LIBRARY Miningan State University This is to certify that the thesis entitled CARBONATED SOFT DRINK DEMAND: ARE NEW PRODUCT INTRODUCTION STRATEGIES A VIABLE APPROACH TO INDUSTRY LONGEVITY presented by Marcus A. Coleman has been accepted towards fulfillment of the requirements for the MS degree in Agriculture Economics Major Professor’s Signature Y/4/fl52 Date MSU is an Affinnative Action/Equal Opportunity Employer PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 5/08 K:IProj/Acc&PreleIRC/DatoDm.indd CARBONATED SOFT DRINK DEMAND: ARE NEW PRODUCT INTRODUCTION STRATEGIES A VIABLE APPROACH TO INDUSTRY LONGEVITY By Marcus A. Coleman A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Agriculture Economics 2009 ABSTRACT CARBONATED SOFT DRINK DEMAND: ARE NEW PRODUCT INTRODUCTION STRATEGIES A VIABLE APPROACH TO INDUSTRY LONGEVITY By: Marcus A. Coleman In an industry dominated by multiple product introductions differentiated at the attribute level, carbonated soft drinks (CSDS) experience demand pressure from all aspects of the beverage industry that go beyond CSDS. The main objective of this paper is to analyze demand for new and sector leading CSDS, which are characterized by multiple product consumer purchasing behavior, firm promotional activity and differentiation at the attribute level. Given the many unique strategies for innovation in CSD new product introductions (NPIS), it is imperative to find out just how effective firm innovation strategies are in using NPIS to stimulate and revitalize demand for CSDS. Using the linear approximate version of the almost ideal demand system that incorporates product attributes through distance metrics, the results of this study Show how consumers react to price increases in both NPIS and sector leading CSDS. The combination of the information gained from both the own-price and cross-price elasticity results as well as the attribute results indicate the relative instability in demand found across the CSD industry, particularly for NPIS. Despite the instability, the results also provide information for product attribute categories where strategies can be formulated to aide in improving the longevity of the CSD industry. ACKNOWLEDGEMENTS First of all I would like to give thanks my lord and savior Jesus Christ, for without him, none of this would be possible. I would also like to extend my deepest gratitude to my thesis advisor, Dr. Dave Weatherspoon. Thank you for all your mentoring and assistance. I would also like to extend my gratitude to my other thesis committee members, Dr. Geoffrey Pofahl and Dr. John Hoehn. I enjoyed working with you and I appreciate your advice and guidance. To my grandmother, Mary Coleman, and mother, Diana Coleman, thanks for being so understanding and teaching me how to make the best out of my life. I also want to thank you for helping me keep my spirits up and for being so supportive. iii TABLE OF CONTENTS CHAPTER 1: INTRODUCTION ...... --1 1.1 Overview .................................................................................................................. 1 1.2 Objectives ................................................................................................................. 4 1.3 Organization of Research ....................................................................................... 5 CHAPTER 2: REVIEW OF LITERATURE 6 2.1 Overview .................................................................................................................. 6 2.2 Competition and Strategy .................................................................................... 11 2.3 Consumer Adoption .............................................................................................. 16 2.4 Demand Studies ..................................................................................................... 17 2.5 Traditional Demand Theory ................................................................................ 18 2.6 Almost Ideal Demand System .............................................................................. 20 2.7 Distance Metrics .................................................................................................... 22 CHAPTER 3: METHODOLOGY - - - ............... 24 3.1 Overview ................................................................................................................ 24 3.2 Demand Model ...................................................................................................... 24 3.3 Distance Metrics .................................................................................................... 25 3.4 Distance Metrics Applied to LA-AIDS ............................................................... 27 3.5 Uncompensated Price Elasticities ........................................................................ 28 3.6 Data ........................................................................................................................ 29 CHAPTER 4: RESULTS 32 4.1 Model Summary Statistics ................................................................................... 32 4.2 Uncompensated Own-Price Results .................................................................... 32 4.3 Uncompensated Cross-Price Results ................................................................... 35 4.5 Regional Effects on Demand ................................................................................ 42 4.6 Results Summary .................................................................................................. 45 CHAPTER 5: CONCLUSION - 47 5.1 Summary ................................................................................................................ 47 5.2 Future Research .................................................................................................... 49 APPENDIX 1: CSD HISTORY 50 APPENDIX 2: TABLES 51 APPENDIX 3: FIGURES 76 REFERENCES 78 iv CHAPTER 1: INTRODUCTION 1.1 Overview The growth of the carbonated soft drink (CSD) industry fi'om 1960 to 1990 was remarkable and greatly outpaced the increase in population growth during the 30 year period. Per capita consumption of soft drinks increased 2.5 times over the 30 year period (Muris et a1., 1993). Even though the industry saw undaunted grth for nearly 30 years, its recent decline has concerned industry analysts. Dube (2005) reported that in 1998 CSDs accounted for 49 percent of total US. beverage gallons consumed per capita, per year. Recently, per capita consumption of CSDs has decreased while that of bottled water and other beverages, such as functional beverages, have steadily increased.1 Sustainability of demand is at the forefront of the discussion of industry longevity for CSDS. Without constant innovation, there would presumably be less consumer incentive to make new purchases of CSDS, especially with the increasing number of healthier substitutes. With competitive pressures from all aspects of the beverage industry, CSDs now battle competitors in which industry analysts would have never 2 This competition brings to question the point of whether the imagined 50 years ago. focus on new product introductions (NPIS) in the CSD industry will foster industry longevity. With new product failure and discontinuation rates of nearly 80 percent across the food and beverage industry, studying demand for new CSDs is an imperative process in formulating innovation strategies to foster future industry success. Beverage World l Functional beverages are those beverages that offer consumers additional benefits from their attributes, i.e. mineral enhancement. 2 The beverage industry includes all drink products, including CSDS. 1 (November 15, 2003) reported approximately 1,235 total new beverage introductions in 2002, where approximately 250 survived by late 2003, which is consistent with a food and beverage industry failure rate of approximately 80 percent. The CSD industry has seen its fair share of product failures and discontinuations throughout the years. A significant failure occurred in 1985 when the introduction of New Coke (the reformulation of Coca-Cola’s flagship product) was deemed one of the riskiest product introductions in CSD history (Greising, 1998). New Coke turned out to be a huge product failure and the original formula was immediate redistributed. The introduction of New Coke proved that without consumer acceptance, new products are doomed from their inception and may be detrimental to a firm’s ability to effectively market to its consumer base. With the product failure rate in the grocery business being close to 88 percent (Remilia, 2000), innovation strategists, marketers and retailers must play equal roles in ensuring effective strategy formulation. With high failure rates, NPIS can have highly negative effects on CSD firms’ ability to appeal to consumers. Conner (1980) offered several ways in which product proliferation can be detrimental. These included: (1) deception from imitation or variants being marketed by all firms, (2) wasteful advertising, and (3) overwhelming consumers with a large number of introductions. Probably the most harmful of these three is flooding the market with a large number of new products. With the large number of CSDS, along with other beverages, being introduced annually, consumer decision making becomes a more strenuous process. Retailers play an important role in heavily influencing the success of NPIS. Luo et al. (2007) explained that in channel structures characterized by powerful retailers, the dominant retailers’ acceptance of a manufacturer’s new product often determines the success of the new offering. For CSDS, it may be more of a fight for shelf space between branded products and private labels. Limited shelf space in most supermarkets and mass merchandisers makes marketing new CSDs a difficult task. The overall role of CSD NPIS is to stimulate demand in the beverage industry, which is characterized by consumers that are beginning to seek more in their beverage acquisitions than just thirst quenching prowess. Buzzell and Nourse (1967) classified NPIS according to their degree of novelty: (1) distinctly new products, (2) brand proliferation/line extensions, or (3) item proliferation, repositioning, or reformulation. For CSDS, introductions stem fi‘om the addition or subtraction of attributes to an existing formulation. Chaney et a1. (1991) believed that innovative behavior is the engine of economic growth and development. Given this school of thought, innovation is a necessary component to keep CSDs active in the product life cycle.3 The CSD industry’s success for over 100 years can partly be attributed to innovation in flavoring, extravagant marketing campaigns and constant adherence to consumer fads.4 The Encyclopedia of Global Industries (EGI) (2007) stated that the global soft drink industry is almost exclusively a marketing phenomenon.5 Given that the actual product is a simple blend of water, sweeteners, flavors and other additives, the EGI (2007) stated that the industry’s genius lies in convincing billions of consumers to drink soft drinks instead of plain water or other beverages. 3 The product life cycle is defined as the stages that an individual product goes through that eventually lead to elimination from the market (product death). The stages include introduction, growth, maturity and decline. 4 Consumer fads can be defined as health trends or society’s acceptance of a particular type of beverage. 5 The term “soft drinks” represents both carbonated and non-carbonated soft drinks. 3 Understanding the impact that NPIS have on the CSD industry at the attribute level will aid in the study of the demand relationships between industry leading CSDS, new products and the constant pressures from the multiple substitutes that are available throughout the beverage industry. Demand relationships among CSD products must be considered to determine if NPIS provide a viable outlet to stimulate demand, both in the short-run and long-run and to determine if they provide a viable strategy for industry longevity. Given the many unique innovation strategies for NPIS exhibited by CSD firms, it is important to find out just how effective these products are in both stimulating and revitalizing demand for CSDS. Innovations are the basis for the future of the CSD industry. With copious CSD introductions annually, well-defined innovation strategies are imperative in marketing a large number of differentiated products to a very diverse group of consumers. 1.2 Objectives The objective of this paper is to analyze consumer demand using a linear demand system which incorporates product attributes into the estimation of the system for new CSDS and sector leading CSDS (which are characterized by multiple product consumer purchasing behavior, firm promotional activity and differentiation at the attribute level). The specific objective of this paper is to examine the substitution effects between NPIS and CSD sector leaders, which possess similar attributes, on a national basis, making full use of the attributes that each product possesses to determine if innovation strategies among CSD firms are a viable approach to industry longevity. 1.3 Organization of Research Chapter I presented the introduction, justification, problem, and the objectives. Chapter II examines relevant literature and gives background information on the study. Chapter III presents the models and data used in the study. The study’s results are presented in Chapter IV. Finally, Chapter V summarizes the results and advances the study’s conclusions. CHAPTER 2: REVIEW OF LITERATURE 2.1 Overview The EGI (2007) stated that the term soft drink was coined to distinguish flavored drinks from hard liquor and were originally designed as a substitute for liquor in an effort 6 For a historical perspective of flavor innovations in to reduce alcohol consumption. CSDS, see Appendix 1. Today, the CSD industry is distinguished by differentiation at the attribute level. Caves (1992) defined product differentiation as distinguishing the differences of a product that set it apart from its main competitors and makes it more attractive to target consumers. Anderson (2008) noted that the key for all of the beverage companies is differentiation. Whatever the strategy, be it a new color, flavor, or formula, CSD firms will strive to create the greatest brand and product awareness with hopes of crowding out their competitors. Lancaster (1990) stated that the degree of product variety increases with the competitiveness of the market. Flavor enhancements, adjustments in calorie content and additions in functional features now lead CSD innovations.7 The US. CSD industry is highly concentrated and characterized by healthy competition. Mintel’s May 2008 Carbonated Drinks Report showed that three companies dominate the CSD market.8 For food stores, drug stores and mass merchandisers (including Wal-Mart) [FDMw], Coca-Cola is the market leader with over 38 percent 6 Soft drinks are non-alcoholic beverages, both carbonated and non-carbonated, containing natural or artificial sweetening agents, natural or artificial flavors, and other ingredients. 7 The addition of functional features to CSDS includes product enhancements such as adding minerals and other healthy alternatives. 8 Mintel provides food and drink research across the world. Mintel GNPD monitors product innovation and retail success in the consumer packaged goods market worldwide. market share followed by PepsiCo with approximately 32 percent Share of the market. Cadbury Schweppes is the third major player in the CSD market and holds 20 percent share of the F DMw market. These three firms dominate the CSD industry in brand recognition and innovative clout. Some of the key historical success factors for the longevity of the CSD industry include: constant product innovation, organizational size and established brand loyalty. The EGI (2007) reported that in the early 20008, the global soft drink industry was dominated by Coca-Cola and PepsiCo at an unprecedented level never seen in international business. Table 1 (Appendix 2) gives recent brand share information for CSDS. The table is broken down by brand and individual products for the top three CSD firms. According to Table l, in 2007 Coca-Cola held 34 percent, Pepsi 32.1 percent and Cadbury Schweppes 21 percent of the food stores, drug stores and mass merchandiser (FDM) market for regular CSDS. Also according to Table 1, Coca-Cola held 45.7 percent, Pepsi 32.8 percent and Cadbury Schweppes 17.4 percent of the FDM market for diet CSDS. Table 2 (Appendix 2) gives current and forecasted CSD sales. According to Table 2, FDMw sales of CSDS are expected to decrease by $867 million from 2002 to 2012. Anderson (2008) stated that as the industry-wide soft drink fight has reached its maturity point, the industry’s giants have begun to rely on new products and non- carbonated beverages for sales growth. This trend has been observed over the past 20 years. Increased awareness and concern about health and dieting, changing consumer tastes and demographics, and increased competition from other beverage products are some of the main drivers causing stagnation in the CSD market. Other changes include: globalization, changing societal concerns, attitudes, lifestyles, and evolving buyer preferences. With new technologies virtually making the world smaller, CSD firms are now creating flavors with a vibrant flare to meet the changing and sophisticated demands of consumers worldwide. CSD firms are innovating to quench consumers’ never-ending thirst for new and innovative products. CSD firms are now adding exuberant flavors to meet the sophisticated demands of consumers. Worldwide there are a few key social changes that are driving demand for specialty foods and beverages. These changes include: a growing consumer interest in world flavors and cuisines; consumers with an increasing drive for indulgence; consumers with more sophisticated palates; consumers with the ability to afford premium products; and consumers’ desire to enjoy premium products on a regular basis (Packaged Facts, 2005). Purcell (2006) reported that today’s consumers between the ages of 25-54 now possess a more sophisticated and adventurous taste for foods and beverages which draws them to some Specialty food items.9 With key changes in consumers’ beverage demand, there are many indicators of consumers’ desire for more bold and stimulating flavors. Table 3 (Appendix 2) reveals that cherry, lemon and orange were the most frequent flavor introductions between 2001 and 2006. Mintel’s May 2008 market report on carbonated drinks claimed that cherry was the top new flavor in CSDS in 2006. They reported that Coca—Cola’s Black Cherry Vanilla Coke, including its diet version, attracted FDM sales of $81 million its inaugural year of 2006. They also reported that the appeal of lemon and lime soda declined among consumers as Coke with Lime, along with its diet counterpart, and Pepsi with Lime, along with its diet counterpart, saw declining sales. During the years 2005 and 2006, 9 The term Specialty food item encompasses new and innovative CSDS. 8 sales of Diet Coke with Lime declined 32 percent, to $77 million, while Diet Pepsi with Lime sales declined nearly 11 percent, to $30 million in the same period (Mintel Group, Carbonated Drinks — US. May 2008). This shows that the combination of multiple flavors, i.e. cherry and vanilla, is becoming more appealing to consumers. As shown in Table 4 (Appendix 2), low sugar, low calorie and natural introductions lead innovative product introductions other than flavor enhancements. ‘0 Table 5 (Appendix 2) gives new CSD introductions by company. According to Table 5, Coca—Cola and Pepsi lead the industry in innovations. Dr. Pepper (a Cadbury Schweppes brand) also had 36 new introductions between 2002 and 2005, which is when the company started its innovative flavor introductions, such as Berries and Cream. Beverage World (November 15, 2003) reported approximately 1,235 total new beverage introductions in 2002, where approximately 250 survived by late 2003. This report is consistent with an industry failure rate of approximately 80 percent. Consumer Reports (December 1, 2005) counted 28 different kinds of Coke and Pepsi on the market, with approximately half being diet. With multiple branded products and NPIS simultaneously entering the CSD market, not only are these introductions creating competition in the beverage industry as a whole, but also are causing cannibalization within CSD brands. This type of competition is seen in Pepsi with products such as Pepsi Cola, Pepsi Vanilla, Pepsi with Lime and Pepsi Summer Mix simultaneously being offered. With high numbers of yearly CSD introductions, it is difficult to determine which products will survive. In Mintel’s examination period (2001-06) for their April 2007 report on carbonated drinks, they found that FDM sales of Diet Coke with Lime, Diet ‘0 Natural CSDS are those that are free from artificial ingredients such as flavors, sweeteners and colors. Pepsi with Lime, and Mountain Dew Live Wire all peaked in the first year and saw sales decline thereafter. Additionally, they reported that Diet Cherry Vanilla Dr. Pepper, Cherry Vanilla Dr. Pepper and Coke with Lime all peaked the second year after their introduction and then saw constant declines. They also indicated that relatively mature brand extensions, such as Mountain Dew Code Red and Diet Pepsi Twist, saw constant declines from 2001 to 2006, which consequently is after their initial introduction. These trends show that new products generally see their highest sales directly after introduction and constant declines thereafter. The sales increases show consumer willingness to try new products but the decline shows the products’ lack of sustainability in consumer tastes and preferences. These sales trends may be an indication that the acquisition of these new or repositioned CSDS was merely trail in nature. This behavior may be an indication of the future success, or lack of success, these innovation strategies may potentially yield. Reduced calorie CSDS have also seen a rise in popularity with increasing trends for healthier beverage and food products found across the US. Mintel’s 2008 report on carbonated drinks reported that regular carbonated beverage sales declined more in FDMw’s than reduced calorie sodas. They stated that sales dropped by more than 15 percent from 2002 to the time of the report and expected declining sales of approximately 2 percent per year for the next five years following the report. These trends are reinforced in Figure 1 (Appendix 3). Part of the increasing trend for reduced calorie CSDS may also be explained by the number of teens now preferring reduced calorie products. Table 6 (Appendix 2) shows teen preference for a number of reduced calorie CSDS that were introduced between 2001 and 2007. Traditional reduced calorie CSDS (i.e. Diet Coke and Diet Pepsi) were the most preferred among teens but the increasing 10 presence of flavor enhanced reduced calorie CSDS, i.e. Diet Coke with Lime, is getting noticeable attention from this same demographic. Another factor in increasing demand for reduced calorie CSDS could be the narrowing distinction between the taste of regular and reduced calorie CSDS. 2.2 Competition and Strategy The EGI (2007) stated that while branded products are at the heart of the soft drink industry, private label soft drinks garner a significant share of many of the world’s markets. The Cott Corporation and its private labels present themselves as strong competitors for the top CSD firms. Private label products typically compete on price and use imitation as a competitive strategy. In 2000, private labels captured nearly 14 percent of CSD volume in the US. and approximately 7 percent of sales (EGI, 2007). According to Business Week (March 21, 2005), the US. market for private label goods stabilized at about 16 percent. Private label manufacturers being able to control 10 to 20 percent of the market presents a significant concern to the top CSD firms. Private label manufacturers typically enjoy free rider effects from industry leaders.11 Imitation is often one of the largest drivers of competition for CSDS. D’Aveni (1994) gave the notion that by imitating an innovators’ action, a firrn’s rivals can enjoy the free-rider effects by sharing in the profits, or reduce the competitive advantage granted to the innovator or both. Imitation is one of the most noticeable competitive strategies for CSD firms. It is up to sector leaders to constantly innovate to sustain their competitive edge, but is this innovation a worthwhile strategy? It is the purpose of this study to provide results that will shed light on this issue. 11 Free rider effects come in the context of private label firms enjoying the innovative technologies that larger firms finance, hence mimicking their product introductions. ll Table 7 (Appendix 2) gives an indication of the constant pressures CSDS face in NPIS from other beverage industries. Table 7 shows that the ready-to-drink (RTD) juice and bottled water industries lead the competitive charge against CSDS, having approximately 1,947 and 524 NPIS, respectively, between 2002 and 2008. Mintel’s May 2008 market report on carbonated drinks speculated that demand for all CSDS will continue to fall as Americans pursue healthier products. They reported that regular CSD consumption dropped notably between 2003 and 2007, with approximately 5.5 million consumers halting consumption. With this trend in CSDS, they also reported that bottled water gained 19 million consumers between 2003 and 2007, indicating the product’s increasing presence in the market and consumers’ ultimate acceptance of a product even though it is readily available to them, un-bottled, in their households. Recent trends in per capita consumption of beverages are shown in Figure 2 (Appendix 3). Figure 2 indicates that per capita consumption of CSDS decreased by 4.1 gallons while that of bottled water increased by 9.8 gallons between 2000 and 2006. With the growing demand for healthier, firnctional beverages, CSD firms are now relying on beverages such as juices and flavored waters to diversify their product portfolios. The Encyclopedia of Emerging Industries (EEI) (2007) reported that the top three beverage industries are soft drinks, bottled water and fruit juices/drinks, respectively. Dubbed the “new aged” drinks, premium bottled beverages have met the new millennium as a product with immense potential, drawing consmners’ taste buds away from soft drinks and alcoholic beverages (EEI, 2007).12 According to the Beverage Marketing Corporation (2002), US. per capita consumption of “new age” beverages ‘2 New aged beverages are classified as such because they offer more benefits to meet the ever changing demands of consumers than beverages of the past. 12 increased 134 percent from 1994 to 2002, reaching approximately 16.9 gallons per year. In wholesale dollars, total revenues for new aged beverages reached $11.6 billion in 2002. The Beverage Marketing Corporation (February 10, 2005) stated that flavored waters’ share of sales could bring in over $800 million by 2009 on the high-end forecast, or nearly $600 million in the medium-growth forecast. Functional beverages in the US. have evolved beyond the niche category of health and wellness drinks. Beverages such as sports drinks, RTD tea and bottled waters have added a new dimension to this market with an increased emphasis on convenience, novelty, fim and image. There has also been an increasing trend towards juices and juice drinks enriched with herbs, botanicals and nutraceuticals (Sorenson & Bogue (2006), Weisberg (2001)). Hasler (2000) noted that key factors driving the interest in functional food include the growing self-care movement, changes in food regulations and overwhelming scientific evidence highlighting the critical link between diet and health. Even though functional beverages have entered the market in a very impactful way, one key feature that has seemingly been neglected in some products is taste and flavoring. In some cases this has deterred consumer acceptance of some functional drinks, which may be a positive demand booster for CSDS (Sorenson & Bogue (2006), Cavallo (2000), Cosgrove (2004), F oote (2002)). Just as flavoring is important for CSDS, the EEI (2007) stated that formulation and flavoring are ongoing preoccupations of premium beverage companies. With growing demand for vitamins, minerals, and other health-related products, the challenge for flavoring companies trying to follow this trend is to provide beverage companies with nutritionally sound, yet tasty products. New functional beverage introductions have steadily increased since 2000. Table 8 (Appendix 13 2) shows the recent new functional beverage introductions. Table 8 indicates that weight control and vitamin enhanced introductions lead the innovative charge against CSDS for functional beverages. This category is evidence of the health trends in beverage consumption in the US. With more healthier, nutritional substitutes available, CSD innovation strategies are even more imperative to make industry longevity possible. Figure 3 (Appendix 3) shows actual and predicted functional beverage sales as forecasted by Mintel’s August 2007 market report on functional beverages. The same report predicted FDM sales of functional beverages to double between 2002 and 2012 (as shown in Figure 3). Mintel’s forecasted sales of carbonated beverages and other non-alcoholic beverages are shown in Table 9 (Appendix 2). Table 9 shows that bottled water, sports and energy drinks, coffee/RTD coffee and tea/RTD tea are all expected to increase in sales up to 2012 whereas CSDS and juice/juice drinks are expected to experience sales declines. The EGI (2007) stated that some industry analysts believe that the traditional concept of equating soft drinks primarily with carbonated beverages, particularly colas, must be revised to reflect the growing popularity of other RTD beverages, such as teas, coffees, herbal beverages, juices, and sports and energy drinks. Table 10 (Appendix 2) exhibits Coca-Cola’s and PepsiCo’s diversification into other RTD markets. Diversifying into these markets has also added another group of competitors for these traditional CSD firms including: Proctor & Gamble (U.S.); Danone (France); Nestle Beverages (Switzerland); and Unilever (England). For Coca-Cola and PepsiCo, these product diversifications have put extra pressure on their CSD lines. Given this increased pressure, impeccable innovation strategies are imperative. l4 Given other market pressures, it is even more important for CSD firms to continue to innovate in flavoring and other CSD attributes to foster industry longevity. Another way Coca-Cola has managed to add to its CSD consumer base is by aggressively marketing some of its reduced calorie CSDS to men. Products such as Coke Zero and Diet Coke Plus help to progress this trend. PepsiCo has not only diversified its portfolio to include stimulating, hydrating, and invigorating products, but it has also introduced cutting-edge products like Tava, Pepsi’s first premium soft drink, and Pepsi Raw, which squarely align with emerging consumer health trends. Offering limited edition products has enabled marketers and strategists to capitalize on seasonal trends without making long-term commitments. PepsiCo has been very strategic in their innovation strategies in this arena. For example, in 2004 the company introduced Mountain Dew Pitch Black for 10 weeks, which had a black grape flavor. In 2005, the company introduced Mountain Dew Pitch Black II, with sour black grape flavoring, which followed the same trend as Mountain Dew Pitch Black. Both of these campaigns ran for 10 weeks leading up to Halloween in their respective years (Mintel Group, Carbonated Drinks — US. May 2008). Other seasonal products from Pepsi included Holiday Spice and Sierra Mist Cranberry Splash, which were Christmas promotions (Mintel Group, Carbonated Drinks -— US. May 2008). Limited edition and seasonal product introductions offer a different flare from that of traditional CSD marketing and innovation. This type of strategy is seen in other industries (i.e. candy and snack cakes) and has proven to be successful in stimulating short-term demand for products themed around a given season or holiday. For CSDS, this is a new territory with immense potential. 15 Some CSD firms are now making an effort to use consumer feedback in formulating new products. In late 2007, Mountain Dew launched a consumer driven campaign aimed at an NPI. Terrned DEWmocracy, online participants were allowed to determine the flavor, color, name, logo and label of a product that they create. Participants eventually would vote on their favorite of many combinations, and three were selected. The selected products, termed Supemovam, Voltage"M and RevolutionTM were put up for a national vote. Participants then promoted their favorite drink to friends using branded campaign tools that are available through the DEWmocracy website and may be used through various social networks. The selected drink will be Mountain Dew’s newest product. Mountain Dew states that the purpose of this campaign is to allow consumers to help create the next Mountain Dew line extension. Their goal is to provide loyal supporters a rich, involving, online experience that serves to bring the community closer together by way of taking them on a journey fi'om deep in a mythic world all the way back to their store shelves (PepsiCo, DEWmocracy). 2.3 Consumer Adoption An important aspect in determining the sustainability of demand for a given product(s) is consumer adoption. Table 11 (Appendix 2) gives some reasons as to why consumers are trying new beverages. From this table it can be inferred that consumers are trying new beverages because of innovative labeling/bottling and flavor enhancements of a preferred brand. Tables 12 and 13 (Appendix 2) both give an indication as to what types of products consumer are trying by demographic. These tables show that women and people between the ages of 18 to 24 are typically more apt to try new, cutting edge beverages. It may be inferred from the information presented in 16 these tables that young women are more inclined to try flavor enhanced beverage than any other consumer demographic. 2.4 Demand Studies Demand studies that apply an array of empirical models to examine retail level market data across a variety of food and beverage industries are a valuable primer in evaluating the overall demand landscape of the CSD product category. The studies here offer different approaches to examine a common interest in demand analysis across all categories. The intuition gained from these studies offers insight for a study of CSD NPIS, particularly as it relates to attribute effects and differences. Draganska and Jain (2005) and Kim et a1. (2002) used yogurt as a product category for study. The differentiation found in yogurt products is comparable to that of CSDS. Draganska and Jain (2005) used retail-level scanner data from the yogurt category and a consumer choice model to determine the effect that the number of variants in a product line has on the selection of a product line. Special attention was given to flavor possibilities and the consumer decision was derived from a utility maximization model. To incorporate consumer heterogeneity, a discrete-choice random coefficients model was used. Kim et al. (2002) proposed a demand model for the yogurt industry based on a translated additive utility structure. Using purchasing data of different varieties of yogurt, the model nested the linear utility structure, while allowing for the possibility of a mixture of comer and interior solutions where more than one but not all varieties are selected. The authors found that some households purchased mostly or exclusively one variety and highly valued popular flavors. They also found that there would be 17 substantial utility loss fiom the removal of popular flavors and heavy compensation would be required for the removal of preferred varieties. In the beverage industry, each category is differentiated at the attribute level. The next few studies exhibit different approaches to demand analysis in various beverage industries. Brown et al. (1994) used weekly retail-level data on juice products. They used the Rotterdam Model and the Wu-Hausman test to examine the possibility of endogeneity of total juice expenditure in conditional demand specifications for individual juices. Xiao et al. (1998) used the Rotterdam model to evaluate patterns in non-alcoholic drink demand. Time-series data encompassing consumption, pricing, and advertising for fluid milk, fruit juices, soft drinks, coffee and tea were used to complete the study. The results showed that the major factor governing the increase in per capita soft drink consumption was structural change. This was found to be the dominant pattern for the last 25-30 years. 2.5 Traditional Demand Theory A common approach to analyzing demand as it relates to product characteristics lies within Lancaster’s demand model. Ratchford (1975) summarized Lancaster’s demand model in a mathematical form which is shown in Equation 1; (1) Max U(z)subjec,,o px SK with z = Bx, where 2 represents a vector of characteristics, p represents a vector of prices, K represents income and B is an (r x n) matrix which transformed the n goods into r characteristics. Goods x are transformed into characteristics 2 through the relation (2 = Bx). Matrix B represents consumption technology. Ratchford (1975) gave an overview of Lancaster’s demand model. This model states that utility is derived from the properties or characteristics which goods possess 18 rather than the goods themselves, as opposed to traditional theory which is not inclusive of product characteristics. Berry (1994) and Anderson et a1. (1992) found empirical and theoretical evidence of this, respectively. In a general sense, traditional demand theory can refer to the analogy of consumer choice under a budget constraint and the consequent production of the change in a consumer’s chosen collection of goods when prices change (Lancaster, 1971). The models outlined by Lancaster (1966, 1971 and 1991) all take demand theory beyond the traditional sense by being inclusive of the rich information that comes from product characteristics. He argued that goods do not give utility, but that consumers derive utility from the characteristics that goods possess. In other words, consumers buy goods based on the attributes they offer. Preference is a function of attributes and must be defined in terms of properties of the good itself, i.e. calories, sugar and flavor. Ratchford (1975) explained several conditions under which Lancaster models are useful. The model also explains the role of price in determining the demand for differentiated products. In the case of the CSD products studied here, the point of differentiation will be in flavor and calorie content. The model also provides a framework for estimating the sensitivity of demand to relative price of a brand. The model also provides a theoretical perspective for brand share determination and gives an economic explanation for the theory of brand loyalty. Given the implications obtained from Lancaster’s work, a model that explicitly takes into consideration product characteristics is beneficial for CSDS. Dube (2004), Berry (1994) and Fader and Hardie (1996) all used a Lancaster based approach to model product alternatives in terms of their underlying product attributes. Random coefficients 19 for product attributes allow for flexible substitution patterns. Dube (2004) assumed that if consumers have a preference for a product, they will tend to substitute the product with similar products. Chan (2006) also found evidence of this. If consumers have a preference for reduced calorie soft drinks, they will potentially substitute these with other reduced calorie soft drinks. Pofahl (2008) stated that new product valuation is inherently dependent on the estimation of substitution patterns between similar products in a category. One of the most difficult processes in traditional consumer theory is the introduction of a new product. The new products used in this study were introduced between 2001 and 2005. The introduction of a new product here simply means the addition or subtraction of one or more attributes to the existing formula. Lancaster (1991) expressed that if a new good possesses characteristics in the same proportion as some existing good that it will simply fail to sell if its price is too high or will completely replace the old good if its price is sufficiently low. This also brings imitation and competing flavors to the forefront. Additionally a model is needed that can be transformed to explain substitutability and complementarity in products. In determining the sustainability in demand of CSD products, explaining cross price relationships between NPIS and industry leaders is vital. 2.6 Almost Ideal Demand System In the arena of demand studies, a number of models exhibit desirable properties according to demand theory. Some of these models include the Rotterdam model, logit demand, linear demand, log-linear demand and the almost ideal demand system (AIDS). The AIDS of Deaton and Muellbauer (1980b) is probably one of the most widely used 20 demand Specifications. According to Deaton and Muellbauer, it is derived as a first-order approximation to any demand function resulting from an individual’s utility maximization. Wilson (1994) stated that an important feature of the AIDS is that the expenditure levels are allowed to impact the distribution of shares. For a study dealing with differentiated products or any products at the market level (particularly with scanner data), the AIDS has proven to be a successful empirical tool in examining such data. Larue et a1. (1991) stated that the AIDS has convenient properties such as exact aggregation and being a first-order approximation to any demand system and has been used in many demand studies of beverage products (they studied alcoholic beverages). Several studies made use of the AIDS or its linear counterpart, the linear approximate (LA) AIDS, include Dhar et al. (2003), Cotterill and Putsis (2000), Carew et al. (2004), Cotterill and Samson (2002), Scale et al. (2002) and Larue et al. (1991). Cotterill and Putsis (2000) and Cotterill and Samson (2002) both applied the LA-AIDS approach to Information Resources Inc. (IRI) market-level data. All other studies made use of IRI market-level data or some other form of scanner data. The only exceptions are Seale et al. (2002) and Larue et al. (1991) which used import data and sales summary report data, respectively. Of these studies, Dhar et a1. (2003) used CSD data in its empirical analysis and Carew et al. (2004), Seale et al. (2002), and Larue et al. (1991) made use of wine data. Alston et al. (1994) stated that even though the AIDS possesses many properties desired in an empirical demand study, it sometimes is difficult to estimate. To simplify, Deaton and Muellbauer (1980b) suggested using the linear approximate version of the AIDS. Cotterill and Putsis (2000) gave a number of reasons why the LA-AIDS is 21 preferable to other demand analyses and functional form specifications. They stated that it is derived from the underlying choice axioms in utility theory. This is where individual behavior can be aggregated to consistently estimate demand parameters fi'om market- level data and it gives a first-order approximation to any “true” demand system functional form. The model is also sufficiently flexible so as not to unduly constrain channel behavior and market power. Alston et al. (1994) stated that the LA-AIDS is, in general, not an integral demand system, but its widespread popularity appears to be based on the fact that it is comparatively easy to estimate, combined with the belief that it is a reasonably good approximation of the true AIDS. For a study of products differentiated at the attribute level, a flexible model should provide an unbiased parameter estimator of demand and elasticities. The LA-AIDS will be used to provide the empirical analysis of the study. 2.7 Distance Metrics For a study of CSDS, a model is needed that is prudent enough in parameter space to handle a large number of differentiated products and also incorporates attributes into the space. The Distance Metrics (DM) approach of Pinkse et al (2002), Pinkse and Slade (2004), Pofahl and Richards (2008) and Pofahl (2008) will enable this study to deal with the differentiation in attributes of CSDS and also a large ntunber of products. It incorporates observable attributes that are important in consumer purchase decisions for soft drinks. In this approach, attribute differences are considered an important driver of consumer demand. The DM approach makes use of the important information found in product attributes and provides a measurable way to estimate their effects on demand. Comparative to other attribute models, the DM approach captures the notion that 22 proximity in attribute space increases the competition between products. With its attribute proximity approach, the DM approach can be used with a representative demand system. Whereas the work of Lancaster was theoretical in nature, the DM approach incorporated into the LA-AIDS will provide an empirical approach to make use of the valuable information found in product attributes. The DM approach, when applied to a representative demand system, allows a large number of differentiated products to be considered. It shows the important role of attribute proximity in determining the competitive relationships among differentiated products. It also reflects the intuition that products possessing similar characteristics compete on price much more than those that are dissimilar. Pofahl and Richards (2008) stated that the fundamental insight of the DM approach is that each product in a category can be viewed as a unique combination of characteristics and that substitution patterns between those products is to be determined by their relative proximity within the multi- dimensional characteristic space. Estimation can then be carried out using standard econometric techniques. 23 CHAPTER 3: METHODOLOGY 3.1 Overview The LA-AIDS is best suited to analyze the retail-level data used in this study. In addition to the work of Cotterill and Putsis (2000) and Alston et a1. (1994) presented in Chapter 2, Deaton and Muellbauer (1980a) stated that the LA-AIDS has the desirable aggregation properties and is a preferred functional form for analyzing market level data. Given the flexible properties of the LA-AIDS, product attributes will be incorporated into Deaton and Muellbauer’s LA-AIDS through distance metrics (Pinkse et al (2002), Pinkse and Slade (2004), Pofahl and Richards (2008), Pofahl (2008)). The incorporation of product attributes allows the analysis of product differentiation in the CSD industry. 3.2 Demand Model Following Pofahl and Richards (2008) and Pofahl (2008), the DM approach will be applied to Deaton and Muellbauer’s (1980a, 1980b) LA—AHDS. The LA-AIDS is presented in Equation 2; N X, (2) wit =al. +jE17y’ln(pjt)+'Biln 1? , where a, [3 and y are parameters, is (1,...,N) is an index of products, te (1,...,T) is a time index, p . t = ( p1 t ,..., p M) is a vector of retail prices, q t = (ql. t ) is a vector 1 ,...,th of product quantities demanded, X t = 21. pl. t ql. t is total expenditure in time t, p . q . t t . . . . . . . . it = 4J— rs expenditure share for product r In time t, lnPt IS a price Index and x t 24 X [4] is the real expenditure level. Here, lnPt * is the log-linear analogue of the * P: Laspeyres price index, which is similar to Stone’s price index of Green and Alston ( 1990). Stone’s price index is used as an empirical approximation to a theoretical translog price index (Moschini, 1995). Typically studies that use the LA—AIDS incorporate Stone’s Price Index but Moschini (1995) cautioned the use of Stone’s price index due to invariance in choice of units of measurement for prices and quantities. This is the primary reason Laspeyres price index is used here. According to Capps et al. (2003) the use of Pt * simplifies the estimation of the demand system. Moschini ( 1995) results showed that the AIDS and LA-AIDS virtually yield the same results. 3.3 Distance Metrics Given that the empirical application of this study that includes 26 products, estimation of the system may be problematic due to low degrees of fi’eedom. To reduce the dimensionality of estimation, product attributes are introduced into the LA-AIDS in a way that reduces the overall parameter dimensions of this model. Distance metrics can either be represented as a discrete or continuous variable. For continuous metrics, the Euclidian distance is used to measure closeness of two products in attribute space. As adopted from Pinkse and Slade (2004), Pofahl and Richards (2008) and Pofahl (2008), Equation 3 expresses the euclidian distance method; —1 L 2 (3)d = 1+2 (Li—Ll.) , 25 where L,- represents product A 1 and Lj represents product A 2. The euclidian distance is mathematically expressed as‘/(Li -Lj)2 . The euclidian distance method can also be written as an aggregate of all attributes as expressed in Equation 4; (14%” = 1 (4) , ”J 1+2\/(Li—Lj)2+(Mi—Mj)2+(Nl.+Nj)2 where L,- = product A 1, Lj = product A 2, M = product B1, M,- = product 32, N,- = product C1 and N} = product C2.l3 Discrete metrics are typically represented by a (0, 1) scale, which equals 1 if a product contains a certain attribute and 0 otherwise. A generalized representation of both continuous and discrete distance metrics is given by Equation 5; Dill (5)g(d;r) = 2101” = mgmrdC). m where I is an indicator firnction, either 1 or 0 for a discrete measure. The discrete metric equals 1 if two products share the same attribute. The variable d0 is a compound discrete measure, where d can equate to m=1 ...D* different values. The variable dc is a vector of continuous metrics, i.e. calorie content. The function used to replace all cross price parameters in the original demand system is g(dl.j;}.k). Here, g() is some function of dij , a vector of distance metrics. The variable A is a vector of parameters corresponding to each distance metric. The variable g() is chosen by the researcher and is a linear function of several discrete and continuous attributes. Pinkse et al. (2002) recommended the use of a semi-parametric technique such as a series expansion method ‘3 The problem with the specification in Equation 4 is that while the effects of each individual attribute can be captured, they cannot be separated from one another. 26 in selecting the specification of g() Pofahl (2008) used a linear function of the discrete and continuous attributes because the author found that the specification of Pinkse et al. (2002) was insensitive to a wide array of choices, which is the case with CSDS. 3.4 Distance Metrics Applied to LA-AIDS As referenced from Pofahl (2008), given the number of products included in the empirical model, estimation of the original LA-AIDS could be problematic from 3 degrees of freedom standpoint. The DM approach reduces the dimensionality of demand estimation. Without imposing any theoretical demand restrictions on the LA-AIDS, N(N+2) parameters would normally be estimated. Imposing symmetry, homogeneity, and adding up reduces the number of parameters to N(N+3)/2-I parameters. Distance metrics reduces the number of parameters to 3N+K parameters (K is the number of distance metrics), assuming that g(-) is specified as a linear function of distance metrics.l4 This is done by modeling the LA-AIDS cross price coefficient (71.1.) as a function of different distance measures between product i and j. The application of the DM approach to the LA-AIDS is mathematically represented by Equation 6; N X (6)w =a +r,.,.1n

EEO .388 as: as. sea 35 .98: 85> sea 85 .3968 5&8 .5 88> 8.26 8.5 828$ .85 use: a: A .308 as: as, 38 $5 .608 8:28 a? £8 85 an. 888.3 588.3 883 8880.3 5883 883 553 .8383 amok 828. 3888.- L858. 888. 838. 8:2. 88:. 888 .3883 $383 5088. 88.83 $883 $883 $883 348:3 «55> 888 888. 888. 288. 82m. 888. 888 8483 is— 58883 8883 2:83 .8883 5:83 :283 5:83 52:3 2: 88:8. 828 888. 8:8. :88. 888 882 88: 5:83 5883 588 $883 @883 8883 8883 E83 850 888. 888. 5.8. 888. 888. 888 838. 882. 8883 5:83 8:83 :823 28.83 38883 3883 $823 2:: 8:8 888. 888 882. 88%. 948 888 882 Esau.— 3888. 3.283 S883 8823 5883 5:83 3833 $883 4.5 888 888 :88. :58. 8:8 228 888- 882. F8883 38203 583 :883 E884 @883 8884 488% >3 8888 :88 288 888. 8:8». 828 8:2; 882. £883 3883 8.83 883 :883 $4883 $883 583 have 838 888. 3.8. 9.88. 888 888 888 4883 5883 88283 5:83 883 €883 €883 8823 E83 BEE. 288 888. 88. 888 828. 8:8 888 882 3.5 :5 22mm 8.5 .5: NO Boa 2:: mo: 8.5 3:6 E 2.5 @3530 2: no 82:55 «30.3.6 :4. E 83.96:— outm we 88.3— 95 .5.“ 8:53— bfisua—H 93.5-89.0 3N «San. 32:5...— co—Eaam .550 we toe—EEOC 63 308 83 .85 2.8 35 .808 «828 as, 3.8 35 8:5 85:8 : £26— 5885 2 .8 .m ._ 05 8 5:355me “oz .. 8888.3 ”828.3 8888.8 5288.8 8888.3 .8884 8883 888.3 25 8888. 888. 888. 82.8. 888. 888. 85.8.- 888. 8.... 8883 8883 8883 888.5 8888.3 8888.3 8883 8888.3 25 8888. 888. 888. 888. 288. 828. 888. 888. .52 .5 8888.3 .8883 8888.3 888.3 8888.3 888.3 88:: 8888.3 .5 h .5 288. 888. 288. 888. £88.- 888. 888. 888. 8888.5 8883 8883 8888.3 :883 58$ 8883 8883 .885 888. 8888. 828. 888. 888. 288. 888. 888. .5 .5 588% :88; 828.8 ”888.5 :88: 58% 8883 3883 was. 888. 828. 888. $88. 828. 8888. 888.- 88:. .5 8888.3 888.3 888.3 88883 8883 8888.3 8883 8828.3 35 888. 888. .828. 888. 888. 8:8. 828. 23.2. 55:88 882% 8888.3 88.85 8888.3 8888.3 8888.3 8883 8823 .88.. 888. 8888. 8.88.- 888. 8888. 888. :88. 882. .5 8883 8888.3 E8884 H8883 ”888.3 88.84 8883 888.3 3:. 888. 888. 8:88. 828. 828. 888. 888.- 882. 3.5 85 3:8 2.5 .55 "-0 28 2:: m5 3.5 3.5 E 3.5 82:5...— uounaum .350 no 52:382— 3552.0 05 a: 82:5...— a_oU-aucU =< 5 8238:— ootm ac 58.5m— 05 be 8:83— »:535 035.895 "30:55:03 5 25¢. 64 Table 22: Cross-Price Elasticity Results for the Effects of Price Increases in Reduced Calorie Pepsi Products on the Quantity Demanded of Reduced Calorie Pepsi Products Diet Mtn. DPV DPL DPT Diet Pepsi Dew Pepsi One .509999 -.515199 -.594285 -.037259 -.033148 -.282834 DPV 1.266103] 1.043427] 1.075550] 1.0031286] 1.0062383] 1.024499] -.734344 .099055“ -.594646 -.037299 -.032899 -.282570 DPL 1.061887] 1.203014] 1.075552] 1.0031299] 1.0062395] 1.024495] -.306697 -.215069 -1.45039 -.015568 -.033007 -.072650 DPT 1.039091] 1.027432] 1.295713] 1.0019765] 1.006233] [.012668 -.811691 -.563259 -.779911 -1 .39073 -.022l42 -.228020 Diet Pepsi 1.068321] 1.047902] 1.088747] 1.055335] 1.007033] 1.028671] Diet Mtn. -.207705 -.143747 -.408564 -.011266 -1.46201 -.022019* Dew 1.039150] 1.0274721 1.075837] 1.002088] 1.084183] 1.014387] -.66321 1 -.464776 -.458327 -.024903 -.007224* -1.06230 Pepsi One 1.056757] 1.039829] 1.062394] 1.002964] 1.004672] 1.111904] * Not significant at the 1, 5, or 10 percent levels 1] Standard Errors Diet Pepsi Vanilla (DPV), Diet Pepsi with Lime (DPL), Diet Pepsi Twist (DPT) Table 23: Cross-Price Elasticity Results for Price Increases in Regular Pepsi Products on the Quantity Demanded of Regular Pepsi Products Pepsi w/ Pepsi Mountain Lime Vanilla Pepsi Dew Pepsi w/ -2.5 1204 .248463 .0044307 .0017824“ Lime 1.427515] 1.069123] 1.0009136] 1.0016662] Pepsi .174947 -1.78136 .001376 -.010784 Vanilla 1.048675 1.273457] 1.0006556] [00210051 -.487276 -.093101* -1.5391 1 .002876“ Pepsi 1.0810813 1.092664] 1.046979] 1.004698] Mountain .047455“ -.532307 -.0087036 -1.56973 Dew 1.05 8395] 1.102571] 1.002080] 1.057183] * Not significant at the 1, 5, or 10 percent levels [] Standard Errors 65 Table 24: Cross-Price Elasticity Results for Price Increases in Regular Pepsi Products on the Quantity Demanded of Reduced Calorie Pepsi Products Pepsi w/ Pepsi Mountain Lime Vanilla Pepsi Dew .200353 .044852“ .001646 .0019390“ DPV 1.049942] 1.115834] 1.0006847] 1.001691] .360553 -.183363 .0014294 .0019524* DPL 1.085782] 1.067923] 1.0006935] 1.001696] .193403 -1.29466 .0015650 0019915" DPT 1.049757] 1.123666] 1.000671] 1.001690] .027329“ -.423721 -.005725 .0025317“ Diet Pepsi 1.059080] 1086859] 1.00240] 1.002767] Diet Mtn. .068051“ .068875“ -.002307 .009542 Dew 1.057377] 1.081082] 1.001022] 1.002605] Pepsi .039156“ -.411728 .003336 .0019711* One [.05 6857] 1.084359] 1.001 1961 1.001702] * Not significant at the 1, 5, or 10 percent levels 1] Standard Errors Diet Pepsi Vanilla (DPV), Diet Pepsi with Lime (DPL), Diet Pepsi Twist (DPT) 66 3092:: as, 3855. .55 s .3: «85$ 26. 89 8:2 was, £835. .559 828$ .5 2:5; .926 85 .Amzpat Pom van: .5 8 A28 25 if 35 .388 as: a? :88 35 .989 85> if 35 1 E25 @8985— _ 238.“ 8883 88983 2%83 88883 @888 $383 5883 83mm: 8833 15a 8882. $28. SMS. 288. 8.88. 83.3. 823.8. 388. 88:8. 88% :883 :883 88883 5283 $5.83 8882: 8:83 2383 888:: 3833 N6 3.5 3&2. $88. 288. 8E8. 888. W888. 88%. 82.8. 82.5. 282. :23: 88883 $883 5883 58.83 88883 $833 #883 8823 8283 Boa 3w: RES 888. @88. 888. 8.83.. 883. «$88. $82 822. 3.5 88:: 82an 82.83 78883 8883 55:: S832: 8883 58:: $883 .56 $92 888. E28. 388. 8:88 £83 8888 :88. 25mm. 382. 888:: $3383 5883 3883 £82: 582: @833 8833 89.8.3 888. 2:5 888. 288. $388.- 88:8 8888 88% 3.83. 238. 288.- 282- $2.5 8882 5883 5883 8283 $883 $883 Romvvi 8883 ”853 39.83 mg $22 888 888. 288. £28. 8:: $38 288. 882. $32. 58:: 82883 5:83 83.88.“ :Eooo._ 88:3 :883 8832: 5383 :323 new: $22 888 288. £88. 2388. 33% 8:8. 83% 8:8. :82. 52:: 88883 83:83 8383 8883 $883 8893 $28; 888:: £884 55:: £83. 888. 8:88. 288. 888.- $3: 888 833. 838. 838. 25 if E: if :5 in any. «55> En 3.3 8: >3 .5: 35 535.32 7.95 \B Each bun-“=0 2: .5 8935.5 83m =< 5 823.5:— 935 be vacuum 2: .8.— 3-33— .3853:— outmameuv "mm 035. 32595 woe-Eam— 350 he ave—.5599 67 .Pmeaséa 855 . 5.5 2:: £8 a 835 .9583898835 885 255% _ _ 20>“: «:88 S 5 .w .3 05 8 anE 8 “oz _. 58:3 8883 38281.8 @883 88838 _ 883 388.3 88:83 8883 888.3 .5 NS 2 _. 888. 888. 888. £88. 888. 882. 888. 2 88. 88:8. 8 :5 888.8 8:883 8883 82.83 583 88:3 8883 $883 8883 $883 ammo: 882. $88.- 888. 888. 888. $88. 28:. 828. 228. 8:8. .5 .5 888.3 8888.3 8888.3 8888.3 8888.3 :883 5:5 #:813 8888.3 :883 3.5 885. :88. :88. 888. 8888.- 88:. 888. 888. 882. :2 8. :5 88:3 E883 89483 58183 3883 58:3 :883 :88: $883 :883 sum».— 88:. 888.- 888. 888. 8:8. 888. 888. 8:2. :88. :88. .5 W823 : 88.83 5:83 8883 E: 83 8883 8888.: 8883 S 883 8883 .5 h 3.82.. 888. 888.- 888. 88:8- :88. 888. 888. 888:- 8:8,- 8823 $883 $8883 8883 88883 ”888: 3.88.3 5.88.3 8223 :883 atmm 8:8: 888. 888. :88. 8888.- 8888. 828. 8888. :88. :88. 8883 :82 :3 8883 :283 :883 882 3 8883 88:88 8883 8:83 3.5 2 :8. 88.8. 2.88. 888. 888.- 88:. 8:8. 888. 888. 8:2. 25 if :5 amon— 65 >:5 WE «85> 55 “a: and >5 .52 3F— ESEEE finen— 8» Each 82695 cow-3.5 .850 no vow—EEK— @3530 2: .8 82695 Each =< E 823.6:— 329 he Buy—um 2: :8 3:83— .Eocufim aura—-820 "Aug-.353 mm 03:. 68 Table 26: Cross-Price Elasticity Results for the Effect of Price Increases in Reduced Calorie Cadbury Schweppes Products on the Quantity Demanded of Reduced Calorie Cadbury Schweppes Products Diet Dr. DCVDP Pepper Diet 7 Up -1.03378 -.011645 -.045864 DCVDP [.183550] [.0039568] [.0097846] Diet Dr. -.046579 -1.45188 -.046842 Pepper [.015811] [.063452] [.010003] -.077052 -.021588 -1.89258 Diet 7 Up [.016013] 1.004194] [.076989] [ ] Standard Errors Diet Cherry Vanilla Dr. Pepper (DCVDP) Table 27: Cross-Price Elasticity Results for the Effect of Price Increases in Regular Cadbury Schweppes Products on the Quantity Demanded of Regular Cadbury Schweppes Products 7UPMB CVDP 7 Up Dr. Pepper -2.48851 .050022 .011733 .0025486 7UPMB [.498436] [.02243 5] [.003632] 1.001218] .17501 1 -1 .29840 -.024940 -.0073775 CVDP [.079136] [.300865] [.005098] [.001816] .266186 -.175189 -2.04088 -.00829 7 Up [.089537] [.035156] [.106243] [.001989] Dr. .130531* -.146013 -.017298 -1.45920 Pepper [.092834] [03634;] [.007003] [.067576] * Not significant at the 1, 5, or 10 percent levels [] Standard Errors 7 Up Mixed Berry (7UPMB), Cherry Vanilla Dr. Pepper (CVDP) 69 Table 28: Cross-Price Elasticity Results for the Effect of Price Increases in Regular Cadbury Schweppes Products on the Quantity Demanded of Reduced Calorie Cadbury Schweppes Products Dr. 7UPMB CVDP 7 Up Pepper .150540 .159016 .0085278 .0062219 DCVDP [.078219] [.042874] [.003348] [.0013926] Diet Dr. .114032“ .075846 .009730 -.018657 Pepper [.080855] 1.025612] [.003896] [.002959] .329324 .035244“ .025366 .001441“ Diet 7 Up [.0882271 [.0224341 [.006377] [.001226] * Not significant at the 1, 5, or 10 percent levels [] Standard Errors Diet Cherry Vanilla Dr. Pepper (DCVDP), 7 Up Mixed Berry (7UPMB), Cherry Vanilla Dr. Pgaper (CVDP) 7O Goa 2:: .23 3.8 35 fine as; if 35 .388 85 a? if .05 .§..B «ass as... :5 .608 «82% £8 £8 35 some.“ .5 85> 826 .8968 3&8 .5 «ass E6 85 .5282; ram 382 n5 8 efim 85.3w 2 Gm_m8._ 4888.8 888% @283 $883 5883 5:83 gun 888. 888. 888. 888. 888m. 888. 883 58:: 8883 88883 5883 $883 :82: 8833 «-0 3.5 $88.- $88. 888. 888. 888. 883:. 88: H8384 @883 ”8:84 8883 588% 533 5883 .5: «888. R88. 888. 828. 88:. 2.88. :88. 8383 $883 @838; 5:83 8884 88:: $883 93 $88. 888. 828. 888. 888. 882. .88; 3.5 :8888 83:83 H883 $283 588. :22: $883 «a: 888. $88. 888. 888. 822. 883. 888. >580.— .8884 $883 :38; 23.83 2883 82:3 8:83 4.5 83.8. :28. 888. 888.- :83. 888. 2:8. @883 5888.8 H888; H8883 $883 8823 H833 2:: .88. ~28. 288. 888.- 2.88. 88:. 883 E 3.6 8883 8888.“ H8283 H8383 W883 822% H8884 >3 8.28. 888. 888. 888.- 88:. :88. 3:88. 5.88% 888% 5283 R883 H888; $888.8 8884 won 888. 888. $88. 888. 82.2. 28:. 3.83 a: b 3.: .280.— 388 .5 b ~86 ~95: as; .535 .5 3933.:— uocaaum .859 ,.o Benn—:0: @3530 .8 825?:— moanotaum bin—EU =< 5 883.55 93.:— uc 88.3— 2: .Su 3:53— 3652mm «3.5-89.0 "an 033—. 71 8&8 .5 «88> 826 .8558 580.. .5 8.53 82.0 .05 .525: E5 882 .5 8 .85 .588 2 888: 3.88: 828: 5888: :88: 888.3 5888: 25 .88.. 8:8. :88. 88:8. 888.- 88.... 88:. 888. E88: 888: 588: 5888: 58:8: 883.: 8:8: 55 888. 8:8.- 888- 888. 888. 888. 88... .5285 3.88: 388.: 88.8: $888: 8888: 3.8:: 888: .89.. .05 :88. 888 R88. :88.- 888. 888. 888. $888 $88.: .8888. 588: 5:8: 58:: 5.8:: 3.535 :88. 888. 888. 888. 888. 8.8:. 888. :888: $888: 2.8.8.: 5.88: 88.8: 88:: 5888.: E5 8...... 888. 888. 88.8. 888.- 882. 88... 53532 .888: 8:8: 8:8: .8888: $88: 88:8: 88:1 8.8 88:. 828. :88. 888. 888. .82.. 8.8.. 5888: 8.88.: 888: 588:: 588: 5888: H882: .80.. 8:2. 888. 888. 888. 888. 88:. :88.- 4888: @888: 58.8: 283.: 58884 5888: 88:: 3.5 88:. 888. 888. 888. 888. 8:8. .88.. 5888.: $888: 88.8: 588: 888: 5888: 88:: 2:55 888. 888. 888. 888. 88:. :83. 888. .80.. .5 h :5 .28.... .88.. .5 .- 55 5865 5:8: .5 .5 .5 82:5...— Uouuaum .550 he toeuafion ©3530 .5 82:5...— munnokgum 5.5—2.0 =< 5 88.9.2: 03...— u: Bow—m— 2: .5.— 8_=mo~— banana—m oath-89.0 5.2.5252 an 2...; 72 Table 30: Distance Metric Results Parameter Estimate Error t-stat P-value Constant (Mo) 1.39E-03 9.87E-05 14.0347 0.000 Brand (Ml) -1.12E-03 1.16E-04 -9.6605 0.000 Flavor (M2) 6.48E-04 2.14E—04 3.03108 0.002 Caffeine (M3)* —5.22E-05 9.12E-05 -0.5724 0.567 Calories (M4) -8.78E-04 1.36E-04 -6.4362 0.000 Sodium (Ms) -2.37E-03 1.93E-04 -12.278 0.000 73 * Not significant at the 1, 5, or 10 percent levels 308 2.5 a? 8.8 35 .58 33 Ea 35 £58 2%; .5 55> 930 .98 as: as, if .05 6:8 83> an»; 35 .608 «2.2% a? 38 $5 .3908 29m .5 «55> .905 «05 $2.58 98 was: a: h £26— 3839 o— .8 .m J 05 8 Ewe-mama «oz ... «8&8? ...vc-&$.” *vo-mvo? .vo-mhv? 8&8. _ - *g-mmad- *vo-mwNN «55> Emma 8&8.” 8&86- 8&8.” 8&mfim 8&_~.m- 8&3.” 8&3.” AUG 8&35 *vo-m _ Em *3&cc.m- *8-5 m. T ...3& _ N” five-m2 .m- *8& 8.N ”U 830 8&3.” 8&8.” *mo-mmc._- awe-m8; 8&8.” .._8&:._- *g-mmng :9 8&2; 8&mm.m 8&8? 8&86 8&2..m mo-mwwd 8&86 PEN 330 8&8.” ‘vo-mvw. _ - «3&va 8&3.” 8&3.” 382%: .a- ave-mm? 7 ago 8&8. T 8&8.” 8&2 . - yvo-mvmfi 8&3.” 3&3 .w- 8&8. _ - 2:5 \B .39.— mc-mvo? mo-mvwd- 8&8? 8&omd 8&3 .m- 8&3.” 8-&_ m.” 13: 15.9 ud *vo-mon? _._8&8._- uvo-mve.” 8&3.” 8&8.” {we-ma.” 2:: \B 8.90 8&3.” 8&2.” mo-mvm.” ...vc-&8.v 8&8? 8&8.” 8&9”. _ - Kr:— 8-&E-. m 8-&n~.m 8&8? 8&8.” *3-&ow.m 8&2 .” ...8-&mw.” won ‘mo-mmn” ovo-ma.” *3-mmmé *mo-memg 8&8.” *g-mmcd- *vo-mond nava— mo-mne. T *3-mmo.m- *mo-mvwé 8&3. H - *mo-mmo. fl - ...vo&cm.v _..8-m8._ mgr—Db 33555 .9550 «Eur— aaaontcz 5.5m ES 855 «3.3.5.5 5.5m 39.0 3.52595:— «2595 39 Z “—5.:— flow 33:330 3.. 3:83— 2:25» :38me SM 03.; 74 £32 Boob S 8 .m ._ 05 an 58$ E 82 ... 888.7 8%.? 883m- 8838 888$ 888$ .888? a 8883 882.”- 288.8 888.? 388.8 888.8 888.8- %: .52 :5 88sz- 88:8- 8.8!":- 8828- 8888- 8888. 888.8 a: h :5 882.8 888.8 8%? 88on- 8888 8838. 888.7 8% 888 8. 828.8. 8888- 888.8 388.8- _._888.m 888.8- 8% 888 o .888... 888.8 888 .8838 88.8 8E8 3.5 .2: 888 8 888.8. 888.8 82 28.8. 82 8.8 8888 888.8. 58 55552 8:88- 22: 3 888.8 2 88.8- 8888- 8888- $88.8- 5888 .5 888.8- 8888- .8888 88:88- N8888- 8882 888.8 a: b 888.8 888.8 888.8. 8888 8-8.8 8-83- 888.” 81%|] 8:88- 888- 888.8- ...8-m1$.- 88:8. 8888- 828.c- : o.— 8388 888.8 “8.8.8- 8588 888.8 288- 8:88 8.5 .wflu—tflcw _a-5EOU mflmwrm uwfiozwhOZ :afléw mov—flwm NEP—Ofi—va 525 E: 320 22.3-— hoauom 83.5 tow 58.85.8850 .88 3:83— 2:52; 3:2qu "mm «Ea-H 75 APPENDIX 3: FIGURES Figure l: Food/Drug Store and Mass Merchandiser Sales and Forecasts of Carbonated Soft Drinks 12.000 10.000 8.000 \ 6,000 * -——Regu|ar CSDs Million 5 p—————-— 4,000 -' 2,000 0 ,\ e .e .3 .5 s e°e°§§e@§§§§§§ * Estimated ** Forecasted Source: Mintel Group, Carbonated Drinks - US - May 2008 - Segment Performance Figure 2: Per Capita Consumption of Carbonated Soft Drinks, Bottled Water and Sports Drinks, 2000-06 40 g —Carbonated Beverages 3 30 w — -BottiedWater a ,.g I'— . 0 __ —- - - 'Sports Dunks ’ 2°‘—_;-‘j"_ 10 -.--------‘-- O Y fir I l' I I 2000 2001 2002 2003 2004 2005 2006 Year Source: Mintel Group, Carbonated Dn'nks - US - April 2007 - Market Drivers 76 Figure 3: Food/Drug Store and Mass Merchandiser Functional Beverage Sales Million 5 14.000 12,000 10.000 8.000 6.000 4,000 2.000 / T I l j 1 l 2002 2003 2004 2005 2006 2007' 2008“ 2009“ 2010“ 2011“ 2012“ * Estimated ""' Forecasted Note: Excludes sales through Wal-Man Source: Mintel Group, Functional Beverages - US - August 2007 77 REFERENCES Alston, J ., K. Foster and R. Green, (1994). “Estimating Elasticities with the Linear Approximate Almost Ideal Demand System: Some Monte Carlo Results.” Review of Economics and Statistics 76 (1994), pp. 351—356. Anderson, Linnea. “Carbonated Beverages”. Hoovers. Retrieved January 14, 2008. http ://premium.hoovers.com/subscribe/ind/overview.xhtml?HIC[D=1 049 Anderson, Simon P., Andre de Palrna, and J acques-Franqois Thisse. Discrete Choice Theory of Product Difi'erentiation. Cambridge, MA: MIT Press. 1992. Anonymous. “Battle of the Diet Cola Clones,” Consumer Reports, Vol. 70, Issue 12; page 9. December 1, 2005. Berry, Steven T. (1994). “Estimating Discrete-Choice Models of Product Differentiation,” RAND Journal of Economics, 25 (Summer), 242-62. Brown, M.G., R. M. Behr and J. Lee (1994). “Conditional Demand and Endogeneity? A Case Study of Demand for Juice Products,” Journal of Agricultural and Resource Economics, 19, 129. Buzzell, Robert D. and Robert E. M. Nourse. Product Innovation in Food Processing, 1954-1964. Harvard Graduate School of Business Administration, Harvard Univeristy. Cambridge, MA. 1967. Capps, Oral, Jeffrey Church and H. Alan Love (2003). “Specification Issues and Confidence Intervals in Unilateral Price Effects Analysis.” Journal of Econometrics, 113, 3-31. Carew, R., W.J. Florowski and S. He (2004). “Demand for Domestic and Imported Wine in British Columbia: A Source-Differentiated Almost Ideal Demand System Approach”, Canadian Journal of Agricultural Economics, Vol. 52, pp. 183-99. Cavallo, J. (2000). "The Challenge of Managing Taste.” Beverage Industry, 91, 75—76. Caves, Richard. American Industry: Structure, Conduct, Performance. 7th edition. Prentice Hall. Upper Saddle River, NJ. 1992. Chan, T. (2006). “Estimating 3 Continuous Hedonic-Choice Model with an Application to Demand for Soft Drinks,” RAND Journal of Economics, Vol. 37(2), pages 466- 482. Chaney, P., T. Devinney and R. Winer (1991). “The Impact of New Product Introductions on the Market Value of Firms,” Journal of Business 64. 573-610. 78 Conner, John M. (1980). “Food Product Proliferation: Part 1.”, National Food Review, No. 10 (Spring), pp. 12-13. Cosgrove, J. (2004). “Can Fortified Beverages Taste Great and Be Good for you?” Beverage Industry, 95, 59—68. Cotterill, R. W., and PO. Samson (2002). “Estimating a Brand-Level Demand System for American Cheese Products to Evaluate Unilateral and Coordinated Market Power Strategies.” American Journal of Agricultural Economics, 817-823. Cotterill, R.W. and WP. Putsis, Jr (2000). “Market Share and Price Setting Behavior for Private Labels and National Brands.” Review of Industrial Organization. 17:17— 39. D’Aveni, R.A. Hypercompetition: Managing the Dynamics of Strategic Maneuvering. Free Press. New York, NY. 1994. Deaton, Angus and John Muellbauer. Economics and Consumer Behavior. Cambridge University Press. New York, New York. 19803. Deaton, Angus and John Muellbauer (1980b). “An Almost Ideal Demand System.” The American Economic Review, Vol. 70, No. 3, pp. 312-326. Dhar, T., J .P. Chavas and B.W. Gould (2003). “An Empirical Assessment of Endogeneity Issues in Demand Analysis for Differentiated Products.” American Journal of Agricultural Economics 85:605—617. Draganska, M. and D. Jain (2005). “Product-Line Length as a Competitive Tool,” Journal of Economics and Management Strategy, 14 (1), pp. 1— 28. Dube, J .P. (2004). “Multiple Discreteness and Product Differentiation: Demand for Carbonated Soft Drinks,” Marketing Science, 23( 1). (2005). “Product Differentiation and Mergers in the Carbonated Soft Drink Industry”. Journal of Economics and Management Strategy, 14(4), 879—904. Fader, Peter S. and Bruce G.S. Hardie (1996). “Modeling Consmner Choice Among SKUs,” Journal of Marketing Research, 33 (November), 442-52. Foote, A. (2002). “What’s New? Global Beverage Flavour Trends.” Beverage World International, 20, 32—34. Green, Richard and Julian M. Alston (1990). “Elasticities in AIDS Models.” American Journal of Agricultural Economics, Vol. 72, No. 2, pp. 442-445. 79 Greising, David. I 'd Like the World to Buy a Coke: The Life and Leadership of Roberto Guizueta. John Wiley & Sons, Inc. New York, NY. 1998. Hasler, C. M. (2000). “The Changing Face of Functional Foods”. Journal of the American College of Nutrition. Vol. 19, No. 5, 499S—506S. Kim, J ., G.M. Allenby and RE. Rossi. (2002). “Modeling Consumer Demand for Variety’ ’, Marketing Science 21, 229—250. Lancaster, Kelvin, (1966). “A New Approach to Consumer Theory. Journal of Political Economy 14, pp. 133—156. . Consumer Demand: A New Approach. Columbia University Press. New York, NY. 1971. (1990). “The Economics of Product Variety: A Survey”. Marketing Science. 9(Summer) 189-206. . Modern Consumer Theory. Edward Elgar Publishing Company. Brookfield, VT. 1991. Larue, B., A. Ker and L. MacKinnon (1991). “The Demand for Wine in Ontario and the Phasing-Out of Discriminatory Mark-Ups.” Agribusiness 7 (5): 475—488. Luo, Lan, P.K. Kannan and Brian T. Ratchford (2007). “New Product Development Under Channel Acceptance”. Marketing Science, Vol. 26, No. 2, pp. 149—163. Matlack, Carol and Rachel Tiplady. “The Big Brands Go Begging in Europe.” Business Week, Issue 3925; page 24. March 21, 2005. Mintel Group. Carbonated Drinks - US May 2008. . Carbonated Drinks - US April 2007. . Carbonated Drinks - US - March 2006. . Consumer Choices in the Beverage Aisle - US - April 2008. . Functional Beverages - US August 2007. . Non-alcoholic Beverages: The Market - US - April 2008. Moschini, G (1995). “Units of Measurement and the Stone Index in Demand System Estimation.” American Journal of Agricultural Economics, 77, pp. 63—68. 80 Muris, Timothy J ., David T. Scheffman and Pablo T. Spiller. Strategy, Structure and Antitrust in the Carbonated Soft Drink Industry. Quorum Books. Westport, CT. 1993. “New Flavored Water Brands Flooding the Market.” Beverage Marketing Corporation, February 10, 2005. http://www.beveragemarketing.com Pendergrast, M. For God, Country and Coca-Cola: The Definitive History of the Great American Soft Drink and the Company that Makes It. Basic Books. New York, NY. 2000. PepsiCo. DEWmocracy. The Journey to Create the Next Mountain Dew. www.dewmocracy.com Pinkse, J ., and Slade, M (2004). “Mergers, Brand Competition and the Price of a Pint.” European Economic Review, Vol. 48, pp. 617-643. , and Brett, C (2002). “Spatial Price Competition: A Semiparametric Approach.” Econometrica, Vol. 70, pp. 1111-1153. Pofahl, Geoffrey M. and Timothy J. Richards (2008). “Valuation of New Products in Attribute Space.” American Journal of Agricultural Economics (forthcoming). Pofahl, Geoffrey M (2008). “What is a Beverage Worth? Arbitrage Pricing and the Value of New Products: An Attribute-Space Approach.” Presented at the 2008 American Agricultural Economics Association Joint Meetings in Orlando, FL. “Premium Bottled Beverages.” Encyclopedia of Emerging Industries. Online Edition. Thomson Gale, 2007. Reproduced in Business and Company Resource Center. F armington Hills, Mich.:Gale Group. 2008. http://galenet.galegroup.com/servlet/BCRC Purcell, Denise. “Specialty Food for the Powerful Twenty-Somethings”, Specialty Food Magazine: Products, Trends, and Your Business Perspective, January 20, 2006. http://www.specialtyfood.com/do/news/ViewNewsArticle?id=2334 “R & D Resources.” (Editor’s Note) Beverage World, v122 ill pS3(1) November 15, 2003. Ratchford, Brian T. (1975). “The New Economic Theory of Consumer Behavior: An Interpretive Essay”. Journal of Consumer Research 2, pp. 65—75. Remilia, Laurence. “There’s No Success Like ‘Failure’” 2000. (Accessed: January 28, 2008). httpzl/www.failuremag.com/news_strategies_europe.html . Strategies Europe. December 81 Riley, J .J . A History of the American Soft Drink Industry: Bottled Carbonated Beverages 1807-195 7. American Bottlers of Carbonated Beverages, Washington, DC. 1958. Scale, J. L., M. A. Marchant and A. Basso, (2002). “Imports versus Domestic Production: A Demand System Analysis of the U.S. Red Wine Market. Review of Agricultural Economics: 25 (1): 187—202. “Soft Drinks and Bottled Water.” Encyc10pedia of Global Industries. Online Edition. Thomson Gale, 2007. Reproduced in Business and Company Resource Center. Fannington Hills, Mich.:Gale Group. 2008. http://galenet.galegroup.com/servlet/BCRC Sorenson, Douglas and Joe Bogue (2006). “Modelling Soft Drink Purchasers’ Preferences for Stimulant Beverages” International Journal of Food Science & Technology 41 (6), 704—71 1. “The U.S. Market for Gourmet and Specialty Foods and Beverages”, Volumes 1 and 2, 6th edition. Packaged Facts. Sep 1, 2005 - ID: LA1087756 http://www.packagedfacts.com/GounnetSpecialty-Foods- l 087 7 5 6/ Weisberg, K. (2001). “More Than a Pick-Me—Up: Functional Beverages.” F oodService Directory, 14, 84. “Wellness & Functional Beverages in the U.S.” Beverage Marketing Corporation of New York. 2002. http://www.beveragemarketing.com Wilson, W. W, (1994). “Demand for Wheat Classes by Pacific Rim Countries”. Journal of Agricultural and Resource Economics 19 (1): 197—209. Xiao, H., H. W. Kinnucan and H. M. Kaiser. (1998). “Advertising, Structural Change, and U.S. Non-Alcoholic Drink Demand.” Research Paper No. 98-01, National Institute for Commodity Promotion Research and Evaluation, Cornell University, Ithaca, NY. 82 till L” l "i flil V Mi " 1293 03062 8881 3 ummmrlmlwl