowed to carry certain new products. The measure of degree of selectivity is adapted from Fein and Anderson (1997) and it uses two items, both with five-point Likert scales. Manufacturer`s returns. Two outcomes are examined in this study: value claimed by the manufacturer and value created from new products. Both of these outcome variables refer to benefits and returns that the manufacturer receives when launching new products in the present, or when launching new products in the near future. Value claimed is defined as the portion of the value that the manufacturer claims on new product launches relative to average manufacturers. This study examines the valued claimed by manufacturers in initial agreements, when the performance of new products is still uncertain. A new measure using two items, each on a five-point Likert scale, was developed to operationalize value claimed by the manufacturer relative to industry average. Value created from new products is defined as the size of total outcomes that the manufacturer can generate from new product launches, relative to average manufacturers. A new measure using three items, each on a five-point Likert scale, was developed to operationalize this construct as the extent to which the retailer provides the manufacturer with opportunities to generate greater value and the extent to which retailers work with manufacturers to generate greater value from new product launches. Control Variables Four control variables included in the model are: power asymmetry between the manufacturer and the retailer, their interdependence, the length of the relationship between the manufacturer and the retailer, and advertising support for new products. These control variables 91 were selected because they influence the retailer`s motivation to reciprocate and thus impact performance outcomes. (Appendix 2.2 presents measure for the control variables.) Power asymmetry between the manufacturer and the retailer is defined as a difference between the manufacturer`s dependence on the retailer and the retailer`s dependence on the manufacturer (Kumar 1995). Their total interdependence is defined as the sum of both firms` dependences (Kumar 1995). The measures for power asymmetry and interdependence are calculated using Kumar (1995). First, dependence of the manufacturer on the retailer (using a three-item, five-point Likert scale) and dependence of the retailer on the manufacturer (using a two- items, five-point Likert scale) were measured. Then, the average scores for the dependence of the manufacturer on the retailer and dependence of the retailer on the manufacturer were calculated. Power asymmetry scores were constructed by taking the difference between the manufacturer and the retailer dependence scores, and interdependence scores were constructed by summing the manufacturer and retailer dependence scores. These control variables were selected because when a manufacturer is dependent on the retailer, the retailer is not easily replaceable and may be less willing to reciprocate manufacturers’ actions regardless of their prior new product introductions. Alternatively, the greater the dependency of the retailer on the manufacturer, the more powerful and dominant the manufacturer becomes (Lusch and Brown 1996), affording the manufacturer the ability to extract greater returns. Additionally, dependency of the retailer on the manufacturer increases the level of competition among retailers, and this enhances the retailer`s motivation to reciprocate in order to retain the manufacturer`s business. Finally, significant interdependence between partners affects the level of conflict, trust, and commitment and alters the motivation to reciprocate (Kumar 1995). 92 The length of the relationship between the manufacturer and the retailer is included because, over time, the nature of reciprocity changes from short term to long term. Under long term reciprocity, any acts of kindness are repaid over the course of the relationship (Ryu and Feick 2007), and this affects the way value is generated and shared over time. Finally, advertising support for new products was included because it plays a very important role in new product launch and success. Should the manufacturer provide substantial advertising support for its new products, the retailer`s willingness to share value and create value from new products increases. This provides an alternate explanation for a change in value claimed and value created and must therefore be controlled for. ANALYSIS AND RESULTS Measurement Model Analysis Confirmatory factor analysis using EQS 6.1 was used to estimate a measurement model comprised of reflective multi-item latent constructs of prior new product success, the innovativeness of manufacturers’ products, frequency of new product introduction, degree of selectivity, value claimed, and value created. Appendix 2.1 presents the results of the measurement model analysis, together with item loadings and composite reliabilities. The overall chi-square goodness-of-fit index for the model is 382.50 with 215 degrees of freedom. The measurement fit indices for the confirmatory measurement model all meet the critical values for a model of good fit (Hu and Bentler 1999): comparative fit index (CFI) 0.949, root mean square error of approximation (RMSEA) 0.062, and standardized root mean square residual (SRMR) 0.044. All factor loadings are large (range: 0.62 to 0.95) and significant (tvalue > 2.00) in support of convergent validity. Discriminant validity was established by 93 examining interconstruct correlations, which should significantly depart from 1.0 (Bagozzi et al. 1991). All correlations among independent variables are significantly smaller than 1.0. The squared correlations were also compared with the average variance extracted (AVE) for each construct. For all independent variables, the AVE is larger than the squared correlations, therefore adequately confirming discriminant validity. The exception, where AVE is not larger than squared correlations, is for value claimed and value created. Since value claimed and value created are dependent variables, and since the model fit did not improve when value claimed and value created were treated as a single construct (chi-square goodness-of-fit index was 391.293 with 220 degrees of freedom, CFI 0.948 and RMSEA 0.062), they are treated in the following analyses as two separate constructs. This was done to explain and understand better the impact of independent variables on value claimed and value created. Lastly, the composite reliabilities (reported in Appendix 2.1) of constructs range from 0.76 to 0.94, indicating acceptable levels of reliability for each construct. Since both independent and dependent measures were obtained from the same source, they are susceptible to common method bias. Three separate tests were conducted to assess the presence of common method bias. First, Harmon`s one-factor test (McFarlin and Sweeney 1992, Sanchez and Brock 1996) was conducted. This test entails entering all of the items of latent variables into a single factor using confirmatory factor analysis. The fit statistics for this model were quite poor (chi-square goodness-of-fit index of 1798.89 with 230 degrees of freedom, CFI 0.523, RMSEA 0.185 and SRMR 0.155) indicating that common method bias is minimal. Second, Lindell and Whitney`s (2001) marker variable assessment technique was employed. A variable (i.e., marker) was identified beyond the scope of the study, assessing its smallest correlation coefficient with theoretical predictors. The marker variable selected was the 94 number of years that the respondents had been working in their current positions. Next, this variable`s coefficient was partialled out from the bivariate correlations. The partialled results were then compared against unadjusted correlations. After partialling out the number of years that respondents had been working in their current positions, all of the significant bivariate correlations among key predictors and outcomes maintained their statistical significance. Lastly, correlations between endogenous and exogenous errors were also examined. Collectively, the results suggest that the risk of common method bias is minimal. Hypotheses testing The conceptual model was tested using structural equation modeling. The model estimated value claimed and value created as a function of prior new product success; innovativeness of manufacturers’ products; frequency of new product introduction; degree of selectivity; the interactions between prior new product success and (a) innovativeness of manufacturer`s products, (b) frequency of new product introduction and (c) degree of selectivity; and the control variables. Since relationship length (a control variable) was nonnormally distributed with a long right tail, the logarithm of relationship length was taken and used in further analysis. The latent variable interactions were estimated following the Ping (1995,2007) singleindicant technique. The following steps were taken in accordance with Ping (1995): (1) verifying of indicator normality, (2) assuming the latent variables were independent of the error terms and of each other, (3) unidimensionalizing each latent variable, (4) centering the observed variables at zero by subtracting the mean (Bollen 1989), (5) estimating loadings and error variances for the linear independent variable indicators using a measurement model, (6) using these estimates to calculate the estimates of the loadings and error variances for the interaction latent variable 95 indicators and (7) specifying these estimates as fixed values in a structural model, then estimating that model. Table 2.2 presents the results of the interaction effects model. The structural model was estimated simultaneously with the measurement model using raw data as input. The overall chi-square goodness-of-fit index of 934.48 with 417 degrees of freedom and the CFI (0.934), RMSEA (0.079), and SRMR (0.196) all indicate that the model has 5 an adequate fit. The results suggest that the impact of prior new product success on value claimed is positive and significant (β1a= 0.222, p < 0.01), in support of H1a, but that the impact of prior new product success on value created is only marginally significant (β1b= 0.111, p = 0.06), providing partial support for H1b. The results also indicate that the interactions between prior new product success and innovativeness of manufacturers’ products on value claimed (β2a = -0.193 p < 0.01) as well as value created (β2b= -0.156, p = 0.01) are negative and significant, which is counter to H2a and H2b. Contrary to the hypothesized effects, the results suggest that the innovativeness of manufacturers’ products moderates the relationship between prior new product success and value claimed and value created negatively rather than positively. Similarly, the interactions between prior new product success and frequency of new product introduction on value claimed (β3a = 0.126, p < 0.05) and value created (β3b= 0.139, p < 0.05) were also counter to H3a and H3b. Contrary to the hypothesized effects, the results suggested that the frequency of new product introduction positively moderates the relationship between prior new product success and value claimed and value created. 5 Combination of CFA < 0.95 and SRMR >0.06 may show potential for a misspecified model. However, since SRMR is sensitive to a sample size (which is relatively small), the results were deemed to indicate an acceptable fit. 96 Table 2.2: Results of the Interaction Effect Model for Study 1 Value Claimed by the Manufacturer Standardized Coefficient (t statistic) 0.222** (2.928) Value Created from New Products Standardized Coefficient (t statistic) 0.111 n.s. (1.564) Frequency of new product introduction -0.103 n.s. (-1.369) 0.073 n.s. (1.020) Innovativeness of manufacturer`s products -0.027 n.s. (-0.350) -0.016 n.s. (-0.219) Degree of selectivity 0.323** (3.495) 0.320** (3.601) Prior new product success x Frequency of new product introduction 0.126* (1.678) 0.139* (1.948) Prior new product success x Innovativeness of manufacturer`s products -0.193** (-2.527) -0.156** (-2.166) Prior new product success x Degree of selectivity 0.057 n.s. (0.727) 0.006 n.s. (0.076) Relationship length -0.039 n.s. (-0.539) -0.047 n.s. (-0.681) Advertising 0.233** (2.494) 0.294** (3.038) Interdependence -0.177** (-2.449) -0.213** (-3.083) Constructs Prior new product success Asymmetry 0.152* 0.074 n.s. (2.095) (1.084) 2 Notes: χ = 934.48, d.f. = 417; CFI = .934; RMSEA = .079, SRMR = 0.196 *p < .05. **p < .01. 97 Finally, no support was found for H4a and H4b, where the interactions between prior new product success and the degree of selectivity on value claimed (β4a= 0.057, p > 0.05) and value created (β4b= 0.006, p > 0.05) were not significant. While these results indicate that the degree of selectivity does not act as a moderator, strong support was found for the direct effect of degree of selectivity on value claimed (β = 0.323, p < 0.01) and value created (β = 0.320, p < 0.01); both effects were positive and highly significant. To improve our understanding of the significant moderating effects of the innovativeness of manufacturers’ products and frequency of new product introduction, post hoc graphical analyses were performed. A plot of the interaction effects is presented in Figure 2.2 and Figure 2.3. This plot was created by adapting the procedure described in Aiken and West (1991), using standardized path coefficients (Cortina et al. 2001). Standardized coefficients were used because the intercept for the unstandardized equation can only be generated from the use of mean structures, which are not provided when using full information maximum likelihood estimation. Figure 2.2 shows the moderating effect of the innovativeness of manufacturer`s products and Figure 2.3 shows the moderating effect of the frequency of new product introduction. As the graphs in Figure 2.2 (panels A and B) indicate, the positive impact of prior new product success on manufacturers’ returns for a low level of the innovativeness of manufacturer`s products is greater than for a high level of innovativeness. Additionally, the graph in Figure 2.3 (panel A) shows that the positive impact of prior new product success on value claimed is greater for a high frequency of new product introduction than for a low frequency. It is interesting to note that overall, value claimed is lower when frequency of new product introduction is high than when it is low. Figure 2.3 (panel B) reveals similar results. Overall, value created is higher when frequency of new product introduction is high. 98 Figure 2.2: Graphical Interpretation of the Moderation Effects of the Innovativeness of Manufacturers` Products (Study 1) Value Claimed A: The Effect of the Innovativeness of Manufacturers` Products on Value Claimed by the Manufacturer 0,5 0,4 0,3 0,2 0,1 0 -0,1 -0,2 -0,3 -0,4 -0,5 low Prior New Product Success high _ 0 Low Innovativeness High Innovativeness B: The Effect of the Innovativeness of Manufacturers` Products on Value Created from New Products 0,4 Value Created 0,3 0,2 0,1 0 -0,1 -0,2 -0,3 low Prior New Product Success _ high 0 Low Innovativeness High Innovativeness 99 Figure 2.3: Graphical Interpretation of the Moderation Effects of the Frequency of New Product Introduction (Study 1) A: The Effect of the Frequency of New Product Introduction on Value Claimed by the Manufacturer 0,3 Value Claimed 0,2 0,1 0 -0,1 -0,2 -0,3 -0,4 -0,5 low Prior New Product Success high _ 0 Low Frequency High Frequency B: The Effect of the Frequency of New Product Introduction on Value Created from New Products 0,4 Value Created 0,3 0,2 0,1 0 -0,1 -0,2 -0,3 low Prior New Product Success _ high 0 Low Frequency High Frequency 100 Discussion of the results of Study 1 The value that new products generate for firms is limited when they are launched through large retailers because a portion of this value must be shared with the retailers, who often demand additional revenues (Iyer and Villas-Boas 2003) and profits (Dukes et al. 2006). Since the ability of manufacturers to negotiate more favorable terms of trade is limited, the purpose of this study is to examine how manufacturers’ new product launch decisions influence retailers’ willingness to share a greater portion of the benefits received from innovation efforts, increasing manufacturers’ returns. The results show that prior new product success and product launch decisions influence retailers’ motivation to reciprocate, and thus manufacturers’ returns. Therefore, this study demonstrates not only that reciprocity can be manifested in performance variables, but also that past decisions manufacturers have made regarding new products can stimulate or inhibit the magnitude of a reciprocal response from retailers and thus influence their returns from current new product launches. Two main drivers of value claimed and value created were found: prior new product success and degree of selectivity. The results suggest that increasing prior new product success and degree of selectivity stimulate retailers’ motivation to reciprocate and lead to a higher value claimed and value created. What is particularly interesting is that the degree of selectivity results in manufacturers claiming and creating greater value from new products regardless of whether they have launched successful products in the past. This supports the notion that increasing one`s selectivity is viewed by retailers as a credible commitment that signals goodwill and invites reciprocal action (Anderson and Weitz 1992), and that incurring costs (Tesser et al. 1968) or risks (Palmatier et al. 2009) on behalf of the relationship stimulates motivation to repay. 101 The main effects of the innovativeness of manufacturer`s products and frequency of new product introduction were not significant. This has important implications for manufacturers who often turn to innovations in hopes of generating increased sales and profits (Abetti 2000, Chandy and Tellis 2000). The results show that manufacturers do not directly improve their returns either by increasing the innovativeness of their products or by increasing the number of new product introductions. The impact on manufacturers’ returns is observed through a moderating role of the relationship between prior new product success and manufacturers’ returns, altering retailers’ motivation to respond positively to past new product performance. While the innovativeness of manufacturers’ products negatively moderates this relationship, frequency new product introductions moderate it positively. Innovativeness of manufacturer`s products. The way that innovativeness of manufacturer`s products moderates the relationship between prior new product success and manufacturers’ returns is surprisingly contrary to the stated hypothesis. Figure 2.2 (Panels A and B) reveal that the magnitude of change in manufacturers’ returns as a result of launching of successful products is greater for low innovativeness than for high innovativeness. This moderating effect applies to both value claimed and value created. These results suggest that the historical success of manufacturers’ new products has a far greater impact on their returns when new products are less innovative than when they are more innovative. One potential explanation for this finding is that retailers are more likely to expect non-innovative products to be successful based on past product introductions than highly innovative products. This is because less innovative products are characterized by minor changes to existing products already on the market (e.g., change in packaging, size, introduction of a new flavor etc.) (Garcia and Calantone 102 2002). If existing products are successful, retailers are more likely to expect that minor changes to these products will also be successful. Alternatively, highly innovative products satisfy new needs, embody new technologies, or include unique features (Calantone et al. 2006, Chandy and Tellis 1998). Innovative new products frequently entail large risks, and the results suggest that retailers are unwilling to carry these risks and increase manufacturers’ returns even when manufacturers have successfully launched such products in the past. Frequency of new product introduction. The finding that frequency of new product introduction positively moderates the relationship between prior new product success and manufacturers’ returns was also surprising. The plots of the interaction effects (Figure 2.3, panels A and B) reveal that when frequency of new product introduction is high, increasing prior new product success results in a larger increase in manufacturers’ returns than when frequency of new product introduction is low. This is contrary to the hypothesized effects. One potential explanation for these findings is that rather than positively reciprocating, retailers engage in negative reciprocity. In other words, the results could be interpreted as follows: when frequency of new product introduction is high, decreasing prior new product success results in a greater decrease in manufacturers’ returns than when frequency of new product introduction is low. Decreasing the success of new product introductions, while at the same time launching a large number of new products, can be very costly to retailers. Since retailers are not receiving benefits, but rather incurring costs, they may engage in negative reciprocity. Negative reciprocity is characterized by giving negative treatment in return for a negative treatment (Cropanzano and Mitchell 2005), resulting in an incentive to reduce partners` payoffs (Falk and Fischbacher 2006). 103 Key implications can be drawn from these results. First, the historical success of manufacturers’ new products has a much smaller impact on their returns when frequency of new product introduction is low than when it is high. Therefore, retailers seem to be far more sensitive to past product performance when manufacturers launch a large number of new products. Manufacturers should consider this finding when determining the number of new products to be launched. Particularly when new product launches have not been historically successful, launching of a large number of new products (hoping that some will succeed) can hurt manufacturers’ returns and negatively impact returns from future new product introductions. Looking at this issue from a retailer`s perspective, since a high frequency of new product introduction increases costs and reduces supply chain efficiency (Berman 2010), unless new products can generate the necessary minimal value, retailers will likely reciprocate negatively and decrease manufacturers` returns. Study 2: Experimental Design The goal of Study 2 is threefold. First, Study 2 attempts to replicate the findings of Study 1. The addition of an experimental design increases the internal validity of the findings, allows for an isolation of hypothesized effects, and provides evidence of the internal validity and causality of the conceptual model. The focus of Study 2 is specifically on the impact of prior new product success and degree of selectivity on value claimed and value created, while controlling for all constructs in the model tested in Study 1. These impacts are studied because the results of the field survey revealed that prior new product success and degree of selectivity are particularly strong drivers of manufacturers’ returns and therefore warrant further examination. Second, Study 2 extends the findings of study 1. The internal validity of experimental designs and the possibility of isolating hypothesized effects allows one to test the underlying 104 causal mechanisms that explain the hypothesized relationships. Positive relationships found in Study 1 between prior new product success, degree of selectivity, and manufacturers’ returns suggest that reciprocity is present in manufacturer-retailer dyads. However, alternate explanations are possible for why positive and significant relationships were found. For example, did manufacturers’ value claimed and created increase as a result of their negotiating power? Did it increase as a result of simple economic behavior, or was reciprocity the underlying mechanism? The second goal of this study is therefore to test whether reciprocity is the underlying mechanism that drives retailers to increase manufacturers’ returns. To test this proposition, two new dependent variables were measured: indebtedness and indebtedness-based reciprocal behaviors. Additionally, Study 2 extends Study 1 by expanding the focus from the food and health-and-beauty product categories to consumer packaged goods (CPG), minimizing any category-specific effects. The third goal is to explore the effects of time on the development of reciprocity in new relationships. In other words, when do retailers reciprocate and increase manufacturers’ returns from new products? For how long do manufacturers need to launch new products successfully or offer exclusive territory before retailers increase their value claimed and created? Sampling and Data Collection The respondents were sampled from a list of managers working for large or mid-size retailers who were responsible for CPGs. The majority of these managers were retail buyers (20%), category managers (11%), or store managers (22%). They had been working in their current positions for an average of 10 years. In total, 137 responses were received from approximately 685 potential respondents, representing roughly a 20% response rate. This approximates response rates in comparable studies administered online. 105 Stimuli and Measures A longitudinal experiment was conducted, using a 2 (high vs. low prior new product success) x 2 (high vs. low degree of selectivity) between subject design. In situations in which it is important to assess how actors behave over time, experiments with successive generations are necessary in order to understand and predict actors` behaviors. Therefore, repeated measures requiring the participants to make decisions at three different points in time over the course of a hypothetical relationship with a manufacturer were used. This allowed for a more accurate examination of how actors behave over time and reciprocity develops. The respondents were randomly assigned to one of four experimental conditions and given hypothetical scenarios that repeated three times, simulating a three year time period (Appendix 2.3 provides the scenario descriptions). The experiment manipulated prior new product success and degree of selectivity while holding frequency of new product introduction, innovativeness of manufacturers’ products, advertising, and relationship length constant across groups. In all conditions, the respondents assumed the role of a retail buyer for a major retail chain responsible for the development of product assortments, management of sales and margins, 6 and negotiation of contracts. Respondents were asked to imagine that Universal Company had become their new supplier and that they had complete control over decision making regarding its new product introductions. A fictitious company was used to avoid any associations that respondents may have had with existing suppliers. In each of the three time periods, participants were asked to determine whether they would change value claimed and/or value created from new products relative to the prior year`s 6 The name Universal Company has been successfully used in prior literature (Weilbaker and Blasiman 1994). 106 contract. Both value claimed and value created were measured using three items each on a sevenpoint Likert scale. These measures were identical to those used in Study 1 (see Appendix 2.4). At the end of the experiment, the levels of indebtedness and indebtedness-based reciprocal behaviors were measured. The respondents were asked to rate the extent to which they felt indebted to the manufacturer for prior new product successes and degree of selectivity (indebtedness) and the extent to which they changed the manufacturer`s returns based on their indebtedness for new product successes and degree of selectivity (indebtedness-based reciprocal behaviors). The measures for both constructs, reported in Appendix 2.4, use six-item, five-point Likert scales, building on Gouldner (1960). Table 2.3 reports the means, standard deviations and correlations for Study 2. Table 2.3: Means, Standard Deviations, and Correlations for Study 2 Construct 1. Value Claimed 2. Value Created 3. Indebtedness 4. Indebtedness based reciprocal behaviors 5. Dependence of the retailer on the manufacturer Mean 4.09 4.99 2.68 2.86 SD 0.92 0.68 1.03 3.06 1 1 0.62 0.54 0.50 2 3 4 1 3.36 0.40 1 0.81 1 3.04 0.79 0.44 0.42 0.46 0.43 5 1 The manipulation checks for prior new product success, degree of selectivity and control variables were also included at the end of the experiment, together with the measure of dependence of the retailer on the manufacturer, which as was used as a covariate in the analysis (these measures are presented in the Appendix 2.5). Significant mean differences in the correct directions were found for low and high prior new product success (2.37 vs. 3.79, t = -10.311, p 107 <0.01) and low and high degree of selectivity (2.87 vs. 3.75, t = -5.387, p <0.01). The manipulation checks for frequency of new product introduction, innovativeness of manufacturers’ products, and advertising (held constant across groups) confirmed that no differences were found across groups (p >0.01). Finally, consistent with the model examined in Study 1, three items using a five-point Likert scale were used to measure the level of dependence of the retailer on the manufacturer as a covariate. These measures were again identical those used in Study 1. Measurement Model Analysis To examine the psychometric properties of multi-item latent constructs, a Confirmatory Factor Analysis, using robust maximum likelihood estimation method, was conducted on the dependent variables: value claimed by the manufacturer, value created from new products, indebtedness, and indebtedness-based reciprocal behaviors. Appendix 2.4 presents the results of the measurement model analysis, together with item loadings, AVEs, and composite reliabilities. The fit of the measurement model for Study 2 meets the critical values for a model of a good fit (Hu and Bentler 1999): chi-square goodness-of-fit index 176.42 with 129 degrees of freedom, comparative fit index (CFI) 0.966, root mean square error of approximation (RMSEA) 0.052, and standardized RMR (SRMR) 0.049. All factor loadings are large (range: 0.60 to 0.910) and significant (t-value >2.00), in support of convergent validity. Cronbach`s alphas of 0.79 or above demonstrate good reliability. Discriminant validity is also confirmed because the average variance extracted for each construct exceeds the square of correlations between constructs (Fornell and Larcker 1981). As in Study 1, AVE was not larger than squared correlations for value created and value claimed. Since value created and value claimed are theoretically distinct constructs, and since the model fit did not improve when they were treated as a single construct 108 (chi-square goodness-of-fit index is 194.185 with 153 degrees of freedom, CFI 0.955 and RMSEA 0.059), they are treated in the following analyses as two separate constructs. Hypotheses testing Replication and Time effects. The first goal of Study 2 is to replicate the results of Study 1, specifically to test the hypotheses that prior new product success and degree of selectivity positively impact value claimed and value created. The second goal of Study 2 is to expand the results of Study 1 and to explore how value claimed and created change over time in new relationships. In other words, do retailers alter manufacturers’ returns immediately, or over time? Two repeated- measures ANCOVA models were used to test the main and the interaction effects of prior new product success, degree of selectivity, and time on (a) value claimed and (b) value created. Dependence of the retailer on the manufacturer was included as a covariate, in order to replicate the model tested in Study 1. The results are summarized in Table 2.4, and they partially confirm findings of Study 1. Table 2.4: Results of Study 2 Source Covariate Dependence of the retailer on the manufacturer Within subject effects: Time Time x PPS Time x DS Between subject effects: Prior new product success Degree of selectivity Prior new products success x degree of selectivity *p < .05. **p < .01. F-values Value Created df Value Claimed 1 12.37** 34.29** 2 2 2 9.86** 12.19** 0.57 n.s. 13.41** 20.95** 1.04 n.s. 1 11.40** 17.22** 1 1 0.26 n.s. 3.25 n.s. 0.55 n.s. 1.17 n.s. 109 Repeated- measures ANCOVA on value claimed found a significant main effect of time (Wilks`s lambda = 0.87, F = 9.63, p <0.01) and a significant interaction between time and prior new product success (Wilks`s lambda = 0.87, F = 10.09, p <0.01). No significant interaction was found between degree of selectivity and time (Wilks`s lambda = 0.99, F = 0.70, p >0.05). Overall, these results suggest that the change in value claimed over time is significant and varies with prior new product success. The test of between-subject effects reveals a significant main effect of prior new product success on value claimed (F = 11.40, p < 0.01), supporting H1a. The marginal means indicate that at low levels of prior new product success, retailers are reluctant to increase value claimed (MVCL = 3.52), while at high levels of prior new product success the change in value claimed over time is greater (MVCL = 4.29). The main effect of degree of selectivity on value claimed is not significant (F = 0.26, p > 0.05), which is contrary to the findings of Study 1. The marginal mean for low degree of selectivity is MVCL = 3.95 and for high degree of selectivity it is MVCL = 3.86. The interaction between prior new product success and degree of selectivity is also not significant (F = 3.25, p > 0.05), confirming the findings of Study 1. The results of repeated- measures ANCOVA regarding value created are identical to those regarding value claimed. A significant main effect of time (Wilks`s lambda = 0.86, F = 11.06, p <0.01) and a significant interaction between time and prior new product success (Wilks`s lambda = 0.78, F = 18.40, p <0.01) were found. Additionally, no significant interaction was found between degree of selectivity and time (Wilks`s lambda = 0.99, F = 0.93, p >0.05). Similarly to value claimed, the test of between-subject effects reveals a significant main effect of prior new product success on value created (F = 17.22, p < 0.01), supporting H1b. The marginal means indicate that at low levels of prior new product success, retailers are reluctant to 110 increase value created (MVCR = 4.09), while at high levels of prior new product success the change in value created over time is greater (MVCR = 4.97). The main effect of the degree of selectivity on value created is again not significant (F = 0.55, p >0.05), contrary to the findings of Study 1. The marginal mean for low degree of selectivity is MVCR = 4.60 and for high degree of selectivity it is MVCR = 4.47. Confirming the findings of Study 1, the interaction between prior new product success and degree of selectivity is not significant (F = 1.17, p > 0.05). To aid the interpretation of changes across time, the marginal means for each independent variable across time were estimated for both value claimed and value created and a series of post-hoc procedures was conducted. Where the results reveal significant differences across groups (or time), Fisher`s Least Significant Distance (LSD) is used to examine which specific means differ. This includes a test of the difference in means between high and low prior new product success in each time period and a comparison of means among the three time periods for high versus low prior new product success (the results are reported in Table 2.5). Second, plots representing the marginal means of value claimed and value created given prior new product success in each time period are also included (Figure 2.4). There are two ways to discuss and interpret the results. Using Table 2.5A as an example, the first way is to test the following: given a specific time period, does value claimed differ across the levels of prior new product success? For example, in the first time period, do the means in value claimed and created differ depending on whether prior new product success was high or low? The second way to discuss and interpret the results (using Table 2.5B) is to test the following: given a particular level of prior new product success, do the means differ across the three time periods? For example, when prior new product success is low, do the means in value 111 claimed and created differ depending on the time period in which these outcome variables are measured? Referring to Table 2.5A, the results reveal that prior new product success does not impact value claimed and value created immediately, but rather over time. The difference in means between high prior new product success and low prior new product success in time period 1 is not statistically significant for both value claimed and value created (p > 0.05). The estimated marginal means are MVCL = 3.94 and MVCR = 4.90 for low prior new product success and MVCL = 4.25 and MVCR = 5.08 for high prior new product success. In time period 2, however, these means are significantly different for value created (p <0.01) and value claimed (p < 0.05). The estimated marginal means are MVCL = 3.59 and MVCR = 4.08 for low prior new product success and MVCL = 4.22 and MVCR = 4.84 for high prior new product success. Although both effects are significant, the impact of prior new product success on value claimed is weaker than on value created. Finally, in the third time period, the differences across groups are highly significant for both value claimed and value created (p < 0.01). The estimated marginal means are MVCL = 3.04 and MVCR = 3.31 for low prior new product success and MVCL = 4.39 and MVCR = 4.99 for high prior new product success. These findings suggest that retailers change value claimed and value created in response to prior new product success only after observing manufacturer`s new product performance for a minimum of 2 years. As stated above, the second way to discuss and interpret the results (using Table 2.5B) is to test whether, given a particular level of prior new product success, the means differ across time. The results reveal that there are no significant differences across time when prior new 112 product success is high for both value claimed (MVCL_1 = 4.25, MVCL_2= 4.22 and MVCL_3 = 4.39) and value created (MVCR_1 = 5.08, MVCR_2= 4.84 and MVCR_3 = 4.99). Alternatively, when prior new product success is low, the means across time differ significantly for both value claimed (MVCL_1 = 3.94, MVCL_2= 3.59 and MVCL_3 = 3.04) and value created (MVCR_1 = 4.90, MVCR_2= 4.08 and MVCR_3 = 3.31). This finding is interesting and suggests that retailers are reluctant, even after a certain period of time, to improve manufacturers’ returns from new products. Even when prior new product success is high, the retailers will not significantly change the value claimed and value created relative to the initial contractual terms. In response to unsuccessful new product launches, however, retailers respond by being less likely to increase value claimed and value created; they are more likely to decrease them. This provides additional support for the presence of negative reciprocity. Table 2.5: Post-hoc Analyses for Value Claimed and Value Created A. Separate Analyses for Time Period 1, Time Period 2, and Time Period 3 B. Sig. Sig. Mean Prior new Product Mean Difference success Difference in Value Claimed Value Created Time Period 1 Low vs. High 0.31 0.191 -0.19 0.361 Time Period 2 Low vs. High -0.64 0.022 -0.77 0.008 Time Period 3 Low vs. High -1.34 0.000 -1.68 0.000 C. Separate Analyses for Low Prior New Product Success and High Prior New Product Success Prior new Product success Low prior new product success High prior new product success Time Period 1 vs. 2 Time Period 1 vs. 3 Time Period 2 vs. 3 Time Period 1 vs. 2 Time Period 1 vs. 3 Time Period 2 vs. 3 Mean Difference in Value Claimed 0.35 0.90 0.55 0.03 -0.14 -0.17 113 Sig. 0.005 0.000 0.000 0.121 0.148 0.142 Mean Difference in Value Created 0.82 1.59 0.77 0.24 0.09 -0.15 Sig. 0.000 0.000 0.000 0.109 0.571 0.277 Figure 2.4: Graphical Interpretation of the Impact of Prior New Product Success on Value Claimed and Created Over Time (Study 2) A: Graphical depiction of the relationships among Prior New Product Success (PPS), Time and Value Claimed 5 Value Claimed 4,5 4 3,5 3 2,5 2 _ Low PPS Performance Time 1 Time 2 HighPPS 0 Time 3 5 Value Claimed 4,5 4 3,5 3 2,5 2 _ Time1 Time2 Performance Low PPS Time 3 High PPS 114 0 Figure 2.4 (cont`d) B: Graphical depiction of the relationships among Prior New Product Success (PPS), Time and Value Created 5,5 Value Created 5 4,5 4 3,5 3 _ Low PPS Performance Time 1 Time 2 HighPPS 0 Time 3 5,5 Value Created 5 4,5 4 3,5 3 _ Time1 Time2 Performance Low PPS Time 3 High PPS 115 0 Reciprocity. The second goal of Study 2 is to test whether reciprocity is the underlying mechanism that drives changes in value claimed and value created. The underlying theoretical argument is that high prior new product success and high degree of selectivity will increase retailers’ feelings of indebtedness and motivation to reciprocate. As a result, the respondents were asked to rate the extent to which they felt indebted to the manufacturer for their prior new product success and degree of selectivity (indebtedness) and the extent to which they changed the manufacturer`s returns based on their sense of indebtedness (indebtedness-based reciprocal behaviors). ANOVA models were used to test the main and interaction effects of prior new product success and degree of selectivity on (a) indebtedness and (b) indebtedness-based reciprocal behaviors. The results of the first ANOVA (summarized in Table 2.6) show that prior new product success has a significant main effect on indebtedness (F = 19.14, p < 0.01). The marginal means indicate that at low levels of prior new product success, retailers do not feel indebted to manufacturers (MInd = 2.36), while their feeling of indebtedness is significantly higher when prior new product success is high (MInd = 2.99). The main impact of degree of selectivity on indebtedness was, however, not significant (F = 0.18, p > 0.05). The marginal mean for low degree of selectivity is MInd = 2.71 and for high degree of selectivity it is MInd = 2.65. The interaction between prior new product success and degree of selectivity is also found to be not significant (F = 2.03, p > 0.05). 116 Table 2.6: Results of Study 2 Source Prior new product success Degree of selectivity Prior new products success x degree of selectivity df Indebtedness 1 19.14** F-values Indebtednessbased reciprocal behaviors 8.17** 1 1 0.18 n.s. 2.03 n.s. 0.50 n.s. 0.04 n.s. ANOVA model testing of the effect of prior new product success and degree of selectivity on indebtedness-based reciprocal behaviors shows the same results. The main effect of prior new product success on indebtedness-based reciprocal behaviors is significant (F = 8.17, p < 0.01). The feeling of indebtedness is significantly higher when prior new product success is high than when it is low (MIndBeh = 2.65 vs. MIndBeh = 3.07). The main effect of degree of selectivity on indebtedness-based reciprocal behaviors is found to be not significant (F =0.50, p > 0.05) with a marginal mean for a low degree of selectivity of MIndBeh = 2.81 and for a high degree of selectivity of MIndBeh = 2.91. Again, the interaction between prior new product success and degree of selectivity is found to be not significant (F = 0.04, p > 0.05). Overall, these findings support the argument that prior new product success affects feelings of indebtedness and motivation to reciprocate resulting in a change in manufacturers’ returns. Degree of selectivity, however, is found to have no impact either on indebtedness or indebtedness-based reciprocal behaviors. Since there is no relationship between degree of selectivity and value claimed and created, further support is found for the argument that reciprocity is the underlying causal mechanism explaining the relationships in the model. When reciprocity is absent, manufacturers’ returns do not change. 117 Discussion of the results of Study 2 The experimental design in Study 2 confirms the internal validity of the survey findings (1) by isolating the hypothesized effects of prior new product success and degree of selectivity on value claimed and value created, (2) by providing evidence of causality in the conceptual model, and (3) by testing the underlying causal mechanisms that explain the hypothesized relationships. Study 2 also expands the findings of Study 1 by generalizing the results to multiple product categories and by examining how reciprocity develops in new relationships. The specific focus is on how long manufacturers need to be successful in launching new products or offer exclusive territory before retailers reciprocate and increase their value claimed and created from new products. Overall, the results of Study 2 partially replicate Study 1`s findings. Consistent with Study 1, change in value claimed and value created is consistently higher when prior new product success is high than when it is low. Contrary to Study 1—and this is a largely surprising finding—degree of selectivity has no impact on value claimed and value created, nor on indebtedness and indebtedness-based reciprocal behaviors. One potential explanation for this finding is that the impact of degree of selectivity on manufacturers’ returns is not robust across multiple product categories. This is because Study 1 examined only the food and health-andbeauty product categories while Study 2 was expanded to include CPG products. For certain CPG product categories, exclusive territory for new products is expected and with some large retailers (e.g. Wal-Mart) an exclusive territory is demanded to gain distribution. In such cases, offering of an exclusive territory does not provide manufacturers with the necessary leverage and therefore does not result in greater returns (e.g., value claimed and value created). The same explanation applies to indebtedness and indebtedness-based reciprocal behaviors. If exclusive 118 territories for new products are expected, demanded, or traditionally offered by the majority of manufacturers, the retailer is not gaining any benefits that they cannot obtain from other manufacturers. As a result, a low degree of selectivity may not always be perceived as a credible commitment (Anderson and Weitz 1992) and therefore may not always be expected to stimulate a sense of indebtedness or motivation to repay. The exploration of the question of when retailers change manufacturers’ value claimed and value created and how reciprocity develops over time in new relationships reveals that this process takes some time. In Study 2, retailers did not change value claimed and created until the later time periods. This finding suggests that although retailers change manufacturers’ returns, this change occurs after a certain period of time and is therefore more likely to occur in more mature relationships. This is consistent with prior literature that suggests that benefits supplied to an exchange partner may not be immediately repaid, but rather be repaid over time (Gouldner 1960, Hoppner and Griffith 2012). Since new products are inherently risky (Ogawa and Piller 2006), this finding is not surprising. Retailers are reluctant to change manufacturers’ returns on new products before consistent results over a longer period of time are generated. An interesting finding is that while significant differences for both value claimed and value created across treatment groups were found in time periods 2 and 3, they were driven primarily by a decrease in manufacturers’ returns in response to low prior new product success. No significant increase in manufacturers’ returns in response to high prior new product success is observed. This suggests that retailers are more likely over time to decrease manufacturers’ returns as a result of unsuccessful product launches than to increase them as a result of successful product launches. This is consistent with prior findings that propensity to punish harmful behavior is stronger than propensity to reward friendly behavior (Fehr and Gachter 2000). Since 119 reciprocity can be either positive or negative (Fehr and Gachter 2000), the results imply that retailers are more likely to reciprocate negatively than positively to new product launches. GENERAL DISCUSSION Drawing on the literature on reciprocity, this essay investigates how past new product launch decisions and prior new product success impact manufacturers’ returns from current new product introductions. The specific focus is on how new product decisions interact with new product performance, altering retailers’ willingness to reciprocate. Since new product introductions are inherently risky (Abetti 2000) and negotiation of favorable terms of trade for new products difficult, reciprocity plays an important role in maintaining relationships over and above economic incentives (Pervan 2009), limiting self-interested behaviors (Bagozzi 1995) and safeguarding manufacturers against power asymmetry (Gouldner 1960). As a result, it is important to understand how manufacturers’ past product launch decisions impact retailers’ willingness to reciprocate and therefore manufacturers’ returns. This essay empirically demonstrates that historically launching successful new products is important in determining manufacturers’ returns from current new product introductions. In addition, this essay demonstrates that the extent to which retailers reciprocate varies with manufacturers’ past new product launch decisions. The following discussion of the theoretical and managerial implications is aligned with the three focal objectives of this essay: (1) to examine whether reciprocity is manifested in performance variables and whether there are circumstances that stimulate rather than inhibit the magnitude of a reciprocal response, (2) to test how manufacturers’ new product launch decisions impact value claimed and value created from 120 new products and (3) to provide guidance to managers on how to improve their returns from new product launches. Theoretical Contributions This essay makes several important theoretical contributions to the marketing literature, particularly to the literature on reciprocity. Since reciprocity has been argued to be at “the core of marketing relationships” (Bagozzi 1995, p. 275), numerous studies have used reciprocity to explain behaviors within relationships (Bagozzi 1975, Anderson and Weitz 1992, Dwyer et al. 1987). While Hoppner and Griffith (2011) empirically test the effects of reciprocity, most studies apply reciprocity theoretically. Formal, empirical tests for the presence and the effects of reciprocity are largely lacking. The first contribution of this essay is the finding that reciprocity is present in competitive exchange relationships and that it influences the way value is created and divided between the exchange partners. The norm of reciprocity dictates that a person should give benefits in return for receiving benefits (Gouldner 1960). It is a norm driven by a feeling of indebtedness leading to a moral obligation to repay (Gouldner 1960). Prior literature did not, however, make a clear distinction between feelings of indebtedness and the willingness of exchange partners to share actual benefits. Prior to this essay, empirical tests have been lacking of whether retailers who gain benefits from manufacturer`s successful new product launches will reciprocate. The finding that reciprocity is manifested in performance variables, specifically in the way value from new products is created and shared, is important because in manufacturer-retailer relationships, a few cents’ difference in the way value is divided can mean large gains or losses. The findings of Study 1 and Study 2 jointly confirm that retailers reciprocate and that this reciprocity is reflected in the way value is claimed and created between the exchange partners. 121 Specifically, change in value claimed and created is found to be driven by prior new product success and degree of selectivity, even though the impact of degree of selectivity on manufacturers’ returns does not appear to be robust across different product categories. In other words, offering an exclusive territory may not always allow manufacturers to increase their returns from new product launches. In product categories where offering of exclusive territories is customary, or where retailers demand exclusive territory as a condition to carry new products, offering of an exclusive territory does not stimulate reciprocity and therefore does not increase manufacturers’ ability to create and claim greater value. Additionally, the extent to which prior new product success impacts manufacturers’ returns has been found to depend on their new product launch decisions. Past decisions regarding the frequency and innovativeness of new product introductions have been found to alter the extent to which retailers reciprocate in response to prior new product success. Specifically, the magnitude of change in value claimed and created is much larger for less innovative products than more innovative products, and for high frequency of new product introduction than for low frequency. These findings extend the literature on reciprocity by improving our understanding of what circumstances stimulate rather than inhibit reciprocal responses in manufacturer-retailer dyads. A noteworthy finding is that in the context of new product introductions, reciprocity develops over time. Since new product introduction is risky and many new products fail (Ogawa and Piller 2006), retailers are reluctant to increase manufacturer`s value claimed and created immediately, before consistent results over longer time periods are generated. Additionally, while the increase in value claimed and created in response to prior new product success appears to be minor, decrease in value claimed and created can be far more severe. The results show that 122 the magnitude of change in value claimed and value created is greater when new product launches are unsuccessful than when they are successful. This suggests that retailers have a greater propensity to punish harmful behaviors than to reward beneficial behaviors (Fehr and Gachter 2000). However, since the effects over time have been explored without formally stating a priori hypotheses, in an initial attempt to assess effects over time, further research should be conducted. Managerial Implications The findings of this essay provide important implications for manufacturers. First, new product introductions should not be considered in isolation. Manufacturers need to realize that each new product is launched in the context of the past new product introductions and future expectations, and that the decisions that manufacturers have made regarding new products in the past will influence their returns from current new product launches. Frequently, manufacturers are shortsighted in their new product introductions. They may launch new products to neutralize competition, believing that a greater number of new products on the market will improve their returns. Or, they may increase new product innovativeness, seeking greater sales and profits (Abetti 2000, Chandy and Tellis 2000). All of these decisions, however, impact retailers’ motivation to reciprocate and therefore manufacturers’ returns from current new product launches. Second, manufacturers have recently realized that with the increasing power of retailers, focusing on consumers alone is not sufficient. To address this problem, manufacturers have begun to incorporate retailers’ criteria into the new product development process in order to increase channel acceptance (Luo 2007). These new product development efforts have, however, been treated in isolation from past and future new product developments. The results of this 123 essay suggest that manufacturers can strategically use their new products to build reciprocal relationships with retailers and improve returns on their future new product launches. Third, an important finding of this essay is that retailers reciprocate and respond to manufacturers’ actions and that reciprocity determines the way value is created and divided between the exchange partners. Unfortunately, retailers are more likely to reciprocate negatively than positively. Low success of prior new product introductions can have a particularly damaging effect on manufacturers’ returns. When this occurs, value claimed and created decrease, and this decrease is further magnified when a manufacturer launches a large number of new products or when these products are not innovative. Manufacturers can, however, minimize the negative impact of unsuccessful new product introductions on value claimed and created by launching fewer or more innovative new products. Alternatively, when a manufacturer launches successful new products, the returns from new product launches increase, and this increase is greater when the frequency of new product introduction is high and innovativeness of new products is low. Fourth, negotiation of favorable terms of trade for new products is one of the most important but also one of the most difficult tasks that manufacturers face. Sellers and category managers working for manufacturers understand this. They also understand how past new product introductions affect manufacturers’ ability to increase value claimed and created from new products. Therefore, it is important to communicate this information closely to innovation centers and new product development teams that can incorporate retailers’ criteria into their new product strategies. Doing so could improve the returns gained from new product launches. 124 Limitations and Future Research Although this essay provides insight into how new product launch decisions and new product success impact retailers’ willingness to reciprocate and manufacturers’ returns, this essay has several limitations that further research could address. First, although using a longitudinal experiment enhances the causal inferences, the cross-sectional nature of the survey instrument limits the determination of the direction of causality. Since reciprocity develops over time, and since the way value is shared and created is determined by past and future actions, further research should undertake a longitudinal study that would explore how manufacturers’ returns vary over time. Specifically, this study did not capture the effects of varying performance of new products over time. For example, how does a retailer respond to a manufacturer who may have introduced successful new products years ago, but who presently struggles to launch products that would exceed average performance? Second, the sample is restricted to retailers and limited to U.S. retailers. The behavior of retailers in other countries could differ, and this may threaten the generalizability of the results. Caution should therefore be used in generalizing these results to other economies. Additionally, the focus of this study is limited to the behaviors in manufacturer-retailer dyads. Other contextual factors such as the behavior of other retailers, competitors, and consumers are excluded. An investigation of the impact of these factors on the way value is generated and shared should provide a better understanding of retailers’ willingness to reciprocate. Third, the dependent variables (value claimed and value created) are general and abstract and may not fully capture the varying ways in which retailers reciprocate and respond to manufacturers’ new product introductions. For example, value can be divided between exchange partners in numerous ways. Manufacturers and retailers need to decide how to divide profit 125 margins, how to set prices, and what slotting fees to set; all of these things impact the division of generated value. Value can also be generated in multiple ways: by increasing new product support, providing better or greater shelf space, or by improving promotion and advertising. Inclusion of more specific dependent variables would enhance our understanding of how retailers respond to manufacturers’ new product offerings. Fourth, due to limited research in this area, there is potential for future research to expand the findings of this essay. Specifically, this essay focuses only on manufacturer- retailer dyads. Future research could include the impact of competition, consumers, or other retailers on the way value is shared and created between exchange partners. Additionally, since manufacturers can launch their new products via multiple channels (e.g., via other retailers, distributors, or online), future research could explore how the choice of these channels impacts how manufacturers create and claim value in their existing relationships with retailers. 126 APPENDICES 127 APPENDIX 1.1 Measures, Factor Loadings, and Composite Reliabilities for Study 1 Source New Product Innovativeness (Fang 2008) Market Uncertainty (Celly and Frazier 1996) Frequency of new product introduction Constructs Antecedents Relative to other products in the same category, how would you describe this new product? - Very ordinary for its category / Very novel for its category - Not creative / Creative - Uninteresting / Interesting - Not at all innovative / Very Innovative - Not challenging to existing ideas in its category / Challenging to existing ideas in its category - Not offering new ideas to its category / Offering new ideas to its category - Not capable of generating ideas for other products / Capable of generating ideas for other products Moderators Which term better describes your expectations about the market for this new product at the time of the launch - certain that selling efforts would pay off / uncertain whether selling efforts would pay off - sales forecasts were likely to be accurate / sales forecasts were likely to be inaccurate - confident of results of marketing actions / unsure of the results of marketing actions In this new product`s category, - the number of new SKUs created yearly by your company for this retailer is roughly - the total number of new products launched yearly by your company through this retailer is approximately -the average number of new products launched annually by your company for through this retailer is approximately 128 λ α/AVE 0.89/ 0.54 0.77 0.82 0.75 0.70 0.70 0.79 0.59 0.80/ 0.59 0.86 0.68 0.75 0.70 0.96 0.99 0.86/ 0.80 Source Constructs Outcomes Performance Ambiguity Contract Specificity With this retailer, - it is easy to monitor the retail performance of new products closely/ it is not possible to monitor the retail performance of new products closely - it is easy to assess the retail performance of new products/ it is not easy to assess the retail performance of new products - it is easy to obtain accurate new product performance evaluations: it is difficult to obtain accurate new product performance evaluation - At launch, the terms of trade for this new product were clearly contractually specified - At launch, the purchasing agreement as a whole was very specific - At launch, the contractual terms for this new product were very detailed and specific 2 Notes: χ = 222.149, d.f. = 144; CFI = .977; RMSEA = .064; SRMR = .055. 129 λ α/AVE 0.89 0.93/ 0.83 0.97 0.87 0.84 0.90 0.87 0.90/ 0.76 APPENDIX 1.2 Measures for the Control Variables for Study 1 Control Variables Transaction Size - The approximate initial monthly purchase amount (in US$) Relationship Length - The number of years that your organization has been in business with this retailer Contract Duration - The approximate duration of this signed agreement (months) Advertising - The total dollars spent advertising for this new product were higher than those spent on other, similar new products - This new product received greater advertising support (in $) than other similar new products 130 APPENDIX 1.3 Study 2 Scenario Descriptions Stage 1 Manipulation Now imagine the new product that you selected has been on the market for 3 months of the 12 month contract, and during this time, the new product consistently generated sales far above (below) the category average. Assume that you have a complete authority over the decision making. How you would treat this product over the next three months? [Insert measures for the dependent variables: relational behaviors] Stage 2 Manipulation Now imagine that another 3 months have passed. The new product has now been on the market for 6 months of the 12 month contract, and continues to generate sales far above (below) its category average. Given this information, how would you treat this product over the next three months? (again assume that you have a complete control over the decision making) [Insert measures for the dependent variables: relational behaviors] Stage 3 Manipulation Imagine that yet another 3 months have passed. The new product has now been on the market for 9 months of the 12 month contract and, contrary to the past, generated sales far above (below) its category average. Again, given this information, how would you treat this product over the next three months? (assume that you have a complete control over the decision making) [Insert measures for the dependent variables: relational behaviors] 131 APPENDIX 1.4 Measures, Factor Loadings, and Composite Reliabilities for Study 2 Source Outcomes Contract Specificity Solidarity Flexibility Information Exchange Relational Behaviors (Hoppner and Griffith 2011) At launch, - the terms of trade for this new product were clearly contractually specified - the purchasing agreement as a whole was very specific - the contractual terms for this new product were very detailed Please rate the extent to which you agree or disagree with the following statements: - You would try to help this supplier to manage this new product - You would be committed to improvements and changes for the benefit of this new product - You would treat this new product as a joint responsibility, rather than an individual responsibility λ α/AVE 0.89 0.92/ 0.81 0.92 0.87 0.74/ 0.51 0.74 0.68 0.72 - To benefit this new product, you would work around the contractual terms - You would be flexible in response to requests to work around the contract to support this new product 0.75 - You would keep this supplier informed about events and changes that may affect this new product - You would provide this supplier with information relevant to this new product more frequently and informally and not only according to pre-specified agreement 0.87 Solidarity 0.99 Flexibility 0.67 Information Exchange 0.90 2 Notes: χ = 59.684, d.f. = 32; CFI = .974; RMSEA = .066, SRMR = 0.049. 132 0.83/ 0.73 0.95 0.78/ 0.68 0.78 0.86/ 0.75 APPENDIX 1.5 Measures for Control Variables and Manipulation Checks for Study 2 Relationship Length Control Variables - The number of years that your organization has been in business with this supplier Transaction Size The approximate initial monthly purchase amount - In US$ - In units New Product Innovativeness Relative to other products in the same category, how would you describe this new product? - Very ordinary for its category / Very novel for its category - Not creative / Creative - Uninteresting / Interesting - Not at all innovative / Very Innovative - Not challenging to existing ideas in its category / Challenging to existing ideas in its category - Not offering new ideas to its category / Offering new ideas to its Category New Product Success Manipulation Checks In the fictitious scenarios given to you about the future performance of the new product: A. In the first three months since launch B. During 3-6 months since launch C. During 6-9 months since launch, - The success of this new product was - The performance of this new product was 133 APPENDIX 2.1 Measures, Factor Loadings, and Composite Reliabilities for Study 1 Source Constructs λ α/AVE Antecedents Prior New Product Success Kabadayi et al. (2007) The innovativeness of manufacturers`products (Fang 2008) Frequency of new product introduction Degree of Selectivity Fein and Anderson (1997) Contribution to sales Over the past three years, this manufacturer`s new products: - generated sales volume - generated sales revenue - performed relative to your sales targets - achieved a sales turnover 0.82 0.83 0.86 0.83 Contribution to profits - generated retail profits - contributed to the category profitability - generated total profits 0.81 0.84 0.86 Moderators Over the past 3 years this manufacturers product were (relative to other manufacturers) - Very ordinary for its category / Very novel for its category - Not creative / Creative - Uninteresting / Interesting - Not challenging existing ideas in the category / Challenging existing ideas in the category - Not offering new ideas to the category / Offering new ideas to the category Over the past three years, in the selected product category - The number of new products introduced by this manufacturer through your chain was - The frequency of new product introduction by this manufacturer through your chain was - The number of new SKUs created for this manufacturer was - The total number of new product introductions by this manufacturer for your chain was Over the past three years, this manufacturer - has given you an exclusive territory for its new products - has voluntarily refrained from carrying its new products through retailers that would compete with you 134 0.94/ 0.70 0.89/ 0.63 0.66 0.83 0.81 0.80 0.85 0.92/ 0.75 0.89 0.89 0.81 0.88 0.79 0.85 0.83/ 0.67 Source Constructs Value claimed and Value created λ α/AVE 0.95 0.76/ 0.64 Outcomes Relative to a manufacturer with average new product performance you now: Value Claimed by the Manufacturer - allow this manufacturer to capture greater share of the value on current new product launched - recognize this manufacturer by increasing its share of value generated from current new product launches Value created from new products - provide this manufacturer with opportunities to generate greater value from current new product launches - engage in activities that generate greater value for this manufacturer from new product launches - work with this manufacturer to generate greater value from current new product launches 2 Notes: χ = 382.50, d.f. = 215; CFI = .949; RMSEA = .0062; SRMR = .044. 135 0.62 0.77 0.85 0.67 0.82/ 0.59 APPENDIX 2.2 Measures for the Control Variables for Study 1 Advertising Dependence Control Variables - The total dollars spent by this manufacturer on advertising for its new products exceed competitors - This manufacturer spends substantial advertising dollars on new products Dependence of the manufacturer on the retailer - In your trade area, other retailers could provide this manufacturer with a comparable access to the market - This manufacturer would incur minimal costs in replacing you with another retailer - This manufacturer could easily replace the sales generated by you with sales ` from other retailers Dependence of the retailer on the manufacturer - Other manufacturers could supply you with similar products - You could manage the cost of switching to another manufacturer Interdependence a. Average scores for the dependence of the manufacturer on the retailer and dependence of the retailer on the manufacturer were calculated b. Interdependence scores was constructed by summing the manufacturer and retailer dependence scores (Kumar 1995) Asymmetry a. Average scores for the dependence of the manufacturer on the retailer and dependence of the retailer on the manufacturer were calculated b. Asymmetry score was constructed by taking the the difference between the manufacturer and retailer dependence scores (Kumar 1995) Relationship length - The number of years that your organization has done business with this manufacturer 136 APPENDIX 2.3 Study 2 Scenario Descriptions Stage 1 Manipulation Imagine that you are a buyer for a major retail chain and it is your job to develop product assortments, manage sales and margins, and negotiate contracts. Last year, Universal Company became your new supplier of branded products in the CPG industry. You were given complete authority over the decision making. The first year contract with Universal Company included standard terms of trade and retail margins that you give to all new suppliers. Over the past year, however, Universal Company launched multiple new products whose frequency of introduction, innovativeness, and advertising support were equivalent to the category average, but whose success was consistently far below (far above) the category average. In addition, you are one of multiple retail outlets in your trade area that carry Universal Company`s products (Universal Company recently granted you an exclusive territory in your trade area for its new products). Presently, Universal Company is planning to launch several new products in your category. Since contractual terms are renewed annually, it is your responsibility to determine and negotiate the new terms of trade and profit sharing conditions (that can significantly vary) for these new products. [Insert measures for the dependent variables: value claimed by the manufacturer and value created from new products] Stage 2 Manipulation Now imagine that another year has passed. Over this year, Universal Company launched additional new products whose frequency of introduction, innovativeness, and advertising 137 support remained equivalent to the category average and whose success remained consistently far below (far above) the category average. In addition, you continued to be one of multiple retail outlets in your trade area that carry Universal Company`s products (Universal Company continued to grant you an exclusive territory in your trade area for its new products). Again, since last year`s contract has expired, it is your responsibility to determine and negotiate the new terms of trade and profit sharing conditions (that can significantly vary) for new products that Universal Company is planning to presently launch. [Insert measures for the dependent variables: value claimed by the manufacturer and value created from new products] Stage 3 Manipulation Imagine that yet another year has passed. Over this year, Universal Company continued to launch additional new products whose frequency of introduction, innovativeness, and advertising support remained equivalent to the category average and whose success remained consistently far below (far above) the category average. In addition, you continued to be one of multiple retail outlets in your trade area that carry Universal Company`s products (Universal Company continued to grant you an exclusive territory in your trade area for its new products). Again, since the old contract has expired, it is your responsibility to determine and negotiate the new terms of trade and profit sharing conditions (that can significantly vary) for new products that Universal Company is planning to presently launch. [Insert measures for the dependent variables: value claimed by the manufacturer and value created from new products] 138 APPENDIX 2.4 Measures, Factor Loadings, and Composite Reliabilities for Study 2 Source Value claimed and value created Outcomes Relative to last year`s contract you would: Value Claimed by the Manufacturer - allow Universal Company to claim greater share of the value on current new product launched - increase Universal Company`s share of value on current new product launches - recognize Universal Company by increasing its share of value generated from current new product launches Value created from new products - provide Universal Company with opportunities to generate greater value from current new product launches - engage in activities that generate greater value for Universal Company from new product launches - work closer with Universal Company to generate greater value from current new product launches Indebtedness (Gouldner 1960) During the experiment, please rate the extent to which, - you felt obligated to repay Universal Company for its past new products - you felt that you owe Universal Company for its past new product introductions - you felt indebted to Universal Company because of its previous new product introductions - you felt indebted to Universal Company because it distributed its products through you - you felt obligated to repay Universal Company for selecting you as its retailer - you felt that you owe Universal Company for its decisions regarding distribution of new products through your competitors 139 λ α/ AVE 0.86/ 0.67 0.78 0.80 0.87 0.82 0.79/ 0.56 0.81 0.60 0.77 0.91 0.91 0.79 0.84 0.84 0.94/ 0.71 Source Outcomes Indebtednessbased reciprocal behaviors (Gouldner 1960) During the experiment, please rate the extent to which you changed (or refrained from changing) the conditions on Universal Company`s new product launches over the past three years: - based on your indebtedness to Universal Company for its past new product success - because you owed Universal Company for its past new product introductions - as a payback to Universal Company for its previous new product introductions - based on your indebtedness to Universal Company for selecting you as its retailer - because you owed Universal Company for its decisions regarding distribution of new products through your competitors - as a payback to Universal Company for distributing its products through you 2 Notes: χ = 176.42, d.f. = 129; CFI = .966; RMSEA = .052 140 λ α/ AVE 0.82/ 0.67 0.79 0.84 0.84 0.79 0.82 0.82 APPENDIX 2.5 Measures for Control Variables and Manipulation Checks for Study 2 Dependence of the Manufacturer on the Retailer Control Variables - It would be difficult to replace the sales and profits generated by Universal Company`s products - The cost of switching to another supplier could be substantial Prior New Product Success Under the given scenarios, - the performance of Universal Company`s new products was - the success of Universal Company`s new products was Degree of Selectivity Under the given scenarios, - Universal Company has given you an exclusive territory for its new products - Universal Company launched its new products only through your retail chain Initial Terms of Trade Under the given scenarios, - the first year contract included terms of trade set - the first year contract included retail margins that were Innovativeness Under the given scenarios, - the innovativeness of Universal Company`s new products was - the newness of Universal Company`s new products was Frequency of New Product Introduction Under the given scenarios, - the frequency of new product introduction was - the frequency with which Universal Company introduced its products was Advertising Under the given scenarios, - the advertising support for new products was - the amount of advertising for new products was 141 REFERENCES 142 REFERENCES Abetti, Pier C. (2000), “Critical Success Factors for Radical Technological Innovation: A Five Case Study”, Creativity and Innovation Management, 9 (December), 208- 221. Aiken, Leona S., Stephen G. West (1991), Multiple Regression: Testing and Interpreting Interactions. Sage Publications. Ailwawadi, Kusum L. and Bari Harlam (2004), “An Empirical Analysis of the Determinants of Retail Margins: The Role of Store-Brand Share,” 68 (January), 147-165. Anderson, Erin (1985), “The Salesperson as outside Agent or Employee: A Transaction Cost Analysis,” Marketing Science, 4 (Summer), 234-254. _______ and Barton Weitz (1992), “The Use of Pledges to Build and Sustain Commitment in Distribution Channels,” Journal of Marketing Research, 1 (1992), 18-34. Armstrong, J. Scott and Terry Overton (1977), “Estimating Nonresponse Bias in Mail Surveys,” Journal of Marketing Research, 51(July), 71-86. Atuahene-Gima, Kwaku (1995), “An Exploratory Analysis of the Impact of Market Orientation on New Product Performance,” Journal of Product Innovation Management, 12 (September), 275-293. Axelrod, Robert and William D. Hamilton (1981), “The Evolution of Cooperation,” Science, 211 (March), 1390-1396. Bagozzi, Richard P. (1975), “Marketing as Exchange,” Journal of Marketing, 39 (October), 3239. _______, Youjae Yi and Linn W. Phillips (1991), “Assessing Construct Validity in Organizational Research,” Administrative Science Quarterly, 36(3), 421-458. _______ (1995), “Reflections on Relationship Marketing in Consumer Markets,” Journal of the Academy of Marketing Science, 23 (4), 272-277. Bass, Frank (1969), “A New Product Growth For Model Consumer Durables,” Management Science, 15(5), 215-227. Becker, Lawrence C. (1986), Reciprocity. New York: Routledge & Kegan Paul. Bercovitz, Janet, Sandy D. Jap, and Jack A. Nickerson (2006), “The Antecedents and Performance Implications of Cooperative Exchange Norms,” Organization Science, 17 (6), 724– 40. 143 Berman, Barry (2010), “Products, Products Everywhere,” Wall Street Journal (accessed August 23, 2010), [available http://online.wsj.com]. Bollen, K. A. (1989), Structural equations with latent variables. New York: Wiley. Bosse, Douglas A., Robert A. Phillips, and Jeffrey S. Harrison (2009), “Stakeholders, Reciprocity, and Firm Performance,” Strategic Management Journal, 30 (4), 447-456. Bowersox, Donald J., Theodore P. Stank, and Patricia J. Daugherty (1999), “Lean Launch: Managing Product Introduction Risk Through Response-Based Logistics,” Journal of Product Innovation Management, 16 (Novermber), 557-568 Calantone, Roger and Mitzi Montoya-Weiss (1993), Product launch and follow-on. In: Managing New Technology Development. William E. Souder and J. Daniel Sherman (eds.). New York: McGraw-Hill, 1994, pp. 217-248. _______, Kwong Chan, and Ann S. Cui (2006), “Decomposing Product Innovativeness and Its Effects on New Product Success,” Journal of Product Innovation Management, 23 (5), 408-421. _______, Nukhet Harmancioglu, and Cornelia Droge (2010), “Inconclusive Innovation “Returns”: A Meta-Analysis of Research on Innovation in New Product Development,” Journal of Product Innovation Management, 27 (December), 1065-1081 Carson, Stephen J., Anoop Madhok, and Tao Wu (2006), “Uncertainty, Opportunism, and Governance: The effects of Volatility and Ambiguity on Formal and Relational Contracting,” Academy of Management Journal, 49 (5), 1058-1077 Celly, Kirti Sawhney and Gary L. Frazier (1996), “Outcome-Based and Behavior-Based Coordination Efforts in Channel Relationships,” Journal of Marketing Research, 33 (May), 200210. Chandy, Rajesh K. and Gerard J. Tellis (1998), “The Overlooked Role of Willingness to Cannibalize,” Journal of Marketing Research, 35 (Novermber), 474-487. _______ and _______ (2000), “The Incumbent`s Curse? Incumbency, Size, and Radical Product Innovation,” Journal of Marketing, 64 (July), 1-17 Connor, John M. (1981), “Food Product Proliferation: A Market Structure Analysis,” American Journal of Agricultural Economics, 63(November), 607-617. Cook, Karen S., Erin Rice (2006), “Social Exchange Theory,” Handbook of Social Psychology, Part 1, 53-76. 144 Cortina, Jose M., Gilad Chen, William P. Dunlap (2001), “Testing Interaction Effects in LISREL: Examination and Illustration of Available Procedures,” Organizational Research Methods, 4 (October), 324-360. Cropanzano Russell and Marie S. Mitchell (2005), “Social Exchange Theory: An Interdisciplinary Review,” Journal of Management, 31 (December), 874-900. Desiraju, Ramarao (2001), “New Product Introductions, Slotting Allowances, and Retailer Discretion,” Journal of Retailing, 77, 335-358. Di Benedetto, Anthony C. (1999), “Identifying the Key Success Factors in New Product Launch,” Journal of Product Innovation Management, 16 (November), 530-544. Dukes, Anthony J., Esther Gal-Or, and Kannan Srinivasan (2006), “Channel Bargaining with Retailer Asymmetry,” Journal of Marketing Research, 43 (February), 84-97. Dwyer, Rober F., Paul H. Schurr, and Sejo Oh (1987), “Developing Buyer-Seller Relationships,” Journal of Marketing, 51 (April), 11-27. Falk, Armin and Urs Fischbacher (2006), “A theory of reciprocity,” Games and Economic Behavior, 54, 293-315. Fang, Eric (Er), Robert W. Palmatier, Lisa K. Scheer, and Ning Li (2008), “Trust at Different Organizational Levels,” Journal of Marketing, 72 (March), 80-98. Fehr, Ernst and Simon Gachter (2000), “Fairness and Retaliation: The Economics of Reciprocity,” The Journal of Economic Perspecitves, 14 (Summer), 159-181. Fein, Adam J. and Erin Anderson (1997), “Patterns of Credible Commitments: Territory and Brand Selectivity in Industrial Distribution Channels,” The Journal of Marketing, 61 (April), 1934. Ferguson, Ronald J., Michele Paulin, and Jasmin Bergeron (2005), “Contractual Governance, Relational Governance, and the Performance of Interfirm Service Exchanges: The Influence of Boundary Spanner Closeness,” Journal of the Academy of Marketing Science, 33 (2), 217–34. Fornell C. and David F. Larcker (1981), “Structural Equation Models with Unobservable Variables and Measurement Error: Algebra and Statistics,” Journal of Marketing Research, 18(3), 382-388 Frazier, Gary L. and John O. Summers (1986), “Perceptions of Interfirm Power and Its Use within a Franchise Channel of Distribution,” Journal of Marketing Research, 23 (May), 169-176. ______ and Walfried M. Lassar (1996), “Determinants of Distribution Intensity,” The Journal of Marketing, 60 (October), 39-51. 145 Ganesan, Shankar, Steven P. Brown, Babu John Mariadoss, and Hillbun (Dixon) Ho (2010), “Buffering and Amplifying Effects of Relationship Commitment in Business-to-Business Relationships,” Journal of Marketing Research, (April), 361-373 Garcia, Rosanna and Roger Calantone (2002), “A critical look at technological innovation typology and innovativeness terminology: a literature review,” Journal of Product Innovation Management, 19 (March), 110-132 Gatignon, Hubert and Jean-Marc Xuereb (1997), “Strategic Orientation of the Firm and New Product Performance,” Journal of Marketing Research, 34 (February), 77-90. Ghosh, Mrinal and Gorge John (1999), “Governance Value Analysis and Marketing Strategy,” Journal of Marketing, 63 (Special Issue), 131-145. _______ and _______ (2005), “Strategic Fit in Industrial Alliances,” Journal of Marketing Research, 42 (August), 346-357. Gouldner, Alvin W. (1960), “The Norm of Reciprocity: A Preliminary Statement,” American Sociological Review, 25 (April), 161-178. Grover, Varun and Manoj K. Malhotra (2003), “Transaction cost framework in operations and supply chain management research: theory and measurement,” Journal of Operations Management, 21 (July), 457-473. Haisley Emily, George Loewenstein (2011), “It`s Not What You Get but What You Get It: The Effect of Gift Sequence on Deposit Balances,” Journal of Marketing Research, 48 (February), 103-115. Heide, Jan B. and George John (1988), “The Role of Dependence Balancing in Safeguarding Transaction-Specific Assets in Conventional Channels,” Journal of Marketing, 52 (January), 2035. ________ and Anne S. Miner (1992), “The Shadow of the Future: Effects of Anticipated Interaction and Frequency of Contact on Buyer-Seller Cooperation,” Academy of Management Journal, 35 (2), 265-291. ________ (1994), “Interorganizational Governance in Marketing Channels,” Journal of Marketing, 1 (January), 71-85. Hoch, Stephen J. and Shumeet Banerji (1993), “When Do Private Labels Succeed?,” Sloan Management Review, 34 (Summer), 57- 67. Hoppner, Jessica J. and David A. Griffith (2011), “The Role of Reciprocity in Clarifying the performance Payoff of Relational Behavior,” Journal of Marketing Research, 48(5), 920-928. 146 Houston, Franklin S. and Jule B. Gassenheimer (1987), “Marketing and Exchange,” Journal of Marketing, 4 (October), 3-18. Hu, L., and Bentler, P. M. (1999), “Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives”, Structural Equation Modeling, 6(1), 1-55. Hultink, Erik Jan, Abbie Griffin, Henry S.J. Robben, Susan Hart (1998), “In search of generic launch strategies for new products,” International Journal of Research in Marketing, 15, 269285. ________, Susan Hart, Henry S.J. Robben and Abbie Griffin (2000), “Launch Decisions and New Product Success: An Empirical Comparison of Consumer and Industrial Products,” Journal of Product Innovation Management, 17 (1), 5-23. Iyer, Ganesh and Miguel J. Villas-Boas (2003), “A Bargaining Theory of Distribution Channels,” Journal of Marketing Research, 40 (February), 80-100. Jap, Sandy D. and Shankar Ganesan (2000), “Control Mechanisms and the Relationship Life Cycle: Implications for Safeguarding Specific Investments and Developing Commitment,” Journal of Marketing Research, 37(May), 227-245. John George and Barton A. Weitz (1988), “Forward Integration into Distribution: An Empirical Test of Transaction Cost Analysis,” Journal of Law, Econocices & Organization, 2 (Autumn), 337-355. Joskow, Paul L. (1987), “Contract Duration and Relationship-Specific Investments: Empirical Evidence from Coal Markets,” The American Economic Review, 77 (March), 168-185. Kabadayi, Sertan, Nermin Eyuboglu, Gloria P. Thomas (2007), “The Performance Implications of Designing Multiple Channels to Fit with Strategy and Environment,” Journal of Marketing, 71 (October), 195-211. Katila, Riitta and Gautam Ahuja (2002), “Something Old, Something New: A Longitudinal Study of Search Behavior and New Product Introduction,” Academy of Management Journal, 45 (December), 1183-1194. Khermouch, Gerry, (1995), “Amid Rampant Product Proliferation, Beverage Marketers Desperately Seeking Simplicity,” Brandweek, 36.22 (May 29), 9. Kim, Stephen K., Richard G. McFarland, Soongi Kwon, Sanggi Son, and David A. Griffith (2011), Journal of Marketing Research, forthcoming. Kumar, Nirmalya, Lisa K. Scheer, and Jan-Benedict E.M. Steenkamp (1995), “The Effects of Perceived Interdependence on Dealer Attitudes,” Journal of Marketing Research, 32 (August), 348-356. 147 Larson Andrea (1992), “Network Dyads in Entrepreneurial Settings: A Study of Governance of Exchange Relationships,” Administrative Science Quarterly, 37, 76-104. Leventhal, G.S. (1969), “Equity, reciprocity, and reallocating rewards in the dyad,” Journal of Personality and Social Psychology, 13 (December), 300-305 Lindell, Michael K. and David J. Whitney (2001), “Accounting for Common Method Variance in Cross-Sectional Research Designs,” Journal of Applied Psychology, 86 (February), 114-121. Luo, Lan, P.K. Kannan, and Brian T. Ratchford (2007), “New Product Development Under Channel Acceptance,” Marketing Science, 26 (March-April), 149-163. Lusch, Robert F. and James R. Brown (1996), “Interdependency, Contracting, and Relational Behavior in Marketing Channels,” The Journal of Marketing, 60 (October), 19-38. McFarlin, Dean B. and Paul D. Sweeney (1992), “Distributive and Procedural Justice as Predictors of Satisfaction with Personal and Organizational Outcomes,” The Academy of Management Journal, 35(3), 626-637. Montgomery, David B. (1975), “New Product Distribution: An Analysis of Supermarket Buyer Decisions,” Journal of Marketing Research, 12(August), 255-264. Mooi, Erik A. and Mrinal Ghosh (2010), “Contract Specificity and Its Performance Implications,” Journal of Marketing, 74 (March), 105-120. Ogawa, Susumu and Frand T. Piller (2006), “Reducing the Risks of New Product Development,” MIT Sloan Management Review, (Winter) 65-71. Palmatier, Robert W., Cheryl Burke Jarvis, Jennifer R. Bechkoff, and Frank R. Kardes (2009), “The Role of Customer Gratitude in Relationship Marketing,” Journal of Marketing, 73 (September), 1-18. Parkhe, Arvind (1993), “Strategic Alliance Structuring: A Game Theoretic and Transaction Cost Examination of Interfirm Cooperation,” Academy of Management Journal, 36 (August), 794829. Pervan, Simon J., Liliana L. Bove and Lester W. Johnson (2009), “Reciprocity as a Key Stabilizing Norm of Interpersonal Marketing Relationships: Scale Development and Validation,” Industrial Marketing Management, 38, 60-70. Ping, Robert A., Jr. (1995), “A Parsimonious Estimating Technique for Interaction and Quadratic Latent Variables,” Journal of Marketing Research, 32 (August), 336–47. ——— (2007), “Second-Order Latent Variables: Interactions, Specifications, Estimation, and an Example,” in Marketing Theory and Applications, Vol. 18, Andrea L. Dixon and Karen A. Machleit, eds. Chicago: American Marketing Association, 286–93. 148 Poppo Laura and Tod Zenger (2002), “Do Formal Contracts and Relational Governance Function as Substitutes or Complements?,” Strategic Management Journal, 23, 707-725. Quelch, John A., David Kenny (1994), “Extend Profits, Not Product Lines,” Harvard Business Review, (September-October), 153-160. Raho, Akshay R. and Humaira Mahi (2003), “The Price of Launching a New Product: Empirical evidence on Factors Affecting the Relative Magnitude of Slotting Allowances,” Marketing Science, 22 (Spring), 246-268. Rindfleisch, Aric and Jan B. Heide (1997), “Transaction Cost Analysis: Past, Present, and Future Applications,” Journal of Marketing, 61 (October), 30-54. Rokkan, Aksel I., Jan B. Heide, and Kenneth H. Wathne (2003), “Specific Investments in Marketing Relationships: Expropriation and Bonding Effects,” Journal of Marketing Research, 40 (May), 210-224. Ryu, Gangseog and Lawrence Feick (2007), “A Penny for Your Thoughts: Referral Reward Programs and Referral Likelihood,” Journal of Marketing, 71 (January), 84-94. Samaha, Stephen A., Robert W. Palmatier, and Rajiv P. Dant (2011), “Poisoning Relationships: Perceived Unfairness in Channels of Distribution,” Journal of Marketing, 75 (May), 99-117. Sanchez, Juan I. and Petra Brock (1996), “Outcomes of Perceived Discrimination among Hispanic Employees: Is Diversity Management a Luxury or a Necessity?,” Academy of Management Journal, 39(3), 704-719. Song, Michael X. and Mark E. Parry (1996), “What Separates Japanese New Product Winners from Losers,” Journal of Product Innovation Management, 17 (September), 422-439. Sorescu, Alina B., Rajesh K. Chandy, Jaideep C. Prabhu (2003), “Sources and Financial Consequence of Radical Innovation: Insights from Pharmaceuticals,” Journal of Marketing, 67 (October), 82-102. ________ and Jelena Spanjol (2008), “Innovation`s Effect on Firm Value and Risk: Insights from Consumer Packaged Goods,” Journal of Marketing, 72(March), 114-132. Stump, Rodney L. and Jan B. Heide (1996), “Controlling Supplier Opportunism in Industrial Relationship,” Journal of Marketing Research, 33 (November), 431-441. Sullivan, Mary W. (1997), “Slotting Allowances and the Market for New Products,” Journal of Law and Economics, 40 (October), 461-493. Tesser, Abraham, Robert Gatewood, and Michael Driver (1968), “Some Determinants of Gratitude,” Journal of Personality and Social Psychology, 9 (3), 233-236. 149 Thomassen, Lars, Keith Lincoln and Anthony Aconis (2006), Retailization: Brand Survival in the Age of Retailer Power, Kogan Page Limited, London and Philadelphia. Wathne, Kenneth H. and Jan B. Heide (2000), “Opportunism in Interfirm relationships: Forms, Outcomes, and Solutions,” Journal of Marketing, 64 (October), 36-51. Wathne, Kenneth H. and Jan B. Heide (2004), “Relationship Governance in a Supply Chain Network,” Journal of Marketing, 68 (January), 73-89. ________ and Jan B. Heide (2000), “Opportunism in Interfirm Relationships: Forms, Outcomes, and Solutions,” Journal of Marketing, 64(4), 36-51. Williamson, Oliver E. (1979), “Transaction-Cost Economics: The Governance of Contractual Relations,” Journal of Law and Economics, 233-261. _____ (1996), The Mechanisms of Governance. New York: Oxford University Press. Wutys, Stefan, Stefan stremersch, Christophe Van Den Bulte and Philip Hans Franses (2004), “Vertical Marketing Systems for Complex Products: A Triadic Perspective,” Journal of Marketing Research, 11 (November), 479-487. Young, Louise C. and Ian F. Wilkinson (1989), “The Role of Trust and Co-operation in Marketing Channels: A Preliminary Study,” European Journal of Marketing, 23(2), 109-122. Zheng, Jurong, Jens K. Roehrich, Michael A. Lewis (2008), “The Dynamics of Contractual and Relational Governance: Evidence from Long-Term Public-Private Procurement Arrangements,” Journal of Purchasing & Supply Management, 14, 43-54. Zhou, Kevin Zheng, Chi Kin (Bennett) Yim, and David K. Tse (2005), “The Effects of Strategic Orientations on Technology – and Market-Based Breakthrough Innovations,” 69 (April), 42-60. 150