Marketing Science Institute Working Paper Series 2020 Report No. 20-132

"Natural" New Products and Brand Distribution

Mitchell C. Olsen, Frank Germann, Meike Eilert

"Natural" New Products and Brand Distribution” © 2020 Mitchell C. Olsen, Frank Germann, Meike Eilert

MSI working papers are distributed for the benefit of MSI corporate and academic members and the general public. Reports are not to be reproduced or published in any form or by any means, electronic or mechanical, without written permission.

Marketing Science Institute Working Paper Series "Natural" New Products and Brand Distribution

Mitchell C. Olsen* Mendoza College of Business University of Notre Dame Notre Dame, IN 46556 USA [email protected]

Frank Germann Mendoza College of Business University of Notre Dame Notre Dame, IN 46556 USA [email protected]

Meike Eilert Gatton College of Business and Economics University of Kentucky Lexington, KY 40506 USA mei224 @uky.edu

July 31, 2020

*Corresponding Author

The authors are thankful for the constructive comments and suggestions provided by Peter Ebbes, Shankar Ganesan, Rajdeep Grewal, John Sherry, Joel Urbany, and Rich Williams. They also thank Natalie Chisam, Jake Eberhart, Erin Jackson, Daniel Kilcullen, and Melanie Langan for their research support. The authors appreciate data from Information Resources Inc. All estimates and analyses in this paper based on Information Resources Inc. data are by the authors and not by Information Resources Inc.

Marketing Science Institute Working Paper Series "Natural" New Products and Brand Distribution

Research Summary Grocery retailers are investing in "natural" product offerings to compete for shoppers. At the same time, many brands' new product offerings claim to be "natural." Such new products appear to be congruent with retail customers' goals. However, the term "natural" is not regulated by any U.S. government agency. Uncertainty remains regarding what "natural" implies to the marketplace and the value it brings to interorganizational exchange between brands and their retail customers. We investigate these issues with a multimethod approach involving in-depth qualitative interviews with 30 managers possessing extensive category management experience, secondary data compiled from 628 brands' new product introductions across 18 consumer packaged goods categories over 11 years, and primary data from a survey-based experiment collected from 101 managers involved with category management.

Results indicate the extent to which a brand is using retailer shelf space productively can determine whether the relationship between a brand's "natural" new product introductions and brand distribution is positive or negative. Category managers find the term "natural" is difficult to evaluate - especially in non-food categories. They navigate this uncertainty by turning to brands' "shelf space productivity" (i.e. the brand's category sales contribution relative to its share of in-store shelf space) as a critical decision-making heuristic. If a non-food brand is using its shelf space productively, it is in position to gain access to greater overall distribution by focusing on "natural" new products. However, if a non-food brand is underutilizing its shelf space, category managers perceive the brand's focus on "natural" new products as more opportunistic, which lowers their trust in the brand's use of the claim. This distrust ultimately results in a larger withdrawal of the brand's access to retailers' distribution resources than had the brand focused more on non-"natural" new products.

These results suggest non-food brand managers should proceed with caution when considering whether their new products will focus on "natural" (vs. non-"natural") offerings. It may seem "natural" products can only help -- 0r at least not hurt-brand distribution. However, we find "natural" claims can be detrimental to distribution if the brand underutilizes its in-store shelf space. It is difficult for brands to escape downward performance spirals, and emphasizing "natural" claims seems to accelerate the descent for unproductive non-food brands.

As part of our analyses, we also propose, validate, and use a novel and straightforward measure that category managers and brand managers will find useful for estimating brand shelf space and shelf space productivity.

Keywords: Natural new products, natural claims, shelf space, productivity, brand distribution

Olsen, Germann, and Eilert 12 Marketing Science Institute Working Paper Series Many consumers say they look for products specifically labeled "natural" while grocery

shopping (Consumer Reports 2016; Schmansky 2019). Consumers expect natural offerings to be

available wherever they shop, and grocery retailers of all formats consider "natural" products a

necessary component of their strategy to compete effectively in the marketplace (Research and

Markets 2017; Sweeney 2019). Evidence suggests "natural," "all natural," or "100% natural"

(hereafter "natural") is a widely-encountered claim in supermarkets (e.g., Rock 2016; Rozin et

al. 2004). For instance, prominent in-store signage touting the availability of "natural" product

options is readily observed in mainstream grocers such as Publix and Kroger, as well as price-

focused supercenters like Walmart. The "natural" claim, specifically, is unique among product

claims given its widespread use across product categories (e.g., Levinovitz 2020).

Grocery retailers' interest in "natural" products offers a seemingly promising opportunity for their brand partners to supply new products explicitly making the claim. Suppliers are dependent on distributors for access to scarce resources (e.g., Lusch and Brown 1996), and research suggests no retailer-supplied1 resource is more valuable to brands' overall performance than distribution (e.g., Hanssens, Parsons, and Schultz 2001; Srinivasan, Vanhuele, and Pauwels

2010; Wilbur and Farris 2014). Interorganizational exchange tends to increase when a supplier's efforts are congruent with the buying organization's goals (e.g., Wathne and Heide 2000). For brands looking to increase overall distribution, the decision to focus on "natural" (vs. non-

"natural") new products2 therefore may offer an appealing opportunity to engage in a behavior congruent with a goal shared across retailers.

However, despite its broad usage, the term "natural" is not regulated by any U.S.

government agency, and uncertainty and skepticism remain regarding what "natural" implies to

1 We will often refer to supermarket or grocery retailers simply as "retailers" in the remainder of the manuscript. 2 We examine the degree to which a brand's new products are composed of " natural" products relative to the rest of the category. For instance , we measure ''Natural New Product Prevalence" in Study 2 as the proportion of"natural" new products launched by the focal brand relative to other brands in the category.

Olsen, Germann, and Eilert 13 Marketing Science Institute Working Paper Series the marketplace (e.g., Dewey 2017). Retailers make their own evaluation of brands' use of the

"natural" claim to avoid misleading their shoppers. Moreover, "natural" products still comprise only a minority of sales in most categories (IRI and SPINS 2020; Nielsen 2019), possibly discouraging retailers from increasing distribution to all brands touting natural product offerings.

The aforementioned issues motivate our investigation into three research questions:

1) Does a brand's focus on "natural" (vs. non-"natural") new products relate positively to brand distribution? 2) Can focusing on "natural" (vs. non-"natural") new products relate to higher levels of distribution for some brands and lower levels of distribution for others? 3) Why may some brands' "natural" new product introductions result in lower overall distribution (i.e., what is the underlying mechanism)?

We investigate these questions in three studies using a multimethod approach involving

qualitative interviews , secondary data, and an experiment. In Study 1, we conduct in-depth

qualitative interviews with 30 managers possessing extensive category management experience

in the U.S. grocery industry to understand how category managers (i.e., the retail employees

directly responsible for making brand-level distribution decisions) make distribution decisions

and whether they view "natural" new products differently than those not marked "natural."

Findings indicate "natural" new products are viewed as uniquely important to retailers and are

associated with distinct evaluations and brand distribution outcomes. Category managers suggest

the "natural" claim is more difficult to evaluate for new products in non-food categories than

food categories. To aid decision-making, they consider the non-food brand's category sales

contribution relative to its share of in-store shelf space (i.e., shelf space productivity).

We examine the interviews through the lens of exchange theory to propose formal

hypotheses, which we investigate in Study 2 by analyzing new product introductions from 628

brands across 18 categories in the U.S. grocery channel over an 11-year period. Results from

Study 2 demonstrate that if a non-food brand uses retailers' shelf space resources productively, it

is in position to gain more overall distribution by focusing on "natural" (vs. non-"natural") new

Olsen, Germann, and Eilert 14 Marketing Science Institute Working Paper Series products. However, managers of unproductive non-food brands face a paradoxical challenge: By focusing on the kind of new products their retail customers presumably want (i.e., "natural" products), they risk losing more access to distribution resources than they would if they focused more on non-"natural" new products. Thus, although it may seem that "natural" products are congruent with retailer goals and can only help-or at least not hurt-brand distribution, we find evidence of a "dark side" to emphasizing the "natural" claim.

Study 3 examines why shelf space productivity plays a significant role in determining whether a non-food brand's "natural" new product efforts ultimately prove beneficial or detrimental to brand distribution. We conduct an experiment with 101 managers involved with category management to examine the underlying mechanism driving distribution outcomes for brands making "natural" offerings prevalent among their new product introductions. We find category managers perceive unproductive non-food brands focused on "natural" new products as more opportunistic, which lowers their trust in the brand's use of the claim. This distrust ultimately results in a larger withdrawal of the brand's access to retailers' distribution resources.

Our research makes three primary contributions to marketing theory and practice. First, our findings contribute to the literature on "natural" products by examining their association with an interorganizational-level outcome. Research on the "natural" product claim tends to examine its consumer-level effects (e.g., McFadden and Huffman 2017; Rozin 2005), which is consistent with the consumer-level focus of the broader literature on "green," or environmentally friendly, claims (e.g., Lin and Chang 2012; Luchs et al. 2010; Olsen, Slotegraaf, and Chandukala 2014).

However, outcomes related to interorganizational exchange, such as brand distribution, have received little attention. For brands relying on external channel partners, it is critical to understand how retailers may react to brands' "natural" new product strategies. We demonstrate there are unique channel-level considerations influencing whether, how, and why the "natural" claim influences brand distribution - both positively and negatively.

Olsen, Germann, and Eilert I 5 Marketing Science Institute Working Paper Series Second, we add to distribution research by examining how "natural" new products serve as an antecedent to brands' overall distribution. Among the four pillars of the marketing mix- product, price, place, and promotion-place (i.e., distribution) is arguably the most important contributor to sales and market share (e.g., Wilbur and Farris 2014). Yet, distribution's antecedents are "substantially under-researched" (Ataman, Mela, and van Heerde 2008, p. 1051).

Understanding drivers of retailer distribution decisions is particularly important when new product performance is uncertain (Kaufman, Jayachandran, and Rose 2006), as it is for "natural" new products. Our multimethod approach includes in-depth qualitative interviews with retail category managers, thereby providing a rare first-person perspective from managers who exert significant influence in industry. The interviews shed light onto category managers' decision- making process, including the finding that shelf space productivity is one of the most important factors category managers consider when making brand-level distribution decisions.

Third, we contribute to research and practice relying on estimations of brands' shelf space productivity. The calculation of shelf space productivity is straightforward with access to internal store planograms and sales data. However, obtainining store-specific planograms from all stores in a trading area, including competitive retail banners, is typically unrealistic for researchers. We propose, validate, and use a novel measure for estimating shelf space and shelf space productivity. Our measure relies only on store scanner data and a common size metric

(e.g., ounces) within a category.

In the balance of the paper, we first provide further background on the "natural" claim and the role of distribution in the brand-retailer exchange. We then present results from our qualitative interviews (Study 1) to better understand the "natural" phenomenon from category managers' perspective. We consider insights from the interviews through the lens of interorganizational exchange theory before investigating our hypotheses with secondary data

Olsen, Germann, and Eilert 16 Marketing Science Institute Working Paper Series (Study 2) and an experiment (Study 3). We conclude with a discussion of our findings'

contribution to marketing theory and practice.

CONCEPTUAL BACKGROUND

The meaning of the term "natural" is open to interpretation. Consumers tend to assume

"natural" products are more environmentally friendly, healthier, and better than products without the designation (Levinovitz 2020; Rozin 2005). However, definitions vary (Evans, de

Challemaison, and Cox 2010; Rozin, Fischler, and Shields-Argeles 2012), and confusion remains when the "natural" claim is encountered on products in daily life (Anstine 2007). Researchers argue the confusion results at least partially from companies' use of the claim across a variety of products and with a diverse set of definitions (McFadden and Huffman 2017).

Manufacturers are not restricted to a narrow use of the word "natural." Since 1991, the

U.S. Food and Drug Administration (FDA) has maintained an informal policy stating "natural"

means "nothing artificial or synthetic (including colors, regardless of source) is included in, or

has been added to, the product that would not normally be expected to be there" and notes "the

term 'natural' is used on a variety of products to mean a variety of things" (US-HHS-FDA

2015). Retail category managers therefore must form their own evaluations of individual brands

claiming to offer "natural" products.

The "natural" claim has been consistently observed over many years (e.g., Rock 2016;

Rozin et al. 2004). Natural products are a specific focus within the grocery retail industry, where

market research firms consider them a unique class of products worth examining across all

category types (e.g., IRI and SPINS 2020) and retailers view them as distinctively important to

store-wide assortment decisions (e.g., Johnsen 2018).

We confirm the "natural" claim's wide use in our sample of secondary data (see Study 2)

by conducting a descriptive analysis across food and non-food product categories (see Appendix

1 for results). While its prevalence varies across categories, "natural" is the most frequently Olsen, Germann, and Eilert 17 Marketing Science Institute Working Paper Series invoked among the 32 green claims identified by Olsen and colleagues (2014) in the Product

Launch Analytics (PLA) database . "Natural" is the most widely used green claim across both food and non-food categories, with 34.56% and 22.02% of all new products in the categories making the claim. Indeed, "natural" is a product claim that deserves specific examination because it (1) appears to be the most common route to going green at the product level, (2) is commonly used across virtually all types of product categories, unlike other claims, and (3) is not a clearly defined term, thereby presenting brands and their customers with a claim that is simultaneously flexibile and ambiguous.

While several studies examine the consumer-level implications of green products (e.g.,

Lin and Chang 2012; Luchs et al. 2010; Olsen et al. 2014), little research examines green new products' association with access to key resources in a business-to-business exchange. Vertical exchange relationships, such as between retailers and brands, can be economically significant for both parties (e.g., Jap and Anderson 2003). By examining how a brand's emphasis on "natural"

(vs. non-"natural") new products is associated with brand distribution, we focus on an exchange outcome of significant importance for both retailers and brands. For retailers, distribution decisions are critical, as allocation can determine the success or failure of a retailer (e.g., Bloom,

Gundlach, and Cannon 2000). For brands, distribution is arguably the most valuable resource to which a retailer can provide access (e.g., Wilbur and Farris 2014).

While all four elements of the marketing mix are critical, researchers contend distribution

(i.e., place) is the most important determinant of brand sales and market share (e.g., Hanssens et al. 2001; Srinivasan et al. 2010). Ataman et al. (2008) report 63% of a new brand's market potential is explained by distribution. Still, research remains scarce on distribution relative to the other marketing mix elements, especially with respect to the antecedents of distribution. Extant research examines retailer adoption of individual new products (Kaufman et al. 2006;

Montgomery 1975; Rao and McLaughlin 1989). At the brand level, prior work examines the Olsen, Germann, and Eilert 18 Marketing Science Institute Working Paper Series relationship between distribution and market share (e.g., Wilbur and Farris 2014) as well as how

new brands' distribution evolves over time (Bronnenberg and Mela 2004).

While it is critical to understand how new brands survive infancy, it is also important to

consider how brands thrive as marketplace incumbents. A significant portion of managers work

with established (i.e., incumbent) brands and must navigate extant exchange relationships with

retail customers. Our research context therefore focuses on a relevant exchange outcome (brand-

level distribution) between incumbent brands and retailers and how it is associated with the

brand's decision to emphasize "natural" (vs. non-"natural") new products.3

STUDY 1: CATEGORY MANAGER PERSPECTIVE ON "NATURAL"

Design and Execution

To gain a greater understanding of the brand-retailer exchange and how "natural" new

products may influence the level of that exchange in the form of brand distribution, we first

identified the issue's key components from a managerial perspective (e.g., Ulaga and Eggert

2006; Flint, Woodruff, and Gardial 2002). We conducted semi-structured in-depth interviews,

allowing respondents to more fully share their opinions and idiosyncratic experiences (Belk,

Fischer, and Kozinets 2013). The results formed a basis for identifying relevant theory,

developing hypotheses, and conducting quantitative analyses.

Employing a triangulation approach, we conducted 30 semi-structured, in-depth

interviews with U.S.-based informants possessing extensive experience in at least one of the

following areas:4 category management responsibilities for a grocery retailer (20 respondents),

senior executive with direct or indirect oversight of category management for a grocery retailer

3 Although luxury brands usually pursue an exclusive distribution strategy where only a select number of retailers are allowed to carry the brand (e.g., Rolex watches) , most CPG brands (e.g., Heinz ketchup) pursue an intensive distribution strategy, where they seek to maximize availability across retailers. 4 The total across the three areas of experience, 43, was greater than the number of interviews, 30, because some individuals had experience in multiple areas. For example, one manager we spoke with was formerly a category manager at a grocery retailer before moving into her current role as a category captain for one of her manufacturer's retail customers.

Olsen, Germann, and Eilert I 9 Marketing Science Institute Working Paper Series (11 respondents with positions at the director, vice president, or C-suite level), and brand partners working with category management teams to make distribution decisions, primarily from a "category captaincy" position (12 respondents). The approach enabled us to cross-check responses from different perspectives, providing a more complete and accurate understanding of retailers' brand-level distribution decision-making processes. Table 1 provides an overview of participant characteristics.

Participants were identified and contacted in one of two ways. The first involved searching for managers on Linkedin with experience in one of the three areas described and sending them interview requests. The approach yielded 18 phone interviews. For the second, we attended a major North American grocery industry trade show and approached attendees with relevant experience, yielding 12 face-to-face interviews. Across both efforts, we ensured a nationally representative sample of firm sizes, headquarters locations, and product category perspectives. Collectively, our respondents had direct responsibility managing virtually all categories found in a typical supermarket. Across the sample, respondents had direct experience working for at least 35 different grocery retailers. Interviews were recorded with handwritten notes and lasted an average of 0.75 hours (range: 0.25-1.75 hours).

Each interview consisted of three sections. We first asked respondents to describe their current and past work responsibilities, functions, and product category experience. The purpose was to understand the individual's relevant experiences and inform subsequent questioning. In the second section, we asked respondents about the most important factors influencing whether and to what extent a brand's overall distribution level would increase or decrease at their retailer.

The purpose was to determine critical brand-level distribution drivers as well as the formal and informal processes used to make distribution decisions.

Each interview's third section asked respondents about their thoughts on new products making the "natural" claim and brands focused on "natural" new product introductions. Olsen, Germann, and Eilert 110 Marketing Science Institute Working Paper Series Particular attention was given to whether category managers evaluate "natural" new products differently than non-"natural" new products and why, the role "natural" products play in the store, and whether "natural" new products are evaluated differently in some categories compared to others. The purpose was to understand managers' genuine thoughts and opinions about the

"natural" claim and how they evaluate brands when "natural" (vs. non-"natural") new products are prevalent. In all sections, questions were carefully phrased in a nondirective and unobtrusive manner to avoid influencing responses through "active listening" (McCracken 1988, p. 21).

Our interview analysis was guided by a grounded theory approach (Glaser and Strauss

1967) and involved a constant comparative method (Spiggle 1994). After each interview, we

assessed thematic patterns and, following a hermeneutic circle of understanding (Dilthey 1957;

Schleiermacher 1998 I 1838), compared subsequent responses to previous interview data and

interpreted the results. We report results and thematic patterns from the qualitative interviews in

an order and structure germane to our research context.

Findings

Evaluating "natural" new products. In our interviews, managers indicated they typically

conduct a full review of any category once per year. Retailers scrutinize each brand's

assortment-level during the reviews and make new brand allocation decisions that are

implemented in a full category reset across their stores following the review. The reviews are

usually months-long, intensive exercises in which category managers forecast how well each

brand will perform going forward. Managers said brands' new products play an important role in

the forecasts. They also stressed new product performance is difficult to assess. New products

generally are risky, as evidenced by their high failure rates (e.g., Urban and Hauser 1993).

Managers unanimously said "natural" new products are more difficult to evaluate than

non-"natural," or conventional, new products. Several managers said products making the

Olsen, Germann, and Eilert 111 Marketing Science Institute Working Paper Series "natural" claim are considered "specialty items" and undergo different evaluations than

equivalent non-"natural" products. As one category manager stated,

"Natural products are evaluated differently ...[they} demand an extra consideration compared to other types of products. "

Respondents gave two separate reasons for "natural" new products' special consideration.

The positive and negative considerations were counterbalanced against each other when retailers made distribution decisions: (1) "natural" product offerings are important to the store, but (2a)

"natural" products are difficult to evaluate because their market saturation point is unknown and

(2b) the claim's meaning is undefined.

All respondents said "natural" offerings play an important role in their stores' strategy.

One C-level executive said "natural" offerings are a "big priority, " while a category manager

with experience at five different retailers said:

"I can say with 100% certainty ...'Yes!' Huge trend overall ...all categories are trying to hit on it to some degree...Everyone [referring to grocery retailers] is trying to hop on the 'natural' trend in some way or form. "

Many managers said the "natural" segment has been a source of growth across their

stores for some time. This common theme was articulated by an executive:

"'Natural' is still a relatively small portion of most category sales, but it's a very fast­ growing segment. Therefore, retailers see this as an area where they need to lean in and invest ."

Several managers used similar phrasing regarding the "need to lean in" and "get ahead"

of the trend on "natural" products, indicating the items may offer an opportunity for brands to

expand distribution by aligning themselves with a retailer goal. Indeed, exchange theory asserts

goal-congruent activities are beneficial for channel relationships (e.g., Wathne and Heide 2000).

Category managers in our sample also stressed the need to tread carefully on where and

to what degree their retail outlets' assortment leans into "natural" in a given category. One

category manager explained,

Olsen, Germann, and Eilert 112 Marketing Science Institute Working Paper Series "['Natural'] is where trends are going...[However] as much as we want to bring those ['natural 'products] in, we only have so much shelf space and have to consider the opportunity cost with what they'd replace ...There 's just a ton of those ['natural'] options out there, so we need to be selective ...they're riskier ...what their long­term presence will be [in terms of their market potential] is still unknown ...There's a lot of risk involved in determining how big of a bet you should place on 'natural'. "

Retail managers consistenly said they were uncertain about the degree to which they

should invest in "natural" offerings, and category managers were skeptical of the term itself.

Virtually all (95%) category managers indicated awareness that the "natural" claim is

unregulated, and the FDA allows it to be used in a variety of ways. According to one category

manager, " There are no guidelines given for 'natural, 'so those products are more difficult to evaluate. Other claims are more ­cut. " Another category manager said, "It's a marketing ploy on their [i.e., the brand's] side." Several others offered similar sentiments: " [' Natural ' is] a tough one. It's not regulated. "

Food and non­food categories. Our interviews indicated retailers view "natural" offerings

as a source of opportunity and frustration. Category managers believe "natural" products are

important for their stores' marketplace strategies, but they are also wary of "natural" products

and brands' use of the claim. A category manager for a large regional grocer groaned when

asked her thoughts about the "natural" claim before saying, "That one really gets me. That's where the claims can really get crazy ...I've seen a lot [of products claiming to be 'natural'] come in...I think that 's a very loose term. " The manager then referred to her past experience in

food and non-food categories as context for a clarification question indicating a distinction

should be made between food and non-food categories. "Are you talking about food or non­food categories? Because that makes a difference. When I managed [food categories], I really paid attention to the ingredients [which helped me determine the validity of the claim]. In non­food,

it's more difficult. It's tough to understand what's meant by that claim. " In other words, whether

Olsen, Germann, and Eilert 113 Marketing Science Institute Working Paper Series the "natural" claim is made in a food or non-food category plays a central role in the extent to

which it can be verified and understood by the retailer.

The sentiment was echoed in other interviews. A category manager from a large grocer

based in the U.S. South said, "For consumers, 'natu ral ' is a vague term, so their expectations

are important, and ...those expectations vary between food and non­food categories when it

comes to products being 'natural. '"

Several managers expressed confidence in their ability to diagnose the validity of a food

product's "natural" claim. In line with research finding individuals think of food and beverages

as inherently natural (Rozin et al. 2012), respondents said food products are expected to be

natural whether they carry the claim or not.

In contrast, research shows non-food items do not necessarily carry the same connotation

(Luchs et al. 2010; Rozin et al. 2004). Andre and colleagues (2019) note "naturalness" represents the absence of human intervention, which aligns poorly with scientifically developed and manufactured non-food product categories. Indeed, it is difficult to visualize products like window cleaners in a state free of human intervention. Due to the ambiguity surrounding the

"natural" claim in non-food categories, we focus our subsequent reporting on how category managers engage in sensemaking of non-food brands' use of the claim while evaluating the level of distribution resources they will lend the brands going forward.

Our interviews further indicated category managers use additional information when

evaluating non-food brands' "natural" new product emphasis. The responses were consistent

with research finding decision makers are averse to uncertainty (Kahneman and Tversky 1979)

and will apply decision heuristics to grapple with the uncertainty of another organization's future

behavior and performance (Montgomery, Moore, and Urbany 2005). One category manager

offered a representative explanation, saying she relies on the brand's "performance to make the

call" when evaluating a non-food brand's "natural" new products. Olsen, Germann, and Eilert I 14 Marketing Science Institute Working Paper Series While brand performance can be measured in a number of ways, our interviews found

category managers believe one metric is most important when adjusting a brand's overall

distribution level-specifically, how well is the brand using its current in-store shelf space?

Shelf space productivity and non­food brands' "natural" new products. Respondents

across the board said how well a brand contributes to overall category sales, given the relative

amount of shelf space it occupies, is a key consideration for brand-level distribution decisions.

The metric was commonly referred to as "shelf space productivity" and "space to sales"

(hereinafter "shelf space productivity" or "productivity"). An overwhelming majority of

respondents volunteered the metric as critical to category assortment decisions, saying: "Number one is the brand's 'shelf space productivity'" and "[Shelf space productivity} is the driving force behind how we set store­level planograms. "

Respondents suggested shelf space productivity is used as a standardized performance

measure for comparing brands in a given category. Unproductive brands hold a greater

percentage of category shelf space than sales, and our interviews suggested the brands' overall

assortment would likely be lowered to resolve this type of imbalance. Conversely, brands with a

relatively low share of category shelf space relative to sales would see an increase in overall

assortment, ceteris paribus. To maximize overall category performance, adjustments are made to

a typically "zero sum game" with respect to the category's total allowable assortment size within

stores. A category manager explained, "There's only so much shelf space, so you really have to prioritize [brands based on how well they are doing on the metric}. " Another category manager

said: "It's very important, because that shelf space is very costly. Your category is landlocked. "

Our interviews suggested category managers rely heavily on shelf space productivity

when evaluating a brand's overall distribution going forward. We therefore hypothesize:

H1: A brand's shelf space productivity is positively associated with overall brand distribution.

Olsen, Germann, and Eilert I 15 Marketing Science Institute Working Paper Series Qualitative Insights through Exchange Theory

The language used by respondents strongly suggested exchange theory as an appropriate

lens for providing structure to their comments in our research context. Interviews suggested

category managers may be wary of opportunistic behavior among brands launching "natural"

new products. The observation points toward a key issue in interorganizational exchange theory,

where opportunism is defined as "self-interest seeking with guile" (Williamson 1975, p. 255).

While retailers can expect all brand partners to be self-interest seeking, their behavior extends into opportunism only when their actions are perceived to be the result of calculated efforts to mislead. Transaction cost economics often conceptualizes opportunistic behavior in "blatant" or

"strong form" manifestations (Masten 1988; Wathne and Heide 2000) clearly violating explicit contracts. However, emergent conceptualizations argue opportunism can occur in weaker forms, such as violations ofrelational contracts (e.g., Wang, Kayande, and Jap 2010; Wathne and Heide

2000). Moreover, brands can engage in active and passive forms of opportunism, characterized as opportunism by commission or omission (e.g., Seggie, Griffith, and Jap 2013).

Our interviews indicated "natural" new products potentially constitute active opportunism, but they are not a blatant or strong form. Introducing new products and making the

"natural" claim are self-interest seeking acts. However, due to information asymmetry between brands and their retail customers, it is difficult for retailers to identify active attempts to mislead.

Category managers are often unsure if a brand is using the undefined, and hence ambiguous,

"natural" claim in a way the retailer and its customers may perceive as misleading if all information from the brand were available. Our interviews made clear "natural" new products are especially difficult to evaluate in non-food categories. In exchange relationships, the focal organization (e.g., the retailer) assesses its suppliers' (e.g., brands') performance to determine how the relationship will expand or contract going forward (e.g., Williamson 1979, 1993). As

Olsen, Germann, and Eilert I 16 Marketing Science Institute Working Paper Series detailed previously, category managers rely on the brand's shelf space productivity as a decision- making heuristic for evaluating non-food brands' "natural" new products.

Respondents indicated category managers may be wary of brands opportunistically claiming their new products are "natural" to exploit retailers' marketwide goal of providing natural offerings to shoppers. If the retailer perceives a brand to be opportunistic, its trust in the brand's use of the claim is expected to erode, resulting in a reduction of access to scarce distribution resources. Per Coleman (1990, p. 91), "situations involving trust constitute a subclass of those involving risk. They are situations in which the risk one takes depends on the performance of another actor."

Interview respondents suggested "natural" (vs. non-"natural") new products from unproductive non-food brands can amplify the distribution loss otherwise experienced based on unproductive use of retailer shelf space. When unproductive brands focus on "natural" new products, their relatively poor stewardship of retailers' shelf space resource is expected to increase suspicions of opportunistic behavior. In tum, the category manager's treatment of the brand going forward may be especially harsh, as illustrated by a category manager with extensive non-food category experience:

"If I have a brand that isn't meeting my performance expectations come to me with a bunch of 'natural' new products, I'd think, 'C'man, it 's time to cut the crap.'I need them to focus on actions that will start turning around their performance in my stores first. "

If a category manager evaluates a brand poorly using its shelf space resources and perceives the brand to be engaged in opportunistic behavior, the manager may reduce exposure to the brand with relatively drastic cuts to the offending brand's distribution. Our interviews indicated a retailer's decision to extend distribution resources to a brand partner exposes the retailer to vulnerability and risk. When opportunism arises in an existing channel relationship, the offended firm holds back valuable resources from the exchange, reducing its exposure to further opportunism from the offending firm (e.g., Jap and Anderson 2003; Williamson 1985). Olsen, Germann, and Eilert 117 Marketing Science Institute Working Paper Series

As Ganesan et al. (2010) argue, opportunistic seller behavior is likely to damage a buyer's trust,

ultimately motivating the buyer to take punitive action.

However, retailers may be less likely to perceive a productive (vs. unproductive) non- food brand's "natural" new product efforts as opportunistic, because the brand has proven itself a good steward of a valuable retailer-owned resource. As wariness of opportunism subsides, retailer trust in the brand's use of the claim is expected to increase, thereby increasing the likelihood the retailer will reward the brand's goal-congruent behavior with expanded distribution. Indeed, exchange also benefits through trust in actions taken by the other party

(Anderson and Weitz 1989; Kronman 1985). In an effort to fulfill the retailer's goal of providing

"natural" products, category managers may reward productive non-food brands' "natural" new product efforts with distribution beyond what would have been received for non-"natural" new products. We therefore expect productive non-food brands to benefit most from retailers' selective investment in "natural" offerings.

In summary, productive non-food brands may benefit from "natural" new products, but

we expect to find a converse to the action for unproductive non-food brands:

H2A: For non-food brands with high shelf space productivity, emphasizing "natural" (vs. non- "natural") new products is more positively associated with brand distribution.

H2B: For non-food brands with low shelf space productivity, emphasizing "natural" (vs. non- "natural") new products is more negatively associated with brand distribution.

ff3: When a non-food brand emphasizes "natural" new products, high (low) brand shelf space productivity (a) reduces (increases) perceptions of opportunistic behavior, which (b) increases (decreases) trust in the brand's use of the "natural" claim, thereby (c) increasing (decreasing) brand distribution.

We test H1 and H2A/B via an empirical analysis using secondary data from brands' new

product introductions (Study 2). We then conduct an experiment involving grocery industry

managers to investigate the mechanism hypothesized in H3 (Study 3).

Olsen, Germann, and Eilert I 18 Marketing Science Institute Working Paper Series STUDY 2: "NATURAL" NEW PRODUCTS AND BRAND DISTRIBUTION

Data

We collect the secondary dataset from three sources. It includes new product

introductions from 628 consumer packaged goods (CPG) brands across 18 categories in the U.S.

grocery channel from 2001 to 2011. The dataset is an aggregation of scanner data from

Information Resources, Inc. (IRI), new product launch information from Product Launch

Analytics (PLA), and advertising expenditures from Kantar Media's Ad$pender database. 2,929

brand-year observations are included in our final analysis. The CPG industry offers a relevant

context, as it invests significantly in new products (Nielsen 2015). Furthermore, grocery retailers

represent a significant portion of the U.S. economy, employing about 4.8 million workers and

generating more than $363 billion in annual economic activity (Food Marketing Institute 2017).

We capture store-level activity via the IRI scanner database described by Bronnenberg,

Kruger, and Mela (2008). The dataset has been used by previous marketing researchers (e.g.,

Datta, Ailawadi, and van Heerde 2017; van Lin and Gijsbrechts 2014) and allows us to capture

weekly distribution, sales, and pricing information at the stock keeping unit (SKU) level within

individual stores. These characteristics are essential for capturing robust distribution measures

and other focal variables in our model, such as shelf space productivity and control variables.

To combine IRI and PLA data, we examine all categories tracked by both databases from

2001 to 2011. Following Datta et al. (2017), we separate ketchup and mustard into distinct

categories. The condiments are treated as separate categories in retail stores, each with its own

clearly defined shelf space - an important consideration for our research. We exclude cigarettes.

While it is a non-food category, cigarettes are inherently different from the other non-food

packaged goods categories in the dataset. We ensure all narrowly defined IRI brand names (e.g.,

Charmin Ultra Soft, Charmin Ultra Strong, Charmin Sensitive) are coded as the parent brand

(e.g., Charmin). Because we investigate antecedents to distribution among incumbent brands, we

Olsen, Germann, and Eilert I 19 Marketing Science Institute Working Paper Series include only brand-years when the focal brand records at least two years of continuous weekly

sales in the IRI dataset (Datta et al. 2017). We do not impose a market share cutoff, creating a

mix of large national brands and small niche incumbents in our study of "natural" new products.

We then collect information on new products and their claims from GlobalData's PLA,

formerly known as Productscan by Datamonitor. PLA, a subscription database tracking CPG

brands' new product introductions across a variety of categories, has previously been used in the

marketing literature (e.g., Lamey et al. 2012; Olsen et al. 2014; Sorescu and Spanjol 2008). For

category-years tracked in PLA, GlobalData offers comprehensive coverage of all brands,

regardless of market share and distribution. PLA tracks product introductions considered new

and unique by consumers. PLA does not record instances when new universal product codes

(UPCs) are issued for pre-existing products undergoing a size change (e.g., price increases are

passed along via a 'down-count'), a minor packaging update, etc. Therefore, it offers a record of

truly new additions to a brand's product line. Because PLA data are merged with IRI data, we

focus on new products introduced in the United States. Finally, we gather information on each

brand's total annual U.S. advertising expenditures via Kantar Media's Ad$pender database.

Our research focuses on the extent to which a brand's new products are composed of

"natural" (vs. non-"natural") offerings. We include brand-years meeting the following criteria:

(1) two years of continuous weekly sales, (2) new products recorded by PLA, and (3) advertising

expenditures tracked by Kantar during the study's time period. If no advertising expenses were

recorded in the focal year, we assume the brand did not advertise that year. In addition to

excluding the cigarette category, we drop two outliers. We exclude Campbell's Soup because its

product line is about two times longer than the next largest brand. We also drop Utz from 2007,

when it atypically launched more than 300 new products. We conduct robustness checks

including cigarettes, Campbell's, and Utz and obtain consistent results.

Olsen, Germann, and Eilert I 20 Marketing Science Institute Working Paper Series Our final dataset includes brand-year observations across the following 18 categories:

Bath tissue, beer, blades and razors, butter, cereal, coffee, deodorant, facial tissue, household

cleaners, ketchup, laundry detergent, milk, mustard, salty snacks, shampoo, soup, toothpaste, and

yogurt. See Appendix 2 for a complete list of brands in each category.

Our unit of analysis is the brand-year (i.e., annual) level, following extant research on

new product strategies (e.g., Olsen et al. 2014; Sorescu and Spanjol 2008) and our qualitative

results, which indicated the annual level is most appropriate for our context. Category managers

strategically evaluate each brand's overall distribution and the extent to which they will increase

or decrease each brands' relative distribution levels during full category reviews. Respondents

overwhelmingly indicated full reviews typically occur once per year in any given category. They

are followed by total category resets where brand-level distribution changes manifest in stores.

Table 2 provides a summary of our measures and data sources. Appendix 3 presents the

sample's descriptive statistics and correlation matrix.

Focal Variables

Brand distribut ion. We capture the full breadth and depth of brand distribution within the

dataset's trading area using IRI scanner data and a measure called total distribution points

(TDPs) (see Ailawadi and Farris 2017 for a comprehensive review of distribution measures). We

make TDPs more managerially relevant by following an emergent trend in the CPG industry, re-

expressing the measure as "equivalized SKUs" (e.g., TABS Analytics 2016). Equivalized SKUs,

essentially the weighted average number of products a brand sells in a typical store within a

market, provides a single managerially intuitive number capturing distribution breadth and depth.

Appendices 4 and 5 illustrate how it is measured and why it is conceptually preferable to

alternative measures. We measure each brand's distribution via the following formula:

.t "b _ (PCVitk) (1) D IS n it- "nL..k=l 100 '

Olsen, Germann, and Eilert I 21 Marketing Science Institute Working Paper Series Where: Distribit = equivalized SKUs of brand i during year t and PCV = quarterly average% product category volume for brand i's product k over year t.

Product k is defined as a unique UPC in brand i's product line. We capture annual distribution as the average equivalized SKU value for brand i across the year's four quarters (i.e., the "quarterly average"). We utilize this quarterly average during year t, as opposed to the sum or maximum distribution achieved for any single week or quarter. By counting a product as being in distribution if it records in-store sales during the quarter, we account for the distribution of slower-moving SKUs in lower volume stores.

"Natural" new product prevalence (NNPP). To capture the degree to which an incumbent brand's new product offerings consist of "natural" new products, we measure the percentage of its new SKUs in year t making the "natural" claim. We make the calculation based on PLA information, where the database captures a claim if it is specifically observable on the product's packaging and/or in communications surrounding the product.

As shown in Appendix 1, "natural" claims are more common in some categories. To account for the variation, we calculate the prevalence of "natural" new products relative to the focal brand's category, measured as the difference between the percent of brand i's "natural" new products and the percent of the rest of the category's "natural" new products. Potential opportunistic behavior is evaluated relative to group norms in interorganizational exchange relationships (e.g., Ganesan et al. 2010; Jap and Anderson 2003), and the NNPP measure captures the extent to which a brand is emphasizing the "natural" claim more or less intensely than the rest of the category. If 50% of a brand's new products contain a "natural" claim, compared to 30% of new products across the rest of the category, the brand's NNPP for the year would be .20 (i.e., .50 - .30 = .20). The focal brand's products are removed from the category number so the naturalness of brand i's new product introductions are compared to the prevalence of "natural" new products in the rest of the category in year t.

Olsen, Germann, and Eilert I 22 Marketing Science Institute Working Paper Series Category type. We indicate whether the focal brand's product category is a food or non-

food category with a categorical dummy variable set to 1 for food categories and O for non-food

categories. Appendix 1 lists the categories by type.

Shelf space productivity. We measure how effectively a brand is utilizing its in-store shelf

space as the degree, expressed as a percentage, to which the brand over- or under-contributes to

total category sales volume given the brand's percentage of in-store shelf space, weighted by

each store's PCV.5 We use the following formula, which converts an index value of productivity

into a percentage (e.g., an index of 1.154 becomes .154; an index of .846 becomes -.154):

(2) ShelfProd-= In_Store_Category_Sales_Shareit _ l It In_Store_Category_Shelf_Shareit '

Where: ShelfProdit = shelf space productivity of brand i during year t; In_Store_Category_Sales_Shareit = quarterly average% share of in-store category volume sales for brand i over year t, weighted by each store's PCV; and In_Store_Category_Shelf_Sharei1= quarterly average% share of total category in-store shelf space for brand i over year t, weighted by each store's PCV.

The measure's validity is contingent on an accurate estimate of each brand's in-store

share of category shelf space. The ideal data source for the measure would be category

planograms for each store in the trading area during the 44 quarters forming the annual measures

of the time period we study. However, detailed planograms across the individual stores are

inaccessible. Therefore, we consider two alternative measures for estimating a brand's in-store

share-of-shelf using scanner data through in-store audits.

We first estimate in-store category shelf-share by calculating a brand's share of a11 unique

UPCs sold in a store over a quarter. This "share-of-SKUs" measure has previously been used to

measure brand distribution depth (e.g., Ataman et al. 2008). Our audits extend this work by

demonstrating it is also a relatively accurate estimate of actual in-store shelf space. However, we

5 We follow best practices in the U.S. grocery industry by weighting all stores by PCV whenever individual stores' relative differences should be taken into account (e.g., not all stores are equally valuable to a brand or retailer).

Olsen, Germann, and Eilert I 23 Marketing Science Institute Working Paper Series argue this approach has a conceptual shortcoming, because it assumes all SKUs occupy the same amount of space on the shelf. This is rarely, if ever, the case in actual stores. While we do not know the number of facings each product has in each store, we can account for relative size differences among SKUs ifwe have a common unit of measurement in each category. The IRI scanner dataset provides a volume-equivalent measurement unit for each SKU based on size units common to a category (e.g., all UPCs in the yogurt category are compared to a standard 16- ounce unit, so an 8-ounce yogurt has a volume-equivalent value of .5). We can therefore account for size differences under the assumption larger items will generally occupy more shelf space.

Consequently, we also estimate a brand's share-of-shelf by dividing the sum of its UPCs' volume-equivalent measurement units by the sum of volume-equivalent values across all category UPCs sold in a store during time t. We use the following equation to calculate the quarterly average share-of-shelf for brand i across all stores g where the brand is sold in year t:

"n (L =l Vol_Eqikgt) PCVgt (3) In_Store_Category_Shelf_Shareit = L..g=i((._,n )x--), L..k=l Vol_Eqmkgt PCVit

Where: In_Store_Category_Shelf_Shareit = quarterly average share of category shelf space in stores where brand i is sold during year t; Vol_Eqikgt = quarterly average of the sum of brand i's volume-equivalent measurement units for all unique products k sold in store g over year t; Vol_Eqmkgt = quarterly average of the sum of category m's volume-equivalent measurement units for all unique products k sold in store g over year t; PCVgt = quarterly average of the% product category volume for store gin year t; and PCVit = quarterly average of the % product category volume for brand i in year t.

Figure 1 provides an overview for measuring the two variables in the context of an example. The figure compares the volume-equivalent measurement approach to our alternative share-of-SKUs estimation . In the hypothetical example, a six-pack of 12-ounce bottles is equal to

1.0 volume-equivalent measurement unit. In this category context where all SKUs also have a single shelf facing, we have perfect alignment between the volume-equivalent measure and each

Olsen, Germann, and Eilert I 24 Marketing Science Institute Working Paper Series brand's actual share-of-shelf. However, the share-of-SKUs approach less accurately captures relative shelf-share when significant size differences exist across SKUs.

We also seek external validation of the two measures and our decision to use the volume- equivalent measure in our subsequent analysis. We conduct 54 in-store audits by auditing all 18 product categories in our sample in three distinct retail formats (i.e., mass merchandiser, mid-tier grocer, and small-format specialty grocer). To estimate each brand's share-of-shelf using

Equation 3, we first record the size of all unique UPCs on the shelf and convert them to a volume-equivalent measurement unit using the IR.I base unit (e.g., 16 ounces for laundry detergent, 100-count for facial tissue). We also calculate each brand's share-of-SKUs. For validation, we record each brand's actual share-of-shelf by measuring the total inches of linear space its products occupy on the shelf. We divide each brand's total linear shelf space (i.e., the sum of space occupied by all its SKUs) by the total amount of linear shelf space devoted to the category in the store. Table 3 summarizes our results for each category and store format.

The average correlation between the volume-equivalent estimate of brand shelf-share and actual shelf-share is .969 (0 = .028, range: .900-1.000). The average correlation between a brand's share-of-SKUs and actual shelf-share is also high, at p = .966 (a= .036, range: .874-

1.000). Both approaches effectively estimate shelf-share. The volume-equivalent measure (i.e., the measure we use in this study) tends to outperform the share-of-SKUs approach in smaller set sizes when stores accommodate a large number of brands with less overall space - settings where size differences among individual items make a noticeable difference. Moreover, the volume-equivalent measure is slightly more precise given its smaller standard deviation and narrower correlation range.

Control Variables

We include several control variables to account for alternative explanations. A summary of the specific measures can be found in Table 2. A detailed description of each variable and why

Olsen, Germann, and Eilert I 25 Marketing Science Institute Working Paper Series we included it as a control can be found in Appendix 6. However, due to space constraints, we

do not elaborate on the control variables here.

Model Formulation

We employ a log-gamma model to estimate the effects of our predictor variables on

brand distribution. 6 Log-gamma models are primarily used when a skewed dependent variable is

continuous and can only take values greater than zero. They can also be used for large count

datasets when the distribution appears continuous (Hardin and Hilbe 2007). Our dependent

variable is continuous but reflects a count-based construct in equivalized SKUs that can take

only values greater than zero. We also use a log-gamma model (i.e., log-link) 7 because the

dependent variable is over-dispersed.

Our dataset follows multiple brands over multiple consecutive years. Thus, we use a

panel log-gamma model with random brand effects. As one would expect brand distribution

levels in one year to influence the next, we confirm the presence of serial correlation in our

dataset with a Wooldridge test (F = 115.96,p < .01). To address this serial correlation, we

estimate our model with an autoregressive correlation structure. In addition, we account for

heteroscedasticity by robust clustering standard errors at the brand level. To reduce concerns of

reverse causality and allow total brand distribution at all retail outlets to respond to new product

introductions, we lag all predictor variables by one year. Moreover, we include time fixed effects

to identify unique year-to-year fluctuations, i.e., macro-economic conditions uniformly

influencing all brands in our sample (with 2002 as the referent year), as well as category fixed

6 Our results are robust to other model specifications (e.g., OLS) as we demonstrate in the robustness checks section. 1 7 The base density function for the gamma distribution is f(y; µ, cj>) = (l)cii exp(-l). In exponential-family yr(cjil µcp µcp

:L (- l nµ ) 1- In 1 } µ, µ -ct> ct>ct> - form, the probability density appears as f(y; cj>) = exp { + In y - Inf (;p) .

Olsen, Germann, and Eilert I 26 Marketing Science Institute Working Paper Series effects to account for category-specific characteristics not already captured (with bath tissue as the referent category). Our baseline model takes the following form:

(4) Distribit = exp(Boi + B1 NNPP it- l + B2 Food Ca t i + B3SSPit- l + B4 MktSha reit- l + B5 N Plit- l + B6 Pr ice Pre mit- l + B7 PLLit- l + B8 Adve rtit- l + B9Promoit-1 + B10YDt + B11CDD, Where: Distribit = distribution of brand i in equivalized SKUs in year t; NNPPit-1 = natural new product prevalence for brand i in year t - 1; FoodCat = dummy variable indicating if brand i belongs to a food category; SSPit-1 = shelf space productivity of brand i in year t - 1; MktSharei1-1 = market share of brand i in year t - 1; NPLt-1 = number of new product introductions by brand i in year t - 1; PricePremir-1 = price premium of brand i in year t - 1; PLL1-1 = product line length of brand i in year t - 1; Advertit-1 = share of category advertising voice for brand i in year t - 1; PromOit-1 = share of category promotional activity for brand i in year t - 1; YD1= vector of year-specific dummy variables; CDi = vector of category-specific dummy variables; and Bai = random intercept for each brand.

We test our hypotheses related to the moderating effects of our focal variables (i.e.,

NNPP, category type, and SSP) with the following version of Equation 4, which also allows for heterogeneity across time variant control variables through interactions with NNPP:

Distribit = exp(Boi + B1 NN PPit- l + B2 Food Cat i + B3 SSPit- i + B4 ( NN PPit- i x FoodCati) + Bs(NNPPit-i x SSPir-i) + B6 ( Food Cat i x SSPir-i) + B7 ( NN PPit- l x FoodCati x SSPir- 1) + B8 MktS hare it- l (5) + B9( NN PPit- i x MktShareit-i) + B10 N Plit- i + B11 (NNPPit-i x NPl it_1) + B12 Price Pr em it- i + B13 ( NNPPit- i x PricePremit-i) + B14 PLLit- i + B15 ( NN PPit- l x PLL it_1) + B16 Adve r t it- i + B17 ( NN PPit- l x Advertir-i) + B18 Pro moit- i + B19 ( NN PPit- l x Promoit-i) + BzoYDr + B21(NNPPit-1 X YDit-1) + B22CDj). Results

Table 4 summarizes our results. All estimates are unstandardized, with standard errors in parentheses. Model 1 shows results for Equation 4, testing the main effects of our predictor variables, including the main effect of shelf space productivity on brand distribution (H1). We then examine H2AJB by adding interactions among the main variables of interest and estimating four different models for consistency: Model 2 (main+ moderation effects only), Model 3

Olsen, Germann, and Eilert 127 Marketing Science Institute Working Paper Series

(control variables' main effects are added), Model 4 (a time trend is added), and Model 5 (i.e., the full model from Equation 5). The direction and statistical significance of our hypothesized effects remain consistent across all models. We compare model fit with Wald x2 values, where the increasing values indicate better model fit as we add controls and interaction terms.

In support of H1, we find evidence that shelf space productivity is associated with greater brand distribution across all models (e.g., main effect in Model 1: p = .296, p < .0 I). Moreover, the relationship between natural new product prevalence, shelf space productivity, and brand distribution is further impacted by category type (i.e., food or non-food). The coefficient of the three-way interaction (i.e., NNPPit-1 x FoodCati x SSPit-1) is significant (p < .05) and negative across all models (i.e., Models 2 - 5). Figure 2 is based on results from Model 5 and shows the moderating effect for an average brand competing in either a food or non-food category and over- or underperforming on shelf space productivity (i.e., all other variables set to the mean).

As Figure 2 demonstrates, food brand distribution benefits slightly more from "natural"

(vs. non-"natural") new product prevalence when the brand overdelivers on its current share-of- shelf (i.e., shelf space productivity). When food brands underdeliver on existing shelf space, there appears to be a relatively small distribution penalty for emphasizing "natural" new products. These results are consistent with our argument that food brands can expect to see subdued distribution effects when focusing on "natural" (vs. non-"natural") new products.

In contrast, "natural" new products have a noticeably larger association with brand distribution in non-food categories, both positive and negative. Consistent with H2A, productive non-food brands' distribution appear to benefit from focusing on "natural" vs. non-"natural" new products. Moreover, and consistent with H2B, for unproductive non-food brands, there is a negative association between a "natural" new product emphasis and brand distribution compared to a non-"natural" new product emphasis. Indeed, as Figure 2 illustrates, while unproductive

Olsen, Germann, and Eilert I 28 Marketing Science Institute Working Paper Series non-food brands also seem to lose distribution when focusing on non-"natural" new product introductions, these same unproductive non-food brands' distribution losses are magnified if they choose to focus on "natural" new products.

Robustness Checks

We assess the robustness of our findings using a variety of model specifications (e.g.,

Germann, Ebbes, and Grewal 2015) and present the results in Appendix 7. First, our primary analysis accounts for serial correlation via an autoregressive correlation structure that includes the t­1 time period. This modeling approach's additional lag-year requirement excludes 177 brands that would otherwise be included in the analysis. Therefore, Model Rl re-estimates the main model using the full sample of 628 brands by removing the ARI restriction. Results remain consistent. We alternately impose a stricter correction for serial correlation by including the t­2 time period in the model's autoregressive correlation structure. Our sample size is further reduced to 346 brands, but results remain consistent (see Model R2).

We also conduct a split half sample analysis in Model R3 by randomly selecting half of our full sample's 628 brands. When we estimate Equation 5 on this sub-sample, 157 brands are ultimately included. In this model, the two-way interaction between NNPP and "food category" becomes statistically significant, and negative, but the qualitative results from the overall three- way interaction are unchanged.

Next, our primary analysis relies on a random effects model due to the time invariant nature of the food category variable. We further account for time invariant brand-effects in

Model R4 by estimating Equation 5 with dummy variables for 450 of the 451 brands in the sample. Our results, including the direction and significance of the three-way interaction term, remain consistent. We note that we also estimate this model with the addition of Gaussian copula terms for the time-variant predictor variables NNPP and SSP. Results remain robust, and both copula terms are non-significant (p > .10), suggesting that, likely due to the other modeling steps Olsen, Germann, and Eilert I 29 Marketing Science Institute Working Paper Series (e.g., control variables), neither variable is endogenous (Gielens et al. 2018; Mathys, Burmester, and Clement 2016).

Finally, we further test whether our model is robust to alternative dependent variable distribution assumptions. While we believe a log-gamma model is most appropriate in our context, we also estimate an ordinary least squares (OLS) regression in Model R5, where we log- transform our skewed dependent variable. Results are robust.

STUDY 3: THE MEDIATING ROLES OF OPPORTUNISM AND TRUST

In Studies 1 and 2, we find the relationship between "natural" new products and brand distribution is contingent on category type (food vs. non-food) and the brand's shelf space productivity. A brand's decision to focus on "natural" new products is particularly impactful in non-food categories, where unproductive brands lose more distribution than they would have by focusing on non-"natural" new products. In contrast, productive non-food brands' new products are associated with more distribution if they emphasize the "natural" claim. We next utilize an experiment to formally investigate H3 and explore the theorized underlying mechanism.

Methodology

We partnered with a prominent B2B market research firm to access a customized panel of

101 individuals involved with category management in the U.S. grocery industry. The scenario- based experiment employed a between-subjects design in which the research firm randomly assigned participants to one of two conditions: unproductive brand or productive brand.

Appendix 8 summarizes the treatments and measures.

Because Studies 1 and 2 suggest productive and non-productive non-food brands see different distribution results when using the "natural" claim, we focused the experimental context on non-food brands with varying levels of shelf space productivity. Study 3 was designed to test for sequential mediation: Does a non-food brand' s shelf space productivity

Olsen, Germann, and Eilert I 30 Marketing Science Institute Working Paper Series influence the extent to which category managers perceive its "natural" new product emphasis to be opportunistic, thereby affecting trust and influencing overall distribution?

All participants were asked to assume they managed a non-food category for a grocery chain and were in the process of conducting a full category review, to be followed by an in-store category "reset" reflecting any assortment changes they made. Participants were randomly assigned to one of the two treatment conditions. In each, participants were asked to respond to statements referring to either an unproductive or productive non-food brand introducing new products emphasizing the "natural" claim more heavily than other brands in the category. As

Appendix 8 details, perceived opportunism was measured through a four-item scale adapted from Jap and Anderson (2003) (a= .76). Trust in the brand's use of the "natural" claim was captured with a three-item measure adapted from Beltramini (1982) and Ohanian (1990) (a=

.94). Finally, respondents indicated whether they would increase brand distribution on a 7-point

Likert scale (1 = strongly disagree, 7 = strongly agree).

Results

We conducted a multi-step mediation analysis using a boostrap procedure with 10,000 samples (Preacher, Rucker, and Hayes 2007; SPSS Process Macro Model 6), the "natural"- focused non-food brand's shelf space productivity as the independent variable, brand opportunism as the first mediator, claim trust as the second mediator, and brand distribution as the dependent variable. The results demonstrate the indirect effect of brand shelf space productivity on distribution through perceived opportunism and trust is statistically significant at a 95% confidence interval (indirect effect= .18, SE= .11, 95% CI [.02, .47]). A non-food brand's decision to emphasize the "natural" claim across its new products is less likely to be perceived as opportunistic when shelf space productivity is high (vs. low) (a1 = -.44, p < .05).

When the brand's behavior is perceived as opportunistic, category managers will trust the brand's use of the "natural" claim less (a3 = ­.50,p < .01). As trust in the claim decreases, the Olsen, Germann, and Eilert I 31 Marketing Science Institute Working Paper Series brand's overall distribution decreases (b2 = .83,p < .01) (Figure 3). The results provide support

for H3. When perceived opportunism and trust are included in the model, shelf space

productivity's otherwise significant direct effect on brand distribution (c = .62,p < .05) becomes

non-significant (c' = .30, p > .10), suggesting full mediation. Figure 3 displays the path model

with estimated coefficients.

DISCUSSION

Our research investigates how a brand's decision to focus on "natural" (vs. non-

"natural") new products is associated with access to a key retailer-supplied resource- distribution. We employ a multimethod approach to develop and test theory related to the role of

"natural" new products in the brand-retailer exchange by conducting 30 in-depth qualitative interviews with practitioners involved in category management, analyzing 18 categories' new product introductions over an 11-year time period and administering an experiment to 101 grocery industry practitioners familiar with category management. Triangulated results indicate focusing on "natural" (vs. non-"natural") new products represents both an opportunity and challenge for brands. Retail category managers indicate they are selective with respect to which brands' "natural" investments they reward, and it is more challenging to evaluate "natural" new products in non-food categories than food categories.

In non-food categories, retailers use a brand's shelf space productivity as a key decision- making heuristic. When a non-food brand over-delivers on shelf space productivity, a decision to focus on "natural" new products is more positively associated with brand distribution than a focus on traditional (i.e., non-"natural") new products. However, we also find evidence of a

"dark side" for "natural" new products. Non-food brands under-delivering on shelf space productivity encounter a paradoxical challenge: By launching the kind of new products retail customers say they want (i.e., "natural" products), they risk a greater loss of access to retailers'

distribution resources than they would if they focused more on traditional (i.e., non-"natural") Olsen, Germann, and Eilert I 32 Marketing Science Institute Working Paper Series new products. Our results indicate category managers view an unproductive non-food brand's

emphasis on "natural" (vs. non-"natural") new products as opportunistic, which decreases the

manager's trust in the brand's use of the "natural" claim and leads them to reduce the brand's

overall distribution.

Implications for Theory

Our research offers several contributions to marketing theory. First, we contribute to

research on "natural" product claims specifically and "green" claims more broadly by

demonstrating how a new product strategy using the "natural" claim is associated with

interorganizational outcomes. Ours is the first study of which we are aware that examines the

B2B implications of "natural" new products. Although extant research examines consumer-level

effects from "natural" (McFadden and Huffman 2017; Rozin 2005) and "green" claims (e.g., Lin

and Chang 2012; Luchs et al. 2010; Olsen et al. 2014), the relationship with interorganizational

exchange outcomes, such as brand distribution, were previously unexplored. For brands relying

on external channel partners to reach consumers, it is critical to understand the unique channel-

level considerations influencing whether, how, and why specific claims influence distribution.

We add to the literature by providing insight into how retailers behave when confronted with

"natural" claims and why they choose to reward some brands' use of the claim while punishing others. "Natural" products may appear to be an appealing option for brands to promote greater exchange with their retailers, but we show it is not always the case.

Second, we contribute to interorganizational exchange theory by investigating how buy-

side partners deal with suppliers possibly engaged in a form of active - though difficult to verify

- opportunism. Extant interorganizational exchange literature tends to focus on blatant, or strong,

forms of supplier opportunism representing clear and objective violations of explicit contracts

(e.g., Masten 1988; Wathne and Heide 2000). However, retail category managers are often

unsure if a brand is using the legally undefined "natural" claim in a way that may ultimately be Olsen, Germann, and Eilert I 33 Marketing Science Institute Working Paper Series seen as misleading if all information were made available. Our research shows buyers use a key

supplier performance metric, shelf space productivity, to inform their perceptions of opportunism

among brands. Shelf space productivity emerges as a critical decision-making heuristic that

category managers use when making brand-level distribution decisions under uncertainty.

Third, our research contributes to scarce literature on distribution's antecedents (Ataman

et al. 2008). By conducting a significant number of in-depth qualitative interviews with category

managers - a critical population of marketing managers not previously interviewed to this extent

in the literature - our grounded theory approach yields important insights into retail category

managers' decision-making process. Among these insights is the finding that shelf space

productivity is one of the most important factors category managers consider when making

brand-level distribution decisions.

Implications for Practice

Our research also contributes to marketing practice. First, we add to the ongoing CPG

industry debate regarding "natural" products' ability to produce positive market-level outcomes

for incumbent brands and whether the products are effective across both food and non-food

categories. The difference between food and non-food categories was highlighted in 2017 during

activist investor Nelson Peltz's ultimately successful quest to gain a seat on the board of

directors at Procter & Gamble (P&G). During the public dispute, Peltz maintained

environmentally-sustainable new products, such as those labeled "natural," are significant

growth drivers for CPG firms, because he had overseen their successful implementation at food-

based companies like H.J. Heinz Co. and Mondelez International, Inc. P&G directors maintained

focusing on such product claims may be appropriate for food categories but not necessarily non-

food categories. As P&G's lead independent director Jim McNemey explained to the Wall Street

Journal, "Either your dishes are clean or they aren't. Or your car is fresh or it isn't. You win

because you have a better functioning brand" (Terlep 2017). Our research suggests both sides are Olsen, Germann, and Eilert I 34 Marketing Science Institute Working Paper Series correct. With respect to the "natural" claim, we find key distinctions should be made between

food and non-food categories when formulating new product strategies.

Second, we propose, validate, and implement a new measure based on differences in

SKU sizes for estimating a brand ' s in-store share-of-shelf and how effectively it uses the space

(i.e., shelf space productivity). As manufacturer and retailer success is dependent on distribution

allocation across brands, marketers on both sides must use the correct metrics to measure the

effectiveness of prior distribution decisions (Ailawadi and Farris 2017). Our qualitative

interviews indicate category managers, brand managers, and sales representatives are interested

in how retail banners allocate shelf space across brands. Many category managers say they

attempt to track how competitors allocate shelf space among brands, with the most common

approach being through ad hoc field audits of competitors' stores. We conduct 54 store audits to

confirm our volume-equivalent measure accurately captures shelf space allocation across stores

in a less labor-intensive manner. The ability to track shelf space and brands' (un)productive use

of that space over time can help inform a category manager's approach to assortment, as well as

a brand manager's approach to retailers in decision areas beyond " natural" new products (e.g.,

how much to invest in trade support programs).

Third, our research has implications for managers ofunderperforming non-food brands.

Extant research demonstrates downward performance spirals are difficult to salvage in an exchange relationship (Wang, Kayande, and Jap 2010). To expand their brand's overall distribution across retailers, managers may see the "natural" claim as a low-cost, low-risk option.

At worst, they may perceive little downside to emphasizing the claim. However, our analyses show it can be a counterproductive new product strategy. For unproductive brands, focusing on

"natural" new products makes the task of maintaining existing distribution even more difficult.

Olsen, Germann, and Eilert I 35 Marketing Science Institute Working Paper Series Post-hoc Analysis

Despite the distribution risk, managers of unproductive non-food brands may

nevertheless wish to proceed with "natural" new product introductions. To partially understand

whether, and how, managers of unproductive non-food brands may mitigate the distribution

penalty their "natural" new products would otherwise receive, we conduct a post-hoc analysis

involving a tactical lever over which brand managers have some control - their brand's relative

presence in the retailer's in-store merchandising programs (i.e., weekly features, displays, and

price reductions). While we control for a brand's level of promotional activity in Study 2, the

post-hoc analysis examines the interactive influence between shelf space productivity, natural

new product prevalence, and a brand's relative category merchandising presence. When making

brand-level distribution decisions, Study 1 found that a brand's effective utilization of the

retailer's shelf space is the most critical consideration. However, retailers also consider other

heuristics when making brand-level distribution decisions. Specifically, category managers also

commonly consider a brand's level of participation in category merchandising programs relative

to its share of category sales. Indeed, when both exchange partners demonstrate relatively high

commitment to the relationship, they are less likely to expect opportunistic behavior from the

other (e.g., Lusch and Brown 1996). Figure 4 summarizes results from the post-hoc analysis.

Appendix 9 provides a table of results and details the modeling approach.

As shown in Figure 4, non-food brands with low shelf space productivity and high

natural new product prevalence suffer more pronounced distribution losses than the other types

of brands. This finding is consistent with the "dark side" implications found in Study 2.

However, the distribution penalty for this type of brand softens as its in-store category merchandising presence increases. Thus, while an unproductive non-food brand focused on

"natural" new products will be hard-pressed to turn its likely distribution loss into a gain,

ensuring the brand's in-store merchandising presence is above its "fair share" should at least Olsen, Germann, and Eilert I 36 Marketing Science Institute Working Paper Series partially offset the steep distribution penalties it would likely encounter otherwise. Interestingly,

Figure 4 also indicates unproductive non-food brands with low "natural" new product prevalence

(dotted gray line) are able to counterbalance the distribution losses they would otherwise incur by ensuring their presence in category merchandising programs is more than their "fair share."

Finally, the relatively flat solid lines in Figure 4 indicate that productive non-food brands do not seem to benefit as much from category merchandising presence as do unproductive brands. This finding supports insights gained during the qualitative interviews (Study 1) where managers indicated shelf space productivity is typically the most important determinant of brand distribution changes. While category merchandising programs are also important, it appears productive brands' distribution levels do not see benefits beyond delivering their "fair share" of in-store category merchandising presence. Therefore, results suggest productive brands may be able to lower their merchandising expenses without adverse effects to distribution.

Limitations and Future Research

While the present research breaks new ground, it has limitations, some of which provide avenues for future research. The focus of our research is on new products making the "natural" claim and how category managers grapple with challenges presented by its undefined nature, particularly in contexts where opportunistic use is difficult to determine. We find category managers tum to an objective measure of brand performance to inform their subjective evaluations and brand-level distribution decisions. Although "natural" is one of the most widely- used undefined claims across various product categories, future research could investigate how brands' use of other undefined claims are evaluated by retailers and how they influence perceptions of opportunism, trust, and exchange levels. We suspect shelf space productivity plays an important role when retailers encounter other ambiguous claims.

We examine how retailers view and respond to brands' "natural" (vs. non-"natural") new product efforts, and the findings expand our understanding of brand distribution's antecedents. Olsen, Germann, and Eilert 137 Marketing Science Institute Working Paper Series

However, we note other possible antecedents are worth studying. For instance, managers likely

also consider a brand's relative contribution to category profits when making brand-level

distribution decisions. We do not have direct access to category profit information, although we

control for price premium as an imperfect proxy. Future research can incorporate product-level

cost data to directly explore how category profit contribution influences brand distribution.

Research in the area would ideally account for different sources of a brand's profit contribution.

For example, do different benefits accrue to brands over-delivering on slotting fees compared to

merchandising allowances?

Across our studies, we find shelf space productivity is an effective way for brands to access further distribution resources. We find it provides the additional brand benefit of lowering category managers' perceptions of opportunistic behavior when evaluating "natural" (vs. non-

"natural") new product introductions, ultimately leading to distribution gains. If unproductive,

brands experience the "dark side" of natural new product prevalence. We expect shelf space

productivity to benefit the brand-retailer exchange in other ways, and future research could

investigate the benefits, in addition to the most effective actions managers can take to help an

ailing brand increase its access to retailer resources.

Olsen, Germann, and Eilert I 38 Marketing Science Institute Working Paper Series Appendix 1 Summary of "Natural" Claim Use across Categories Percentage of all new Rank of "natural" products making a among all possible Category "natural" claim "green" claimsa Food categories Cereal 44.55% 1 Yogurt 41.70% 1 Salty snacks 39.45% 1 Milk 38.59% 2 Soup 36.68% 1 Ketchup 29.14% 1 Butter 26.38% 1 Mustard 26.24% 1 Coffee 18.25% 2 Beer 7.47% 1 All food categories 34.56% 1 Non-food categories Shampoo 30.74% 1 Household cleaners 27.76% 1 Toothpaste 25.66% 1 Deodorant 21.15% 1 Laundry detergent 20.95% 3 Cigarettesb 8.54% 1 Bath tissue 3.79% 4 Blades and razors 2.61% 6 Facial tissue 2.20% 4 All non-food categories 22.02% 1 All categories 30.66% 1

Notes: Numbers are based on all new product activity recorded by Product Launch Analytics from 2001 to 2011. a Ranking is based on use of the "natural" claim compared to 31 other green claims identified within the Product Launch Analytics database by Olsen, Slotegraaf, and Chandukala (2014). b The cigarettes category is included in a robustness analysis.

Olsen, Germann, and Eilert I 39 Marketing Science Institute Working Paper Series Appendix 2 List of Categories and Brands in Sample

Bath tissue Heineken Red Stripe Angel Soft Henry Weinhard' s Rogue Capri Hoegaarden Rolling Rock Charmin Icehouse Samuel Adams Cottonelle Imperial San Miguel Marcal Iron City Saranac Quilted Northern Keystone Schlafly Scott King Cobra Schlitz Kona Sea Dog Labatt Shiner Beer La Crosse Shipyard Abita Lagunitas Shock Top Alaskan Landshark Sierra Nevada Anchor Steam Leinenkugel Singha Anderson Valley Lindemans Smirnoff Asahi Long Trail Smithwick's August Schell / Schell's Lowenbrau Smuttynose Beck's Magic Hat Southern Tier Big Sky Michelob Spaten Blue Moon Michigan Squatter Boulder Mickey's Steinlager Bridgeport Mike's Hard Lemonade Stella Artois Budweiser Miller Summit Busch Modelo Troegs Carlsberg Molson Tsingtao Carolina Moosehead Twisted Tea Coors Murpheys Unibroue Corona Narragansett Victory Dundee New Belgium Warsteiner Estrella Newcastle Widmer Firestone New Holland Yuengling Flying Dog Odell's Zima Foster's Otter Creek Full Sail Pacifico Genesee Paulaner Blades and razors Genny Light Peroni Bic George Killian's Irish Red Pete's Gillette Goose Island Pilsner Urquell Old Spice Gordon Biersch Point Personna Grolsch Presidente Schick Guinness Pyramid Super Max Hacker-Pschorr Redbridge Wilkinson Sword Harpoon Redhook

Olsen, Germann, and Eilert I 40 Marketing Science Institute Working Paper Series Butter Golden Grahams illy Bestlife Grape-Nuts Johann Wulffs Breakstone's Honey Bunches of Oats Luzianne Challenge Honeycomb Marques De Paiva Crystal Farms Kashi Maxwell House Earth Balance Kix Melitta Fleischmann's Life Millstone I Can't Believe It's Not Lucky Charms New England Coffee Butter! Malt-O-Meal Newman's Own Organics Keller's Nature Valley Seattle's Best Coffee Land O'Lakes Nature's Path Starbucks Olivio Oatmeal Crisp Tully's Coffee Parkay Oatmeal Squares White Cloud Coffee Promise Oreo O's Yuban Smart Balance Pebbles Quaker Oats Raisin Bran Deodorant Cereal Rice Krispies Arm& Hammer All-Bran Seitenbacher Arrid Alpha-Bits Selects Avalon Organics Apple Jacks Shredded Wheat Back to Nature Smart Start Ban Barbaras Smorz Certain Dri Bear Naked Special K Coty Bob's Red Mill Sunbelt Crystal Cap'N Crunch Toast Crunch Degree Cascadian Farm Total Cheerios Trix Dry Idea Chex Uncle Sam English Leather Cinnabon Weetabix Gillette Cocoa Puffs Weight Watchers Herbal Clear Cookie Crisp Wheaties Mennen Com Flakes Zoe's Mitchum Com Pops Nature's Gate Count Chocula Old Spice Cracklin' Oat Bran Coffee Power Stick Crispix Arbuckles Right Guard Crunch Cafe Bustelo Secret Crunchy Nut Cafe La Llave Soft N Dri Eggo Chock full o'Nuts Erewhon Don Francisco's Sure Fiber One Eight O'Clock TAG Froot Loops Folgers Tom's of Maine Frosted Flakes Gavina Frosted Mini-Wheats Green Mountain Fruit Harvest Hills Bros.

Olsen, Germann, and Eilert I 41 Marketing Science Institute Working Paper Series Facial tissue Seventh Generation Clover Stometta Farms Kleenex Simple Green Cream O'Weber Puffs Soft Scrub Crowley Scotties Spic and Span Darigold Tilex Dean's Weiman Farmland Household cleaners Whink: Garelick Farms Arm&Hammer Windex Golden Guernsey Bissell X-14 Hershey's Bon Ami Hood Carbona Horizon Organic Citrus Magic Ketchup Kemps Clorox Heinz Knudsen CLR Hunt's Lactaid Comet Red Gold Land O'Lakes De-Solv-It Lifeway Disposer Care Nesquik Drano Laundry detergent Oakhurst Earth Friendly Products All Oberweis Dairy Easy-Off Ariel Organic Valley Ecover Arm & Hammer Parmalat Fabuloso Biokleen Promised Land Fantastik Cheer Purity Festival Country Save Reiter Formula 409 ECOS Roberts Glade Era Schneider's Glass Plus Fab Shamrock Farms Goo Gone Gain Silk Granite Gold Method Skinny Cow Greased Lightening Oxydol Slammers Holloway House Purex Smart Balance Kaboom Seventh Generation Smith's Krud Kutter Sun Stonyfield Lime-A-Way Stremick's Heritage Foods Lysol Tide SunMilk Method WIN Tuscan Dairy Farms Mr. Clean Upstate Farms Mrs. Meyer' s Woolite Murphy Xtra Orange Glo Mustard OxiClean Colman's Pinalen Milk East Shore Pine-Sol Anderson Erickson French's Pledge Barber's Grey Poupon Scott ' s Liquid Gold Borden Griffin's Scrubbing Bubbles Byrne Dairy Gulden's

Olsen, Germann, and Eilert 142 Marketing Science Institute Working Paper Series

Haus Barhyte Kettle Snack Factory Jack Daniel's Kettle Classics SnakKing Koops' Kraft Snyder of Berlin Lance Snyder's of Hanover Morehouse Lay's Southern Recipe Robert Rothschild Farm Lundberg Family Farms Stacy's Stonewall Kitchen Martin's SunChips Terrapin Ridge Farms Mediterranean Snacks SunRidge Farms Woeber's Miguel's TERRA Mikesell's Terrell's Mi Ranchito Tim's Salty snacks Mission Tom's Act II Mr. Krispers Torengos Andy Capp's Munchies Tostitos Barcel Nabisco Troyer Farms Barrel O'Fun Newman's Own Organics Tumaros Boulder Canyon New York Style Turkey Creek Snacks Brad's Oberto Uncle Ray's Brim's Old Dutch Utz Bugles Oogie's Vitner's Cabo Chips Orville Redenbacher's Wege of Hanover Cape Cod Padrinos Wise Cheetos Pepperidge Farm Zapp's ChexMix Pirate's Booty Combos Planters Corazonas Plocky's Shampoo Cracker Jack Poore Brothers ABBA Crunch 'n Munch popchips Alagio Doritos Popcorn, Indiana Alberto VOS Eatsmart Snacks Poppycock American Crew Energy Club Pop Secret ARTec Evans Pringles Aussie Flat Earth Quaker Avalon Organics Food Should Taste Good Riceworks Aveda French's Ricos Aveeno Active Naturals Fritos Robert's American Axe Garden of Eatin' Gourmet Back to Basics Gardetto's Rold Gold Bain de Terre Genisoy Route 11 Biolage Glenny's Ruffles BioSilk Guiltless Gourmet RW Garcia Burt's Bees Hain Salem Baking Co. CHI Herr's Sensible Portions Citre Shine Jays Shearer's Creme ofNature John Wm. Macy's Skinny Denorex Just the Cheese Smartfood Dermarest

Olsen, Germann, and Eilert I 43 Marketing Science Institute Working Paper Series

Desert Essence Sexy Hair Dove Shikai Natural White Finesse Softsheen Orajel Fekkai Suave Oral-B Freeman Pearl Drops Gamier Fructis Te Tao Rembrandt Giovanni ThermaSilk Sensodyne Head & Shoulders TIGI Tom's of Maine Herbal Essences TRESemme Ultrabrite Infusium 23 UltraSwim Zooth Isoplus Vavoom Jason Vidal Sassoon John Frieda Wella Yogurt Johnson's White Rain Anderson Erickson Joico Blue Bunny Kerastase Breyers Kiss My Face Soup Brown Cow KMS Baxters Cabot Creamery L'anza Healthy Choice Cascade Fresh L'Oreal Heinz Chobani Mane 'n Tail Hormel Colombo Marc Anthony Imagine Crowley Motions Juanita's Foods Dannon Naked Naturals Emmi Nature's Gate Lucini Italia Fage Neutrogena Maggi Hiland Nexxus Manischewitz Horizon Organic Nioxin Mori-Nu Karoun OGX Muir Glen Kemps Pampers Kandoo Pacific La Creme Pantene Progresso LaLa Paul Mitchell Snow's La Yogurt Pert Walnut Acres Liberte Physique Wolfgang Puck Mountain High Prell Nancy's Progaine Old Home Psoriasin Toothpaste Redwood Hill Farm Pureology Aim Roberts Redken Aquafresh Stonyfield Revlon Arm&Hammer The Greek Gods Rusk Biotene Voskos Salon Grafix Burt's Bees Wallaby Organic Samy Close Up Weight Watchers Sebastian Professional Colgate Yami Sedal Crest YoCrunch Selsun Jason Yoplait

Olsen, Germann, and Eilert 144 Marketing Science Institute Working Paper Series

Appendix 3 Simple Statistics and Correlation Matrix Variable Mean St. Dev. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) 1. Brand distribution 8.002 11.405 1 2. Natural new product prevalence -.050 .392 -.133 1 3. Shelf space productivity -.266 .409 .393 -.238 1 4. Food (vs. non-food) category .625 .484 -.171 .022 -.168 1 5. Market share .045 .087 .733 -.121 .453 -.204 1 6. New product introduction s 5.315 6.704 .372 .025 .035 -.052 .226 1 7. Price premium .711 1.393 -.229 .097 -.379 -.128 -.229 -.041 1 8. Product line length 38.119 39.979 .712 -.123 .304 -.140 .509 .320 -.156 1 9. Advertising .007 .051 .458 -.046 .289 -.159 .555 .093 -.147 .378 1 10. Promotional activity .005 .045 .223 -.008 .260 -.081 .539 .016 -.010 .040 .251 1 Note: Correlations greater than +/- .05 are significant at p < .05.

0 lsen, Germann, and Eilert I 45 Marketing Science Institute Working Paper Series

Appendix 4 Illustrative Example Comparing Measures of Distribution Breadth

Hypothetical scenario: Two salty snack brands (A, B) competing in a market consisting of four total stores (S1, S2, S3, S4).

Store 1 Store 2 Store 3 Store 4 Total Total store sales, i.e., all commodity groups (quarterly average) 1,500,000 2,000,000 2,500,000 4,000,000 10,000,000 Store percentage of all commodity volume (ACJi; 15% 20% 25% 40% 100% Total salty snack category sales (quarterly average) 125,000 75,000 175,000 125,000 500,000 Store percentage of product category volume (PCJi; 25% 15% 35% 25% 100% Is brand A in distribution at store k during year t? Yes No Yes No ls brand B in distribution at store k during year t? No Yes No Yes

Given the data above, brands A and B compare on distribution breadth as follows: Measure of Distribution Breadth Brand A Brand B Count of stores 2 2 50% Percentage of stores 50% (2 of 4 total) (2 of 4 total) 40% Percentage of ACV 60% (15% + 25%) (20% + 40%) 60% Percentage of PCV 40% (25% + 35%) (15% + 25%)

0 lsen, Germann, and Eilert I 46 Marketing Science Institute Working Paper Series Appendix 5 Illustrative Example: Measuring Brand Distribution via Equivalized SKUs

Brand A Brand B

Brand Distribution Breadth % Product category volume (PCV) 99 99 Distribution Breadth, by SKU, in each Brand's Product Line (Measured via SKU-Level PCV) SKU 1 78 95

SKU2 47 87 SKU 3 43 75 SKU4 34 SKU5 16

Total distribution points (TDPs) 218 257 (sum of each individual SKU's PCV)

218 / 100 = 257 I 100 = Equivalized SKUs (TDP / I 00) 2.18 2.57

Notes: SKU = Stock keeping unit. In this example, both brands have equivalent distribution breadths, as they are both in stores contributing 99% of total category sales in the trading area. While Brand A has a longer product line than Brand B (five SKUs versus three), Brand B's total distribution in equivalized SKUs is greater than Brand A's (2.57 vs. 2.18). Thus, Brand B has 2.57 SKUs in a typical store within the specified trading area on average, while Brand A has 2.18 SKUs. While we calculate "equ ivali zed SKUs " based on PCV weights, we note it can alternati vely be calculated using ACY weights.

Olsen, Germann, and Eilert 147 Marketing Science Institute Working Paper Series

Appendix 6 Description of Control Variables Used in Study 2

Market share. Extant research finds evidence of a positive relationship between market share and distribution (e.g., Bronnenberg, Mahajan, and Vanhonacker 2000; Reibstein and Farris

1995; Wilbur and Farris 2014). It is also important to account for the dependence one partner in an exchange relationship has another (e.g., Jambulingam, Kathuria, and Neven 2011). In our context, retailers' dependence on a high market share brand is arguably greater than their dependence on a low market share brand. Therefore, we control for the brand's share of category sales volume in year t.

New product introductions. New products offer an opportunity for brands to expand usage context in consumers' minds (Aaker 1996), and we expect new products to have a positive association with brand distribution. While our NNPP variable measures the percentage of new products making "natural" claims, we control for the number of new product introductions, measured as the number of new products launched by brand i in year t according to PLA.

Price premium. The price at which products are sold at retail can impact distribution outcomes (Rao and McLaughlin 1989). Moreover, reputation plays a role in exchange relationships (e.g., Jap and Anderson 2003) and our qualitative interviews suggested a brand's relative contribution to category profit is considered when making distribution decisions. While we do not have access to profit data directly, price premium is argued to capture a brand's marketplace equity (e.g., Aaker 1996) and be a driver of margin premiums (e.g., Mam, Roegner, and Zawada 2003). Previous researchers have measured price premium as the percentage price difference between the focal brand and category private label (e.g., Steenkamp, van Heerde, and

Geyskens 2010). The approach assumes consumers consider relative price differences in percentages, rather than absolute dollars (Monroe 1973). We adapt this approach to our context

Olsen, Germann, and Eilert I 48 Marketing Science Institute Working Paper Series by also recognizing what is considered a high or low price should be relative to the focal store's pricing strategy and mix of brands carried. Moreover, not all stores in our sample carry a private label in each category. Therefore, we compare the focal brand's price to the average price in the category within each store.

By using the store's average category price instead of private label price, we include stores without a private label in the focal category. The approach also accounts for private label brands that are strong brands in their own right (e.g., Kroger's Private Selections). We measure price premium as the annual average of the quarterly in-store percentage difference between the focal brand's volume-equivalent average dollar price and the category's volume-equivalent average dollar price, weighted by store PCV. We compare the focal brand's price only to the brands with which it directly competes. We do not compare the focal brand to prices in stores where the focal brand is not sold.

Product line length. While long product lines do not necessarily coincide with more distribution (Ataman et al. 2008), they may put a brand in better position to expand distribution.

To account for the influence of long product lines on distribution, we control for the quarterly average number of unique UPCs sold by brand i in year t.

Advertising: Share of voice. A manufacturer's advertising efforts can positively impact distribution outcomes (Montgomery 1975). We control for share of advertising voice, calculated as a brand's percentage share of the category's total advertising dollars (McAlister et al. 2016) recorded by Kantar Media's Ad$pender during year t. The variable is highly correlated with market share (p = .85, p < .01). Therefore, we adopt a residual approach (e.g., Hennig-Thurau,

Houston, and Heitjans 2009) and use the residuals after market share is regressed on share of advertising voice to capture remaining variance not due to market share.

Olsen, Germann, and Eilert I 49 Marketing Science Institute Working Paper Series Promotional support: Share of activity. Manufacturer channel support programs have a positive impact on distribution intensity (Frazier and Lassar 1996). Bi-lateral relationship specific investments have received considerable attention in the interorganizational exchange literature (e.g., Anderson and Weitz 1992, Wathne and Heide 2000, Williamson 1993), and in- store promotion programs represent a relevant form of bi-lateral relationship specific investments in our research context. While promotional pass-through rates vary considerably across brands and categories (Ailawadi and Harlam 2009), greater promotional funding is generally associated with higher promotional spending. Since we do not have access to actual promotional funding levels, we control for brand i's share of in-store category promotional activity using the average quarterly share(%) of the focal store's in-category promotional activities (i.e., feature, display, and price reductions) for year t, weighted by each store's PCV. We capture the variable as follows: A SKU on display for one week engages in one promotional activity. A SKU on display with a one-week price reduction, defined as a temporary price at least 5% lower than the everyday price, engages in two promotional activities. A SKU also on feature for one week engages in three promotional activities. As shown in Equation 1, the number is then divided by the average count of category promotional activities across all products in the focal store and weighted by the store's PCV: (1) "n ((PromoActivitiesikgt) PCVgt) Promo- = L­ _ x-- it k-l (PromoActivitiesmgt) PCVit '

Where:

PromOit = share of category promotional activities (i.e., feature, display, and price reductions) for brand i during year t; PromoActivitieSikgt = average quarterly count of weekly promotional activities for brand i's product kin store g over year t (e.g., 0 = no promotional activity; 3 = feature, display, and price reduction); PromoActivitiesmgt = average quarterly count of weekly promotional activities for all products in category m in store g over year t; PCVgt = average quarterly % product category volume for store g over year t; and

Olsen, Germann, and Eilert I 50 Marketing Science Institute Working Paper Series PCVir = average quarterly% product category volume for brand i over year t.

Similar to advertising, the resulting variable is highly correlated with market share (p = .89, p <

.01). Therefore, we use the residuals after market share is regressed on share of promotional

activity to capture remaining variance not due to market share.

Olsen, Germann, and Eilert I 51 Marketing Science Institute Working Paper Series Appendix 7 Summary of Robustness Analyses

DV: B rand Dist ributi on

Rl : No R2: Serial Main Se rial Corre lation R3: Split R4: Brand RS: OLS, DV: Variables Hypothes es Mode l Corre lation Correc tion at Half Sample Fixed Effects ln(BrandDis tr) Correc tion AR2-l eve l Main Effe cts and Moderators -.434*** Na tural new product prevalence (NNPP) -.585** -.417** .859 -.529*** -.901*** (.Ill) (.291) (.125) (786) (.125) (.313) . 715*** Shelf space productivtiy (SSP) H1 1. 10 7*** .422*** 2.146*** . 122 I.164*** (.124) (.192) (.134) (.519) (.254) (.298) .302 Food catego ry -. 120 .71 ]*** .249 .579** -.379 (.251) (.249) (.230) (.727) (.247) (.302) .484*** NNPP x SSP 1.712*** .673*** 3.853*** .552** 1.132** (. 142) (.387) (.170) (I.239) (.224) (.484) .022 NNPP x Food category -.006 -.11 6 - 1.528** -.041 . 213 (.086) (.223) (.101) (.701) (.073) (.251) -.303** Food category x SSP -.643** -.248* -1.841*** -.170 -.464 (./42) (.204) (.152) (.492) (.294) (.320) -.327** NNP P x Food category x SSP H 2A/B -1.391 *** -.745*** -3.8 I I * ** -.422** - 1.025** (. / 47 ) (.393) (.175) (/.201) (.216) (.498) Control Variables 8.511*** Market share 4.323*** 9.670*** 3.606 7.305*** 4.632*** (.719) (.935) (.737) (2.469) (J.192) (J.196) 3.513*** Market share x NNPP -1.284 4.157*** l.766 -.656 -1.910 (.962) (/ .762) (l.D21) (9.522) (.432) (2.106) -.000 New product introductions (NPis) .010*** .003** .023* .000 .012 (.001) (.003) (.001) (.013) (.001) (.OJ I) -.003 NPis x NNPP -.009 .011 -.039 .004 .012 (.004) (.007) (.007) (.027) (.003) (.0ll) -.070*** Price premium -.108*** -.112*** -.048 .026 -.122*** (.014) (.033) (.035) (.050) (.042) (.038) .090*** Price premium x NNPP .085* .066*** -.050 .061 .060 (.022) (.051) (.023) (.140) (.043) (.066) .011*** Product line length .018*** .009*** .025*** .006*** .018*** (.001) (.001) (.001) (.003) (.001) (.002) .004*** Product line length x NNPP .015*** .003*** .024** .000 .01 ]*** (.001) (.002) (.001) (.OJ I) (.001) (.004) .687** Advertising .849 .353 -6.461*** .455** .608 (.308) (J.003) (1.097) (J.565) (.206) (J.269) 3.362*** Advertising x NNPP 2.509 .883 3.275 1.338*** 3.638 (.850) (2.172) (1.097) (9.2 IO) (.447) (2.661) -1.100 Promotional activity 1.088 .043 -1.938 -.441 1.1 86 (.774) (J.597) (.870) (2.259) (.872) (J.975) -2.721*** Promotional activity x NNPP -3.764 2.365** -12.566 .220 4.245 (.974) (2.415) (I.105) ( I 0.763) (.734) (3. I 79) Year fIXed effects Inc luded Inc luded In cluded Inc luded Inc luded Inc luded Time trend (Year x NNPP) Included Included Included Included Inc luded Included Category fIXed effects Included Included Included Included Included Included Brand fIXed e ffects Included .286* Constant .584*** .139 1.002* .537*** .311 (.175) (.200) (.172) (.550) (.195) (.270) Account for serial corre lation Yes No Yes (AR2) Yes Yes Yes Brand-year obse rvations 2,575 2,929 2,260 921 2,575 2,929 Brands 451 628 346 157 451 628 Standard errors in parentheses. *** p < .01; **p < .05; *p < . IO. Note: "Ma in Model" corresponds to Model 5 in Table 4.

Olsen, Germann, and Eilert I 52 Marketing Science Institute Working Paper Series Appendix 8 Scale Items and Reliabilities from Experiment with Grocery Industry Managers

All 101 participants were asked to assume they are managing a non-food category for a grocery chain and that they are in the process of conducting a full category review, which will be followed by a full category reset that will reflect any changes they may make to the category's current assortment. Participants were then randomly assigned to one of two treatment conditions, where we manipulated the brand's shelf space productivity:

Please indicate the extent to which you agree with the following statements for an (1) unproductive I (2) productive brand introducing new products to the market that emphasize the claim "natural" much more heavily than other brands in its non­food category. Productivity refers to whether a brand's contribution to category sales is over­ or under­ delivering based on the shelf space it has currently.

The mediating variables and brand distribution were measured using a seven-point scale with anchor points at 1 = strongly disagree and 7 = strongly agree.

Perceived Opportunism (a= .76) Adapted from Jap and Anderson (2003)

When this (1) unproductive I (2) productive brand's new products heavily emphasize the claim "natural, " the brand is ... 1. Making hollow promises. 2. "Window dressing" its efforts to improve. 3. Providing false information. 4. Expecting us to provide more than our fair share of resources to support them.

Trust in the "Natural" Claim (a= .94) Adapted from Beltramini (1982) and Ohanian (1990)

When this (1) unproductive I (2) productive brand's new products heavily emphasize the claim "natural, " the claim is... 1. Credible 2. Trustworthy 3. Sincere

Brand Distribution

Going forward, I will increase the total number of products my stores offer for sale from this (1) unproductive I (2) productive non­food brand which is introducing new products heavily emphasizing the claim " natural. "

Olsen, Germann, and Eilert I 53 Marketing Science Institute Working Paper Series Appendix 9 Post-hoc Analysis: The Role of Category Merchandising Presence in Non-Food Categories

DV: Brand Distribution Variables Model 1 Model 2

Main Effects and Moderators -.025 -.271*** Natural new product prevalence (NNPP) (034) (083) .33 1 ** .256* Shelf space productivity (SSP) (137) (148) . 191 *** .170*** Category merchandis ing presence (CMP) (040) (066) -.077*** -.062 CMP2 (008) (039) .468*** NNPP x SSP (098) .170 NNPP xCMP (137) -.157 SSP x CMP (122) .356 NNPP X SSP X CMP (231) -.248*** NNPP x CMP2 (079) -.062 SSP x CMP2 (059) -.300** NNPP x SSP x CMP2 (121) Control Variables 8.465*** 7.075*** Market share (J.036) (J.463) .611 Market share x NNPP (651) .001 .001 New product introductions (NPis) (001) (001) .003 NPis x NNPP (003) -.080*** -.066*** Price premium (010) (012) .056*** Price premium x NNPP (022) .003*** .002* Product line length (001) (001) -.000 Product line length x NNPP (002) -.134 .253 Advertising (431) (447) 1 .587** Advertising x NNPP (758) Year fixed effects Included Inclu ded Time trend (Year x NNPP) Inclu ded Category fixed effects Included Included .484** .669*** Constant (194) (242)

Wald x2 870 1,181 Sta ndard errors in parentheses. ***p < .01; **p < .05; *p < .10. N = 955 brand-years, 159 brands, 8 non-food product categories. Notes: Referent category-year is bath tissue, 2002. Estimates generated using a panel lo g-linked gamma model.

Olsen, Germann, and Eilert I 54 Marketing Science Institute Working Paper Series We use the same model and estimation procedure in our post-hoc analysis as we do in

Study 2 (i.e., Equation 5), with three notable exceptions. First, given our focus on implications for non-food brands, we estimate the model for non-food brands only, thereby excluding the food category variable and its interactions. Second, the three-way interactions now involve category merchandising presence (CMP) along with the other two focal variables (i.e., natural new product prevalence and shelf-space productivity). Third, extant literature finds an inverted U- shaped relationship between promotional spending and performance (e.g., Jia et al. 2018; Little

1966). Thus, we also include the quadratic main and moderating effects of category merchandising presence (CMP) in our post-hoc analysis. 8 The results are presented in the above table.

8 Results remain consistent when category merchandising presence's linear and quadratic terms are included as control variables in the main model from Study 2.

Olsen, Germann, and Eilert I 55 Marketing Science Institute Working Paper Series REFERENCES (Appendix)

Aaker, David A. (1996), "Measuring Brand Equity Across Products and Markets," California Management Review, 38 (3), 102-20. Ailawadi, Kusum L. and Bari A. Harlam (2009), "Findings - Retailer Promotion Pass-Through: A Measure, Its Magnitude, and Its Determinants," Marketing Science, 28 (4), 782-91. Anderson, Erin and Barton Weitz (1992), "The Use of Pledges to Build and Sustain Commitment in Distribution Channels," Journal of Marketing Research, 29 (1), 18-34. Ataman, M. Berk, Carl F. Mela, and Harald J. van Heerde (2008), "Building Brands," Marketing Science, 27 (6), 1036-54. Beltramini, Richard F. (1982), "Advertising Perceived Believability Scale," in Proceedings of the Southwestern Marketing Association, D.R. Corrigan, F.B. Kraft, and R.H. Ross, eds. Wichita State University, Wichita, KS: Southwestern Marketing Association, 1-3. Bronnenberg, Bart J., Vijay Mahajan, and Wilfried R. Vanhonacker (2000), "The Emergence of Market Structure in New Repeat-Purchase Categories: The Interplay of Market Share and Retailer Distribution," Journal of Marketing Research, 37 (1), 16-31. Frazier, Gary L. and Walfried M. Lassar (1996), "Determinants of Distribution Intensity," Journal ofMarketing, 60 (4), 39-51. Jambulingam, Thani, Ravi Kathuria, and John R. Nevin (2011), "Fairness-Trust-Loyalty Relationship Under Varying Conditions of Supplier-Buyer Interdependence," Journal of Marketing Theory and Practice, 19 (1), 39-56. Jap, Sandy D. and Erin Anderson (2003), "Safeguarding Interorganizational Performance and Continuity under Ex Post Opportunism," Management Science, 49 (12), 1684-1701. Jia, He, Sha Yang, Xianghua Lu, and C. Whan Park, "Do Consumers Always Spend More when Coupon Face Value is Larger? The Inverted U-shaped Effect of Coupon Face Value on Consumer Spending Level," Journal of Marketing, 82 (4), 70-85. Little, John D.C. (1966), "A Model of Adaptive Control of Promotional Spending," Operations Research, 14 (6), 1075-97. Mam, Michael V., Eric V. Roegner, and Craig C. Zawada (2003), "The Power of Pricing," McKinsey Quarterly, 3 (July), 40-49. McAlister, Leigh, Raji Srinivasan, Niket Jindal, and Albert A. Cannella (2016), "Advertising Effectiveness: The Moderating Effect of Firm Strategy," Journal of Marketing Research, 53 (2), 207-24. Monroe, Kent B. (1973), "Buyers' Subjective Perceptions of Price," Journal of Marketing Research, 10 (1), 70-80. Montgomery, David B. (1975), "New Product Distribution: An Analysis of Supermarket Buyer Decisions," Journal ofMarketing Research, 12 (3), 255-64. Ohanian, Roobina (1990), "Construction and Validation of a Scale to Measure Celebrity Endorsers' Perceived Expertise, Trustworthiness, and Attractiveness," Journal of Advertising, 19 (3), 39-52. Olsen, Mitchell C., Rebecca J. Slotegraaf, and Sandeep R. Chandukala (2014), "Green Claims and Message Frames: How Green New Products Change Brand Attitude," Journal of Marketing, 78 (5), 119-37. Rao, Vithala R. and Edward W. McLaughlin (1989), "Modeling the Decision to Add New Products by Channel Intermediaries," Journal of Marketing, 53 (1), 80-88.

Olsen, Germann, and Eilert I 56 Marketing Science Institute Working Paper Series Reibstein, David J. and Paul W. Farris (1995), "Market Share and Distribution: A Generalization, a Speculation, and Some Implications," Marketing Science, 14 (3), Gl90- G202. Steenkamp, Jan-Benedict E.M., Harald J. van Heerde, and Inge Geyskens (2010), "What Makes Consumers Willing to Pay a Price Premium for National Brands Over Private Labels?" Journal of Marketing Research, 47 (6), 1011-24. Wathne, Kenneth H. and Jan B. Heide (2000), "Opportunism in Interfirm Relationships: Forms, Outcomes, and Solutions," Journal of Marketing, 64 (4), 36-51. Wilbur, Kenneth C. and Paul W. Farris (2014), "Distribution and Market Share," Journal of Retailing, 90 (2), 154-67. Williamson, Oliver E. (1993), "Opportunism and Its Critics," Managerial and Decision Economics, 14 (2), 97-107.

Olsen, Germann, and Eilert I 57 Marketing Science Institute Working Paper Series REFERENCES (Main Manuscript)

Ailawadi, Kusum L. and Paul W. Farris (2017), "Managing Multi-and Omni-channel Distribution: Metrics and Research Directions," Journal of Retailing, 93 (1), 120-35. Anderson, Erin and Barton Weitz (1989), "Determinants of Continuity in Conventional Industrial Channel Dyads," Marketing Science, 8 (4), 310-23. Andre, Quentin, Pierre Chandon, and Kelly Haws (2019), "Healthy Through Presence or Absence, Nature or Science? A Framework for Understanding Front-of-Package Food Claims," Journal of Public Policy & Marketing, 38 (2), 172-91. Anstine, Jeffrey (2007), "Organic and All Natural: Do Consumers Know the Difference?" Journal of Applied Economics & Policy, 26 (1), 15-28. Ataman, M. Berk, Carl F. Mela, and Harald J. van Heerde (2008), "Building Brands," Marketing Science, 27 (6), 1036-54. Belk, Russell, Eileen Fischer, and Robert V. Kozinets (2013), Qualitative Consumer & Marketing Research. Thousand Oaks, CA: SAGE Publications. Beltramini, Richard F. (1982), "Advertising Perceived Believability Scale," in Proceedings of the Southwestern Marketing Association, D.R. Corrigan, F.B. Kraft, and R.H. Ross, eds. Wichita State University, Wichita, KS: Southwestern Marketing Association, 1-3. Bloom, Paul N., Gregory T. Gundlach, and Joseph P. Cannon (2000), "Slotting Allowances and Fees: Schools of Thought and the Views of Practicing Managers," Journal of Marketing, 64 (2), 92-108. Bronnenberg, Bart J. and Carl F. Mela (2004), "Market Roll-Out and Retailer Adoption for New Brands," Marketing Science, 23 (4), 500-18. ---,Michael W. Kruger, and Carl F. Mela (2008), "The IRI Marketing Data Set," Marketing Science, 27 (4), 745-48. Coleman, James (1990), The Foundations of Social Theory. Cambridge, MA: Harvard University Press. Consumer Reports (2016), "Consumer Reports Survey Show 73 Percent of Consumers Look for 'Natural' Labels at Grocery Stores - and Many are Unwittingly Misled," press release, (accessed November 21, 2019), [available at https://www.consumerreports.org /media- room/press-releases/2016/05/consumer-reports-survey-show-73-percent-of-consumers- misled-by-natural-labels-at-the-grocery-store /]. Datta, Hannes, Kusum L. Ailawadi, and Harald J. van Heerde (2017), "How Well Does Consumer-Based Brand Equity Align with Sales-Based Brand Equity and Marketing-Mix Response?" Journal of Marketing, 81 (3), 1-20. Dewey, Caitlin (2017), "The Raging Legal Battle over What Makes a Food 'Natural'," The Washington Post, (August 30), (accessed March 21, 2019), [available at https://www.washingtonpos t.com/news/wonk/wp/2017/08/30/the-raging-legal-battle-over- what-makes-a-food-natural/?utm_term=.2857106f0ed3]. Dilthey, Wilhelm (1957), Gesammelte Schriften. Stuttgart: Teubner. Evans, Greg, Blandine de Challemaison, and David N. Cox (2010), "Consumers' Ratings of the Natural and Unnatural Qualities of Foods," Appetite, 54 (3), 557-63. Flint, Daniel J., Robert B. Woodruff, and Sarah Fisher Gardial (2002), "Exploring the Phenomenon of Customers' Desired Value Changes in a Business-to-Business Context," Journal of Marketing, 66 (4), 102-17. Food Marketing Institute (2017), "FMI Measures Food Retail's Significant Economic Impact," (accessed May 21, 2018), [available at https://www.fmi.org /news room/news- archive/view/2017/10/11/fmi-measures-food-retail-s-significant-economic-impact].

Olsen, Germann, and Eilert I 58 Marketing Science Institute Working Paper Series Ganesan, Shank:ar, 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, 47 (2), 361-73. Germann, Frank, Peter Ebbes, and Rajdeep Grewal (2015), "The Chief Marketing Officer Matters!" Journal of Marketing, 79 (3), 1-22. Gielens, Katrijn, Inge Geyskens, Barbara Deleersnyder, and Max Nohe (2018), "The New Regulator in Town: The Effect of Walmart's Sustainability Mandate on Supplier Shareholder Value," Journal of Marketing, 82 (2), 124-41. Glaser, Barney G. and Anselm L. Strauss (1967), The Discovery of Grounded Theory. Chicago: Aldine. Hanssens, Dominique M., Leonard J. Parsons, and Randall L. Schultz (2001), Market Response Models: Econometric and Time Series Analysis, 2nd ed. , MA: Kluwer Academic Publishers. Hardin, James W. and Joseph M. Hilbe (2007). General Linear Models and Extensions, 2nd ed. College Station, TX: Stata Press. Hennig-Thurau, Thorsten, Mark B. Houston, and Torsten Heitjans (2009), "Conceptualizing and Measuring the Monetary Value of Brand Extensions: The Case of Motion Pictures," Journal of Marketing, 73 (6), 167-83. IRI and SPINS (2020), "Supporting the Natural Products Consumer," white paper, (accessed June 9, 2020), [available at https://www.iriworldwide.com/en- US/Insights /Publications/Natural-Products-Performance]. Jap, Sandy D. and Erin Anderson (2003), "Safeguarding Interorganizational Performance and Continuity under Ex Post Opportunism," Management Science, 49 (12), 1684-1701. Johnsen, Michael (2018), "Kroger Hosts 2nd Natural Foods Innovation Summit," Drug Store News, (February 8), (accessed June 16, 2020), [available at https://drugstorenews.com/retail- news/kroger-hosts-2nd-natural-foods-innovation-summit]. Kahneman, Daniel and Amos Tversky (1979), "Prospect Theory: An Analysis of Decision Under Ris k," Econometrica, 47 (2), 263-91. Kaufman, Peter, Satish Jayachandran, and Randall L. Rose (2006), "The Role of Relational Embeddedness in Retail Buyers' Selection of New Products," Journal of Marketing Research, 43 (4), 580-87. Kronman, Anthony T. (1985), "Contract Law and the State of Nature," Journal of Law, Economics, & Organization, 1 (1), 5-32. Lamey, Lien, Barbara Deleersnyder, Jan-Benedict E.M. Steenkamp, and Mamik G. Dekimpe (2012), "The Effect of Business-Cycle Fluctuations on Private-Label Share: What Has Marketing Conduct Got to Do with It?" Journal of Marketing, 76 (1), 1-19. Levinovitz, Alan (2020), Natural. Boston, MA: Beacon Press. Lin, Ying-Ching and Chiu-Chi Angela Chang (2012), "Double Standard: The Role of Environmental Consciousness in Green Product Usage," Journal of Marketing, 76 (5), 125- 34. Little, John D.C. (1998), "Integrated Measures of Sales, Merchandising, and Distribution," International Journal of Research in Marketing, 15 (5), 473-85. Luchs, Michael G., Rebecca Walker Naylor, Julie R. Irwin, and Rajagopal Raghunathan (2010), "The Sustainability Liability: Potential Negative Effects of Ethicality on Product Preference," Journal of Marketing, 74 (5), 18-31. Lusch, Robert F. and James R. Brown (1996), "Interdependency, Contracting, and Relational Behavior in Marketing Channels," Journal of Marketing, 60 (4), 19-38. Masten, Scott E. (1988), "Equity, Opportunism , and the Design of Contractual Relations," Journal of Institutional and Theoretical Economics, 144 (1), 180-95. Olsen, Germann, and Eilert I 59 Marketing Science Institute Working Paper Series Mathys, Juliane, Alexa B. Burmester, and Michel Clement (2016), "What Drives the Market Popularity of Celebrities? A Longitudinal Analysis of Consumer Interest in Film Stars," International Journal of Research in Marketing, 33 (2), 428-48. McCracken, Grant (1988), Qualitative Research Methods Series: The Long Interview, Vol. 13, Newbury Park, CA: SAGE Publications. McFadden, Jonathan R., and Wallace E. Huffman (2017), "Willingness-to-Pay for Natural, Organic, and Conventional Foods: The Effects of Information and Meaningful Labels," Food Policy, 68, 214-32. Montgomery, David B. (1975), "New Product Distribution: An Analysis of Supermarket Buyer Decisions," Journal of Marketing Research, 12 (3), 255-64. ---,Marian Chapman Moore, and Joel E. Urbany (2005), "Reasoning about Competitive Reactions: Evidence from Executives," Marketing Science, 24 (1), 138-49. Nielsen (2015), "Global New Product Innovation Report," white paper, (accessed February 23, 2018), [available at http://www.nielsen.com/content/darn/nielsenglobal /dk/docs/Nielsen%20Global%20New%20 Product%20Innovation%20Report%20June%202015.pdf]. --- (2019), "A 'Natural' Rise in Sustainability around the World," white paper, (accessed July 8, 2019), [available at https://www.nielsen.co m/us/en/insights/article/2019/a-natural- rise-in-sustainability-around-the-world/]. Ohanian, Roobina (1990), "Construction and Validation of a Scale to Measure Celebrity Endorsers' Perceived Expertise, Trustworthiness, and Attractiveness," Journal of Advertising, 19 (3), 39-52. Olsen, Mitchell C., Rebecca J. Slotegraaf, and Sandeep R. Chandukala (2014), "Green Claims and Message Frames: How Green New Products Change Brand Attitude," Journal of Marketing, 78 (5), 119-37. Preacher, Kristopher J., Derek D. Rucker, and Andrew F. Hayes (2007), "Addressing Moderated Mediation Hypotheses: Theory, Methods, and Prescriptions," Multivariate Behavioral Research, 42 (1), 185-227. Rao, Vithala R. and Edward W. McLaughlin (1989), "Modeling the Decision to Add New Products by Channel Intermediaries," Journal of Marketing, 53 (1), 80-88. Research and Markets (2017), "Whole Foods, Trader Joe's, and Natural Channel Grocery Shopping: The Future of Food Retailing," white paper, (accessed November 18, 2019), [available at https://www.researchandmarkets.com /research/rtv3jw/whole_foods]. Rock, Andrea (2016), "Peeling Back the 'Natural' Food Label," Consumer Reports , (January 27), (accessed April 18, 2019), [available at https://www.consumerreports.org /food- safety/peeling-back-the-natural-food-label/]. Rozin, Paul, Mark Spranca, Zeev Krieger, Ruth Neuhaus, Darlene Surillo, Amy Swerdlin, and Katherine Wood (2004), "Preference for Natural: Instrumental and Ideational/Moral Motivations, and the Contrast between Foods and Medicines," Appetite, 43 (2), 147-54. --- (2005), "The Meaning of "Natural": Process More Important than Content," Psychological Science, 16 (8), 652-58. ---,Claude Fischler, and Christy Shields-Argeles (2012), "European and American Perspectives on the Meaning of Natural," Appetite, 59 (2), 448-55. Schleiermacher, Friedrich (1998 I 1838), Hermeneutics and Criticism and Other Writings. New York, NY: Cambridge University Press. Schmansky, Sarah (2019), "The Golden Growth Opportunity for U.S. Drugstores," Nielsen, (September 19), (accessed October 28, 2019), [available at https://www.nielsen.corn/us /en/insights/article/2019/the-golden-growth-opportunity-for-u-s- drugstores/]. Olsen, Germann, and Eilert I 60 Marketing Science Institute Working Paper Series Seggie, Steven H., David A. Griffith, and Sandy D. Jap (2013), "Passive and Active Opportunism in Interorganizational Exchange," Journal of Marketing, 77 (6), 73-90. Sorescu, Alina B. and Jelena Spanjol (2008), "Innovation's Effect on Firm Value and Risk: Insights from Consumer Packaged Goods," Journal of Marketing, 72 (2), 114-32. Spiggle, Susan (1994), "Analysis and Interpretation of Qualitative Data in Consumer Research," Journal of Consumer Research, 21 (3), 491-503. Srinivasan, Shuba, Marc Vanhuele, and Koen Pauwels (2010), "Mind-Set Metrics in Market Response Models: An Integrative Approach," Journal of Marketing Research, 47(4), 672-84. Sweeney, Jennifer (2019), "Report: Natural Products Growth Outpaces Total Food and Beverage," Grocery Dive, (September 9), (accessed September 19, 2019), [available at https://www.grocerydive.com/news/report-natural-products-growth-outpaces-total-food-and- beverage/562353/]. TABS Analytics (2016), "The Most Useful Sales & Marketing Metrics that are Unknown - Part l," (accessed August 10, 2017), [https://www.tabsanalytics.com /blog/the-most-useful-sales- marketing-metrics-that-are-unknown]. Terlep, Sharon (2017), "Procter & Gamble vs. Nelson Peltz: A Battle for the Future of Big Brands," The Wall Street Journal, (October 8), (accessed March 11, 2019), [available at https://www.wsj.com/articles/p-g-vs-nelson-peltz-a-battle-over-the-future-of-big-brands- 1507485229]. Ulaga, Wolfgang and Andreas Eggert (2006), "Value-Based Differentiation in Business Relationships: Gaining and Sustaining Key Supplier Status," Journal of Marketing, 70 (1), 119-36. Urban, Glen L. and John R. Hauser (1980), Design and Marketing of New Products. Englewood Cliffs, NJ: Prentice-Hall, Inc. US-HHS-FDA (2015), "Use of the Term "Natural" in the Labelling of Human Food Products; Request for Information and Comments," Federal Register, 80 (218), 69905-09. van Lin, Arjen, and Els Gijsbrechts (2014), "Shopper Loyalty to Whom? Chain versus Outlet Loyalty in the Context of Store Acquisitions," Journal of Marketing Research, 51 (3), 352- 70. Wang, Qiong, Ujwal Kayande, and Sandy Jap (2010), "The Seeds of Dissolution: Discrepancy and Incoherence in Buyer-Supplier Exchange," Marketing Science, 29 (6), 1109-24. Wathne, Kenneth H. and Jan B. Heide (2000), "Opportunism in Interfirm Relationships: Forms, Outcomes, and Solutions," Journal of Marketing, 64 (4), 36-51. Wilbur, Kenneth C. and Paul W. Farris (2014), "Distribution and Market Share," Journal of Retailing, 90 (2), 154-67. Williamson, Oliver E. (1975), Markets and Hierarchies: Analysis and Antitrust Implications. New York, NY: The Free Press. ---(1979), "Transaction-Cost Economics: The Governance of Contractual Relations," The Journal of Law & Economics, 22 (2), 233-61. ---(1985), The Economic Institutions of Capitalism. New York, NY: The Free Press. --- (1993), "Opportunism and Its Critics," Managerial and Decision Economics, 14 (2), 97- 107.

Olsen, Germann, and Eilert I 61 Marketing Science Institute Working Paper Series

FIGURE 1 Estimating Brand Shelf-Share with Scanner Data: Illustrative Example and Explanation of Measurement Approach How to Estimate Brand Shelf Space with Scanner Data Brand i's share of in-store shelf space is operationalized as the sum of volume-equivalent measurement units (e.g., a six-pack of 12- ounce bottles) for all UPCs sold by the brand in store g during time period t, divided by the sum of volume-equivalent measurement units for all UPCs sold in the focal category in store g during the same time period The values are weighted by each store's product category volume and summed to create an aggregate estimate of brand i's share of in-store linear shelf space during time period t.

Notes: The example represents a situation in which the measure is peifectly aligned with actual shelf­share, because all SKUs have the same number of shelf facings. Our measure is validated against real store planograms through 54 in­store audits and compared to brands' actual shelf­ share in 18 categories across three grocery formats (mass­merchandiser, mid­tier grocer, and small­format specialty grocer). The correlation between actual and estimated shelf­share is .969 on average ((J = .028, range: .900­1.000). The example compares the measures with a share­ of­SKUs estimate.

0 lsen, Germann, and Eilert I 62 Marketing Science Institute Working Paper Series

FIGURE2 Moderating Effect of Shelf Space Productivity and Category Type on the Relationship between Natural New Product Prevalence and Brand Distribution

Notes: Brand distribution gain/loss reflects the change in equivalized SKUs for a brand with a current distribution of8.0, the mean level in our dataset. To help visualize the relationship, "low" and "high" levels are set at the following: For shelf space productivity, low = 75% underperformance and high = 75% overperformance; for natural new product prevalence, low = 50% below the rest of the category and high = 50% above the rest of the category.

0 lsen, Germann, and Eilert I 63 Marketing Science Institute Working Paper Series

FIGURE3 Mediation Results from Experiment with Grocery Industry Managers for a Non-Food Brand Emphasizing "Natural" New Products

=.83***

and ibution

Notes: ***p < .01; **p < .05; *p < .10. N = 101 managers familiar with category management at U.S. grocery retailers.

0 lsen, Germann, and Eilert I 64 Marketing Science Institute Working Paper Series FIGURE4 Moderating Effect of Natural New Product Prevalence and Shelf Space Productivity on the Curvilinear Relationship between Category Merchandising Presence and Brand Distribution

Notes: Brand distribution gain/loss reflects the change in equivalized SKUs for a brand with a current distribution of 9.9, the mean level for non-food brands in our dataset. To help visuali ze the relationship, "low" and "high" levels are set at the following: For shelf space productivity, low= 75% underperformance and high= 75% overperformance; for natural new product prevalence, low= 50% below the rest of the category and high = 50% above the rest of the category. The X-axis provides a range of values for category merchandising presence, which is measured as follows : Percentage to which brand i' s average quarterly share of all in-store merchandising programs (i.e., weekly feature, display, and price reductions) in category m is above or below its share of in-store category sales. For example, if a brand comprises 15% of category merchandising programs and contributes 10% to in-store category sales, its category merchandising presence is +50%.

Olsen, Germann, and Eilert I 65 Marketing Science Institute Working Paper Series

TABLE 1 Descriptive Statistics of Qualitative Interviews Number Individuals interviewed: 30 Direct experience in the following areas: Category management 20 Retail senior executive management 11 Brand-side category analyst (e.g., category captain) 12 Unique retailers for which the sample has direct category management experience: 35

Direct experience in the following categories: Percent Food categories 93% Non-food categories 87% Both food and non-food categories 80% Retailer size (current employer, if working for a retailer):

Large (1,000+ stores) 38% Medium (100-999 stores) 42% Small (less than 100 stores) 21% Headquarters location (current employer's region of the U.S.): Northeast 20% South 13% Midwest 60% West 7% Company size (current employer's annual revenue): Large ($1OB+) 40% Medium ($1B-$10B) 40% Small (<$1B) 20% Note: Annual company revenue based on publicly available estimates.

Olsen, Germann, and Eilert I 66 Marketing Science Institute Working Paper Series TABLE2 Operationalization of Variables in Secondary Data Analysis Variable Symbol Operationalization Source Brand EqSKUit Quarterly average of brand i's equivalized SKUs in year t, IRI distribution interpreted as the average number of brand i's SKUs sold per store across a trading area, weighted by each store's percentage of category volume (PCV). Specifically, equivalized SKUs are measured as brand i's total distribution points (i.e., sum of the quarterly average % PCV for each SKU k for brand i across year t) divided by 100.

Natural new Percentage of brand i's new products making a "natural" claim in Product product year t minus the percentage of the rest of the focal category's new Launch prevalence products making the claim in year t. Analytics Category type FoodCati Dummy variable set to 1 if brand i belongs to a food category; Coded by non-food categories are set to 0. authors Shelf space Percentage to which brand i delivers more or less category sales IRI productivity relative to its in-store category share-of-shelf.a Calculated as the ratio of brand i's quarterly average share in year t of in-store category volume sales, weighted by each store's PCV, to its quarterly average share of in-store category shelf space, minus one. For example, if a brand contributes 15% of category sales and occupies 10% of the in-store shelf space devoted to a particular category, its shelf space productivity is 50%. Control Variables Market share MktShareit Brand i's percentage share of category sales volume in year t. IRI New product NPlit Number of new products launched by brand i in year t captured Product introductions by Product Launch Analytics (PLA). PLA tracks only products Launch considered to be new and unique additions to a brand's product Analytics line from the consumer's perspective. Price premium PricePremit Quarterly average across year t, weighted by each store's PCV, of IRI brand i's in-store price premium (% difference) compared to the volume-equivalent dollar price of all products sold in the category by the focal store in each quarter during year t. Product line PLLit Quarterly average of number of unique UPCs sold by brand i in IRI length year t. Advertising: Advertit Residual of regressing market share on share of advertising voice Kantar Share of voice (i.e., brand i's percentage of the category's total advertising Media's dollars in year t). Ad$pender Promotional Promoit Residual of regressing market share on brand i's average quarterly IRI support: Share share(%) of the focal store's in-category promotional activities of activity (i.e., weekly feature, display , and price reductions), weighted by each store's PCV, in year t. a More details on the measure for in-store category share-of-shelf can be found in Figure 1. Note: In addition to the variables described here, we control for time- and category-effects via year and category dummy variables as well as brand fixed-effects in a robustness analysis .

Olsen, Germann, and Eilert I 67 Marketing Science Institute Working Paper Series TABLE3 Field Evidence of Different Measures' Accuracy of Estimating Brands' In-Store Share of Category Shelf Space

Retail format Retail format Category cha racterist ics and Small-format Category characte ristics and Small- forma t measures for estimating Mass Mid-tier specialty measures for estimating Mass Mid-tier specialty Category brands' in-s tore shelf space merc handiser grocer grocer Category brands' in-store shelf space merchandiser grocer grocer Total linear feet 124 82 19 Tota l linear feet 28 21 7 Number of brands 9 9 2 Number of brands 5 I I 10 Bath Tiss ue Ketchup Share-of-SKUs' acc uracy 0.940 0.958 1.000 Share-of-SKUs' a ccuracy 0.997 0.995 0.918 Volume-equiva lent's accuracy 0.964 0.976 1.000 Volume-eq uivale nt's accuracy 0.999 0.998 0.942 To tal linear feet 217 352 95 Total linear feet 176 140 26 Number of brands 71 114 69 Laundry Number of brands I I 13 9 Beer Share -of-SKUs' accuracy 0.900 0.949 0.972 dete rgent Share-of-SKUs' accuracy 0.993 0.997 0.874

0.919 0.973 0.944 0.993 0.996 0.955 Volume-equivalent's accuracy Volume-equiva te nt's accuracy Total linear feet 80 43 1 Total linear feet 90 165 78 Blades & Num ber of brands 8 5 2 Number of brands 1 4 1 7 1 7 Milk razors Share-of-SKUs' acc uracy 0.973 0.997 1.000 S hare-of-SKUs' accuracy 0.987 0.908 0.998 Volume-equivalent's accuracy 0.975 0.951 1.000 Volume-equivalent's accuracy 0.996 0.957 0.998 Total linear feet 33 22 3 Total linear feet 24 30 12 Numb er of brands 14 10 2 Number of brands 11 21 12 Butter Must ard Share-of-SKUs' acc uracy 0.896 0.993 1.000 Share-of -SKUs' accuracy 0.928 0.893 0.950 Volume-eq uivalent's accuracy 0.915 0.980 1.000 Volume-equivalen t's acc uracy 0.970 0.936 0.920

Total linear feet 258 207 127 Tota l line ar feet 262 475 239 Number of brands 55 56 25 Number of brands 32 43 45 Cerea l Salty snacks Share-of-SKUs' accuracy 0.965 0.973 0.998 Share-of-SKUs' accuracy 0.998 0.910 0.999 Volume-e quivale nt's acc uracy 0.957 0.954 0.999 Volume-equivalent's acc uracy 0.994 0.928 0.996

Total linear feet 96 96 72 Tota l linear feet 167 78 47 Nmnber of brands 17 16 18 Nmnber of brands 52 32 20 Coffee Shampoo Share-of-SKUs' acc uracy 0.960 0.909 0.914 Share-of-SKUs' accuracy 0.989 0.994 0.973 Volume-equivalent's accuracy 0.945 0.929 0.954 Volume-equivalent 's acc uracy 0.963 0.952 0.955

Total linear feet 89 62 16 Total linear feet 193 139 46 Number of brands 21 16 14 Num ber of brands 7 II 4 Deodo rant So up Share-of-SKUs' accuracy 0.985 0.983 0.957 Share-of-SKUs' accuracy 0.999 0.995 0.992 Volume -equ ivale nt's accuracy 0.971 0.983 0.946 Vo lume-equival ent's accuracy 0.995 0.996 0.992 Total linear feet 68 48 4 Total linear feet 135 72 23 Numbe r of brands 3 3 2 Number of brands 14 12 15 Facial tiss ue Toothpaste Share-of-SK Us' accuracy 1.000 0.989 1.000 Share-of-SK Us' accuracy 0.984 0.992 0.993 Vo lume -equivale nt's accuracy 0.977 0.996 1.000 Volume -equivale nt's accuracy 0.998 0.996 0.989 Total linear fe et 110 95 20 Total line ar feet I l l 129 86 Household Number of brands 24 42 II Number of brands 17 15 20 Yogurt clea ners Share-of-S K Us' accuracy 0.903 0.981 0.930 Share-of-SK Us ' accuracy 0.969 0.965 0.929 Volume-equivale nt's accuracy 0.900 0.973 0.980 Volume-equivale nt's accuracy 0.955 0.965 0.917

Notes: "Ac curacy" refers to the correlation of brands ' actual share of in-store category shelf space with estimates from the "s hare ofSKUs" or "vo lume-equivale nt" meas ure. Across all 54 category-store audits, the average correlation between a "share of SKUs" estimate and brands' actual s hare of catego ry shelf space is .966 (CJ = . 036, range: .874-1.000). The average correlation betwee n the volume-equivalent est im ate used in Study 2 and brand shelf space is .969 (CJ = .028, range: .900-1.000).

0 lsen, Germann, and Eilert I 68 Marketing Science Institute Working Paper Series

Table 4 Model Comparisons: Relationship between Natural New Product Prevalence and Brand Distribution

DV: Brand Distribution Variables Hypotheses Model 1 Model 2 Model3 Model 4 Models Main Effects and Moderators -.022 Natural new product prevalence (NNPP) -.177 .041 -.140** -.434***

(016) (290) (034) (058) (111)

.296*** Shelf space productivity (SSP) 3.284*** .332*** .366*** .715*** H1 (088) (219) (153) (135) (124) .385 Food category -.630*** .377 .317 .302 (289) (150) (284) (280) (251) 2.085*** .335*** .380*** .484*** NNPP xSSP (541) (081) (135) (142)

NNPP x Food category .358 -.036 -.100** .022

(347) (040) (044) (086)

Food category x SSP -1.874*** -.024 -.050 -.303** (272) (184) (163) (142)

-1.712*** -.299*** -.353*** -.327** NNPP x Food category x SSP H 2N B (636) (088) (087) (147) Control Variables 10.441*** Market share 10.559*** 10.213*** 8.511*** (1.060) (988) (857) ( 719)

3.513*** Market share x NNPP (962) .000 New product introductions (NPis) .000 .000 -.000

(001) (001) (001) (001) NPis xNNPP -.003

(004) -.055*** -.052*** -.061*** -.070*** Price premium (.011) (.012) (OJ 1) (014)

.090*** Price premium x NNPP (022) .005*** Product line length .005*** .006*** .011*** (.001) (001) (.001) (.001)

.004*** Product line length x NNPP (.001) -.069 Advertising -.160 -.222 .687**

(.221) (.233) (.244) (.308) 3.362*** Advertising x NNPP (.850) -.993 Promotional activity -1.148 -1.041 -1.100 (.855) (851) (.803) (.774)

-2.721*** Promotional activity x NNPP (974)

Year fixed effects Included Included Included Included

Time trend (Year x NNPP) Included Included Category fixed effects Included Included Included Included .207 Constant 2.251*** .196 .237 .286* (.184) (.105) (175) (.161) (.175)

Waldx,2 441 385 520 698 1,590

Standard errors in parentheses. ***p < .01; **p < .05; *p < .10. N = 2,575 brand-years, 451 brands, 18 product categories. Notes: The referent category-year is bath tissue, 2002. Estimates are generated using a panel log-linked gamma model. Olsen, Germann, and Eilert I 69 Marketing Science Institute Working Paper Series