ANALOGICAL LEARNING AND CATEGORICAL IDENTITY DURING MARKET EMERGENCE
Jesper B. Sørensen and Mi Feng
ABSTRACT
We examine how the organizational identity of established firms affects their strategic outcomes during the emergence phase of a new market. Drawing on cognitive theories of analogical learning, we build theory about how the established identities of producers influence the fluency with which consumers make sense of novel products, and hence affect valuations. We illustrate this theory through an empirical study of consumer evaluations of de alio entrants during the emergence of the digital camera industry.
Keywords: Analogical learning; innovation; market entry; producer identity; market emergence
INTRODUCTION
The emergence of new product categories and markets often begins with technical innovations that combine existing ideas and technologies in novel ways (Fleming & Sorenson, 2001; Schumpeter, 1950). While many recombinant innovations may be subsumed within established markets, new markets are more likely to emerge when the prior ideas and technologies exist in technological domains or product categories thought to be distinct and separate. The fact that such innovations draw from different domains often creates substantial challenges, however, for the commercialization process, and hence for the viability of the emergent category. These challenges increase with the innovation’s novelty; the greater the degree to which new products borrow from different and distant domains, the greater difficulties consumers may have with comprehension and valuation. As Rosa, Porac, Runser-Spanjol, and Saxon (1999, p. 68) note, “When new products and new uses for old products are significantly different from existing knowledge structures … behavioral adjustments must be made and conceptual systems are destabilized.”
Such situations present firms possessing expertise and experience in the pre-existing domains with the opportunity to leverage existing competencies into an emerging domain. As a result, the process of market emergence may bring together former specialists in the established markets and technologies, resulting in “shifting industry boundaries and the convergence of firms from previously distinct domains” (Benner & Tripsas, 2012, p. 277). A substantial body of research in organizational theory has focused on how the behavior and fates of such de alio entrants differs from entrepreneurial start-ups, or de novo entrants (Carroll & Hannan, 2000; see Khessina & Carroll, 2008 for a review), and has shown that de alio producers bring distinct competencies (Carroll, Bigelow, Seidel, & Tsai, 1996; Helfat & Lieberman, 2002; Klepper & Simons, 2000) and perspectives (Benner & Tripsas, 2012) to the emerging market.
While this literature has primarily focused on the differences between de alio and de novo entrants, differences among de alio entrants have also been studied. The general approach in such studies has been to emphasize the ways in which differences in industry of origin shape the resources and capabilities that entrants bring to the new market. For example, Carroll et al. (1996, p. 121) suggest that entrants from industries with “relevant specialized transferable skills and knowledge” will have lower mortality rates than other entrants. Similarly, Helfat and Lieberman (2002, p. 727) suggest that “it is the match between the market entered and the firm’s pre-entry resources and capabilities that matters.”
By virtue of their market experience in existing product categories, de alio entrants also bring established producer identities to the emerging market. We posit that such identities are consequential, since a producer’s identity “serves as handles for frames of reference that audiences use in their evaluations” (Negro, Koçak, & Hsu, 2010, p. 17). A substantial literature in economic sociology has in recent years demonstrated the impact of identity on market processes, particularly with respect to valuation (Hannan, Pólos, & Carroll, 2007; Podolny, 1993; Zuckerman, 1999). As in the case of resources and capabilities, it is reasonable to expect that the “match” between the origin identity of an established firm and the emerging market shapes its performance. This is particularly true given that emerging markets generally lack well-established performance metrics and high-consensus dimensions of valuation. But how is that “match” to be conceptualized? Existing theory in strategy and economic sociology provides little guidance.
In this article, we build theory about how existing identities influence the ability of producers to navigate the commercialization process in emerging markets defined by recombinant innovation, and illustrate this theory through an empirical study of consumer evaluations during the emergence of the digital camera industry. Digital cameras combined technologies from computers and consumer electronics with technologies from traditional film photography. As a result, digital cameras were a potentially appealing new market for firms with expertise in these different industries, and entrants into the nascent digital camera market was almost exclusively by de alio firms from these different source industries (Benner & Tripsas, 2012). Importantly, the components for digital cameras could be easily sourced from well-established suppliers; as a result, entrants differed little in their relevant capabilities.
A critical strategic challenge for entrants into this emerging market was that digital cameras created a need for sensemaking among consumers faced with a novel product that combined features of a film camera, a computer, and a consumer electronics device. Indeed, studies from the early history of the industry show that there was substantial ambiguity among consumers about how to best categorize a digital camera (Moreau, Markman, & Lehmann, 2001). Similarly, Benner and Tripsas (2012) demonstrate that producers from different industries conceived of the nascent industry in very different ways. It was therefore not obvious that the digital camera would be viewed as a “camera” rather than, say, a computer peripheral; this is the ambiguity that creates an opening for the identities of established firms to play a role in market definition and performance. The fact that digital cameras are viewed as “cameras” today is the outcome of processes that we seek to explain here.
Our focus is on deepening our conceptual understanding of whether and how the fates of de alio firms from different source industries differ during the process of market emergence, and how such identity-based effects may change as the market matures.1 In order to develop intuitions concerning the effects of different pre-existing categorical identities in this market, we draw on research on categorization and valuation processes (Moreau, Lehmann, & Markman, 2001; Sujan, 1985). In particular, we focus on models of analogical learning (Gentner, 1989; Gregan-Paxton & John, 1997; Gentner & Smith, 2012) for insights concerning how fluency in mapping and inference between an established domain and a novel domain shapes comprehension and valuation.
ESTABLISHED IDENTITIES IN AN EMERGING MARKET
Technological change destabilizes existing markets by posing adaptive challenges for both producers and consumers. The disruptive potential of new innovations is particularly high when innovations draw on and combine distinct technology and product domains. Often, such recombinant technologies form the basis for the emergence of new markets with new parameters of competition. For example, the cellular phone combined technologies from radio transmission, consumer electronics, and telephones. Similarly, the digital camera arose from the combination of technologies from computers and consumer electronics with elements from film photography. As a result, recombinant innovations may constitute both an opportunity and a threat for existing firms; they present firms with expertise in the technologies borrowed from the original domains with substantial opportunities for growth, as well as the potential threat of substitution.
A substantial literature in organizational theory and strategy has examined the organizational factors that shape the disruptions that established firms may encounter in the face of technical change. This literature has emphasized the difficulties due to constraints arising from a firm’s existing competencies, resource investments, organizational structures and processes, and managerial cognition (Henderson & Clark, 1990; Tripsas & Gavetti, 2000; Tushman & Anderson, 1986). Only recently, however, have scholars begun to examine the role of organizational identity in technological transitions, despite the fact that “even seemingly minor shifts from a technological standpoint may challenge the existing organizational identity” (Tripsas, 2009, p. 441) and that the identity may shape the responses of both internal and external organizational constituents.
A central component of an organization’s identity derives from the products it offers; these offerings locate the firm within the pre-existing conceptual schemes and interpretations of audiences of consumers and critics (Hannan et al., 2007). This categorical identity may have a particularly strong influence on the ability of established firms to take advantage of innovations that build on the technologies and features they already know from their existing industry. From a technical perspective, de alio firms should generally have an advantage when an innovation builds on features with which they are familiar (Carroll et al., 1996; Helfat & Lieberman, 2002). However, established firms bring not only their technical competencies to the new product, but also their organizational identity in the eyes of buyers and critics; the firm’s offering is viewed through the lens of their established identity.
This established identity is particularly consequential early in the history of a new technology, when consumers are struggling with “both the formation of concepts with which to understand the product, and the development of criteria to be used in evaluation” (Clark, 1985, p. 244). When consumers lack a clear set of evaluative criteria, the success of a de alio firm’s offering in the new arena depends on whether audiences can make sense of the new product and value it appropriately within their established conceptual systems that link categories of products to different uses (Hargadon & Douglas, 2001; Rindova & Petkova, 2007; Rosa et al., 1999; Tripsas, 2009).
DE ALIO IDENTITY AND ANALOGICAL LEARNING
The organizational identities of existing firms are important in the sensemaking process during market emergence because they activate consumers’ existing product category knowledge, which in turn shapes their responses to the de alio firm’s novel product. When they assess new products, the prior knowledge of consumers affects both their ability to comprehend the product, as well as their assessments of the pros and cons of the product (Moreau, Lehmann, et al., 2001; Rosa et al., 1999; Sujan, 1985). Audiences have an easier time comprehending and valuing a product to the extent that they can interpret it within their existing category structures: “if a new stimulus can be categorized as an example of a previously defined category, then the affect associated with the category can be quickly retrieved and applied to the stimulus” (Sujan, 1985, p. 31). On the other hand comprehension and appeal suffer if the new product forces consumers to process and make sense of discrepant information relative to their existing categorical understandings.
We argue that the identities of de alio producers influence the nature of the comparisons that audiences make when trying to comprehend and value a novel product. Psychologists and students of consumer behavior argue that the sensemaking around a novel product occurs through an analogical learning process that involves the transfer of knowledge from a base domain (e.g., a well-understood product category) to the target domain (e.g., the novel product). This process occurs through three stages: access, mapping, and transfer (Gentner & Smith, 2012). The access stage involves the activation of an existing mental representation (the base domain) that is used to structure inferences about the novel product. For example, a smartphone may be compared to the consumer’s existing knowledge of laptop computers.
Once a base domain has been accessed, learners attempt in the mapping phase to create a one-to-one correspondence between the characteristics of the base domain and the target, in terms of similarity in either attributes or of relations between attributes (Gregan-Paxton & John, 1997; Moreau, Lehmann, et al., 2001). Thus a smartphone may share with a laptop the attribute of possessing a screen with a graphical user interface, and the relation of the screen allowing the individual to read email messages.
Finally, once mapping has occurred, additional knowledge from the base domain is applied to the target, leading to more general inferences about the novel product under the assumption that domains similar in certain respects (as established in the mapping phase) will also be similar in other respects. Thus based on the observed similarities between smartphones and laptop computers, the consumer may infer that the smartphone can also be used to browse the internet, compose documents, etc.
This framework suggests several points at which the organizational identities of de alio producers may influence sensemaking around innovative products, impact comprehension and valuation, and ultimately determine the acceptance of a firm’s offering.
ACTIVATION OF A BASE DOMAIN: ATTRIBUTES AND CATEGORY LABELS
Consider first the access stage. A firm’s success or failure in an emerging market depends on whether the new product activates a base domain that facilitates comprehension and valuation. For highly innovative products, it is not clear ex ante which base domain will be activated, and in the absence of a clear organizational identity for the producer the new product may fail to effectively activate any base domain. Consumers may then switch to piecemeal evaluation of the product’s attributes, but this has been shown to require greater cognitive effort than categorical evaluation (Sujan, 1985). As a result, the new product is more likely to be ignored.
In the case of recombinant innovations, multiple existing knowledge structures have the potential to be accessed. Which of these base domains is activated sets the terms for subsequent sensemaking. For example, since smartphones combine elements of both cellphones and laptop computers, both product categories may serve as plausible base domains. The individual’s comprehension and valuation of the smartphone depends on which base domain is activated.2
Access is a function of (1) common attributes between the base and target domains (Gregan-Paxton & John, 1997; Nisbett & Ross, 1980) and (2) category labels (Moreau, Markman, et al., 2001; Yamauchi & Markman, 2000). Firms can, therefore, through product design, marketing, and sales efforts attempt to shape which base domain is activated. Choices concerning the physical design (such as form, materials, and color) can facilitate access to a particular base domain by emphasizing common attributes with known product categories, and hiding or obscuring potentially discrepant features (Hargadon & Douglas, 2001; Rindova & Petkova, 2007; Tripsas, 2009).
A particularly powerful means of shaping access to a base domain is to associate the novel product with a category label, rather than, or in addition to, emphasizing common attributes. Category labels exert a discrete influence on learning that is qualitatively different from the effects of shared attributes. Emphasizing category labels encourages individuals to perceive the object as a whole (rather than assess attributes in a piecemeal fashion) and focuses attention on features within the category, while directing attention away from discrepant information (Moreau, Markman, et al., 2001; Yamauchi & Markman, 2000; yamauchi 2008 category). As noted by Yamauchi and Markman (2000, p. 792), “information about category membership molds the way people infer a characteristic of an object … people pay particular attention to the feature value most prevalent in the members of that corresponding category.” As a result, expectations for a new product will be inferred from the products in the category invoked by the category label.
For example, Moreau, Markman, et al. (2001) primed participants through category labels to associate a digital camera either with the film camera category (“works like a camera”) or with the computer peripheral category (“works like a scanner”). (Note that the study was completed before digital cameras were a well-defined product category, when they did not fit unequivocally into any existing product categories.) Participants were then asked to identify where, on a map of a hypothetical electronics store, they would expect to find the product, and to assess their expectations for picture quality by choosing between printed photos of varying pixel density. Participants exposed to the film camera category were more than twice as likely to categorize the product as a camera than a computer peripheral, indicating the importance of the initial labeling for comprehension. Furthermore, labeling shaped valuation (through categorization): participants who categorized the product as a camera had significantly higher expectations for the quality of the photographs it would produce than participants who categorized it as a computer peripheral.
Similarly, in a case study of a successful entrepreneurial venture, Tripsas (2009) illustrates how one of the original producers of flash memory products for digital cameras (“Linco”) deliberately shaped the analogical learning of consumers, retailers and investors by positioning their product as “digital film”:
Instead of calling the product a flash memory card like the competition did, Linco called it “digital film” and put the words digital film on the package. Linco designed its packaging to be bright gold, closely matching Kodak analog film. Analog film has a speed rating, so Linco created one for its digital film. (Tripsas, 2009, p. 449)
Tripsas argues that the firm’s choice of identity had a strong positive effect on consumer adoption and drove its dominance of the new product category.
Linco was an entrepreneurial venture with no pre-existing products or organizational identity, and therefore had substantial latitude to shape the analogical learning process of external audiences. Similarly, no organizational identity was cued in the advertisements devised by Moreau, Lehmann, et al. (2001) and Moreau, Markman et al. (2001). By contrast, established firms launching products during the market emergence stage are often constrained by their established identities. In particular, we argue that the de alio firm’s categorical identity determines the base domain that is activated in the analogical learning process. Moreover, the categorical association invoked by the firm’s established identity may be difficult to override through marketing efforts or other intentional claims by the firm. For example, Moreau, Lehmann, et al. (2001) and Moreau, Markman, et al. (2001, 2001, p. 496) show that when two competing labels are shown in sequence for the same ambiguous product, the first label “has a disproportionate influence on consumers’ inferences about and preferences for a new product.”
As a consequence, identical products introduced by firms with different categorical identities may differ in how they are comprehended and valued by consumers, because they activate different base domains. The categorical identity of de alio firms may therefore be an advantage or a liability in the competitive dynamics of emerging markets depending on the extent to which consumers have difficulty interpreting the new product in terms of the activated base domain.
MAPPING AND INFERENCE
When an unfamiliar object is linked to a familiar mental representation through analogy, the structure of the base domain leads individuals to emphasize comparable features, attempt to impose the relational structure from the base domain, and try to draw correct (or useful) inferences based as a result (Gentner & Smith, 2012; Gregan-Paxton, Hibbard, Brunel, & Azar, 2002). If consumers have difficulty constructing a mapping, they will struggle to make sense of the product, or draw incorrect inferences about the product’s functionality. For example, someone who interprets a smartphone through the lens of a laptop computer may be disappointed by how difficult it is to compose emails on a smartphone, or may find the inability to create spreadsheets problematic.
Fluency in mapping is hindered when the target and the base domain differ substantially in attributes and relations between attributes. The relational structure is critical, while common attributes may activate a base domain, attributes alone are unlikely to support a successful mapping to the target domain. For example, many objects have a button (an attribute); however, these buttons have different consequences when pressed (i.e., relations – ringing a doorbell, taking a photo, turning off an appliance). As a result, mapping based on attributes alone is difficult. Instead, “people will align two domains based on their common relations.… [W]e prefer to match large, deep-connected systems” (Gentner & Smith, 2012, pp. 131–132).
A common relational structure provides better support for inferences about the target domain. When faced with gaps in their knowledge of the target, people borrow known relations from the base domain to fill in missing elements of the target domain, provided the candidate relations are not in conflict with relations in the target. As a result, relational mappings often have greater explanatory power than attribute mappings. Knowing that the target and the base both have buttons tells us less than being confident in the inference that pressing the button will result in a snapshot. Gregan-Paxton and John (1997, p. 271) argue that “comparisons based on relational mappings produce more goal-relevant inferences than comparisons based on attribute mappings” and are preferred for that reason.
In addition to a common relational structure, fluency in mapping between domains is enhanced to the extent that corresponding elements in the relational structure are similar. Gentner and Smith (2012) term this transparency. Mapping is easier to the extent that objects are similar when they play the same roles in the shared relational structure, and dissimilar when they play different roles. Again, shared attributes are not enough; the attributes that are shared should have similar relations.
This model of fluency in mapping and inference provides a means of conceptualizing how continuous an innovation is relative to existing knowledge (Moreau, Lehmann, et al., 2001); continuity is a function of transparency and, in particular, a common relational structure. Importantly, continuity may depend on the base domain that is activated. An innovation may be fluently mapped from base domain A because there is a high degree of transparency and overlap in relational structure, while mapping may be more difficult from domain B. In this case, the new product may be more easily seen as an incremental change relative to base domain A but a discontinuous or radical change relative to domain B. For example, a computer-controlled car is more likely to be seen as a new kind of automobile than as a new kind of computer; our conceptualization of an automobile is more useful for drawing goal-related inferences than is our concept of a computer.
The literature on analogical learning suggests, in short, that the comprehension and valuation of a recombinant innovation depends on the continuity between the base domain and the target. As our discussion of the access phase demonstrated, this in turn means that the categorical identities of de alio producers play a critical role in the commercialization process during market emergence because different categorical identities activate different base domains. De alio producers whose categorical identities facilitate fluent mapping from the implied base domain will be advantaged in an emerging market.
EMPIRICAL SETTING AND DATA
We focus on the response of organizational audiences to new products arising from a recombinant innovation that combines technical elements and product features from existing products and technologies. Specifically, we focus on digital cameras. The digital camera industry has a number of appealing features as a strategic research site:
- First, digital cameras rapidly became an established product category and gained widespread acceptance among consumers. The fact that this occurred primarily in the past fifteen years provides an unusual opportunity to gather rich data tracking the evolution of consumer responses to products in an emerging market category.
- Second, research during the early history of the industry indicates that there was real interpretive ambiguity about the product among consumers. For example, a study conducted during the early history of the field concluded that “a digital camera does not fit unequivocally into any existing product class, and thus, the digital camera is not automatically categorized as a camera” (Moreau, Lehmann, et al., 2001; Munir, 2005).
- Third, entry into the digital camera space consisted almost entirely of established firms from three established industries – (film) photography, computing and consumer electronics – with little if any entry from de novo producers (Benner & Tripsas, 2012). This allows us to focus on the role of pre-established producer identities in consumer evaluation.
- Fourth, as Benner and Tripsas (2012) show, the existence of a well-developed supply chain for the core technologies in a digital camera minimized the importance of pre-existing differences in producer capabilities; in particular, while there may have been firm-level differences in competence, producers from one industry were not at any particular capability disadvantage relative to firms from another industry.
We analyze user ratings of a large population of consumer digital camera models. Our main analytic interest is in how the ratings of products from firms with different categorial identities differ. A key aspect of our research design is that we have highly detailed data on the technical characteristics of each camera. This is important because at any given point in time, cameras from different producers may differ technically, as firms compete to introduce new features and catch up to features offered by their competitors. Benner and Tripsas (2012) show that prior industry affiliation affects the innovative focus of digital camera producers. Yet our interest is only in the effects of organizational identity. In the absence of measures of each camera’s technical characteristics, observed differences in ratings might be due to differences in specifications and capabilities. Controlling for detailed technical characteristics allows us to more confidently draw inferences about the effects of categorical identity on consumer evaluations.
The producers in our sample are listed in Table 1, along with our coding of each producer’s organizational identity. (We discuss the construction of the sample below.) Producers listed as “Film” are firms that produced film-based cameras priory to their entry into the production of digital cameras, while the remaining producers originate in the computer (e.g., HP) and consumer electronics industries (e.g., Sony). As is apparent from the table, 11 of the 21 producers observed in our dataset had prior experience in the film camera industry. Similarly, film producers account for 52% of the reviews in our sample. However, they account for 60% of the products.
Table 1. Digital Camera Producers.
We collected detailed product-level technical information, as well as consumer product evaluations, on comprehensive sample digital cameras produced between 1996 and 2010. Although the history of digitalized image capturing and processing can be dated back in the 1960s for the aerospace industry, it was not until the mid-1990s that digital camera producers began to offer quality image at affordable price to the consumer market. The key technology that underpinned the digital imaging performance to commercial standard is the charge-coupled device (CCD), which detects and records the light intensity into electronic information when integrated with a photoelectric sensor. The widespread consumer adoption of personal computers during this period, along with the growth of the Internet, spurred the growth of the product category, relegating traditional film cameras to an increasingly peripheral position in the consumer market.
We rely on three major consumer websites devoted to digital photography for product-level information:
Collectively, these data sources contain detailed technical specifications for almost all consumer digital cameras products in the United States since 1996. In that year, sales of digital cameras in the United States totaled 400,000 units, compared to 15.1 million film cameras. By 2004, the corresponding figures were 18.2 million for digital cameras and 6.7 million for film cameras. In 2010, 34 million digital cameras were sold in the United States Thus our data cover digital cameras as they shifted from being a niche product to reaching a mass audience.
Our sample is limited by the fact that we only include in our analysis cameras that were reviewed on the Digital Photography Review website. The process that determines whether a camera is reviewed by users on the website is opaque to us, but it seems reasonable to surmise that the likelihood of a camera appearing in our dataset is a function of the sales and popularity of the manufacturer. Consistent with this, our list of producers in Table 1 is shorter than the list in Benner and Tripsas’ (2012) more comprehensive study. Benner and Tripsas document 83 firms operating in the digital photography market between 1991 and 2006. By contrast, our dataset contains 21 firms in the period from 1996 to 2010, with no new producers entering our sample after 2006. These firms released 924 products (including unreviewed cameras) in the decade before 2006, while Benner and Tripsas collected data on 1,629 products between 1991 and 2006.
A comparison of the producers listed in Table 1 with those listed by Benner and Tripsas (2012, Table 1) suggests that our sample excludes a number of firms that entered very early in the history of the industry, and niche and low-end producers. Our sample is more likely to include film companies: where our dataset includes 13 of the 25 film companies listed by Benner and Tripsas, coverage of non-film companies is worse, with 8 of 43 companies included. However, our sample of producers covers a large portion of the market by unit volume. For example, in 1999 the top six digital camera producers – all of them in our sample – accounted for 78.5% of unit volume in the United States. By 2010, the top six producers accounted for 76.1% of worldwide market share. In short, our sample covers the vast majority of the U.S. digital camera market.
Figure 1 presents the number of product releases per year in our dataset, by year of release, for those cameras reviewed on DPR. A very small number of digital cameras in our sample were introduced in the first two years of our data – 4 in 1996 and 10 in 1997. Of these 14 products, 10 were introduced by film camera companies (Agfa, Canon, FujiFilm, Nikon, and Olympus), while Casio and Epson introduced two digital cameras each. Product releases doubled in each of the following two years and accelerated consistently until an initial peak of 134 models in 2004. A gradual decline began in 2007, perhaps due to the growing popularity of smartphones and cellphone cameras.
Fig. 1. Product Releases per Year.
Figure 2 presents the number of reviews posted on the Digital Photography Review website by year of the review. These data, which form the basis for our dependent variable, are truncated since we do not have any reviews prior to the launch of the DPR website was in 1999. We therefore are not able to gauge the opinions of the consumers who were the earliest adopters in 1996–1998.3
Fig. 2. Number of Reviews per Year.
We see in Fig. 2 a rapid take-off in the number of reviews between 2001 and 2002. While this tracks the increased number of digital cameras available (by the end of 2001, 221 cameras had been introduced), the pattern is difficult to interpret since at least some of this increase may be due to increases in the popularity of the DPR website. However, it does not appear to be the case that the increase in the number of reviews on DPR came at the expense of a rival reviewing site. It thus appears that 2002–2005 were the years of peak consumer critical engagement in the sensemaking process around digital cameras. Following 2005, the number of reviews drops off rapidly, suggesting that the conceptualization process has stabilized and the new category has become more institutionalized.4
MEASUREMENT
Dependent Variables
Our dependent variable of interest is consumer ratings of each product. We collected reviews from the Digital Photography Review website, which has gathered voluntary consumer reviews since 1999.5 For each review, consumers provide numeric ratings (from 1 to 5) of the product along five dimensions: construction, image quality, features, ease of use, and value for money. As an overall score we follow DPR and use the arithmetic mean of the five scores as our main dependent variable. Users of the website (e.g., people in the market for a camera) can browse the site and see the overall rating across all reviews for a particular product, as well as each individual review. The website also records the date on which each review is posted.
To guard against the possibility that our results are driven by one particular dimension of assessment, we also estimate models using the ratings along each separate dimension as dependent variables.
For ease of presentation we multiply the dependent variable by 100.
Independent Variables
Our primary goal in selecting independent variables for our models is to control, to the extent possible, for objective differences in the technical features and relative quality of each camera. Our first step in doing so is to include in our models measures characterizing each camera’s technical features. Combining the information from the three websites, we are able to gather information on 34 technical specifications (or features) for each product.6 These specifications include aperture range, shutter speed, resolution, pixel density, redeye deduction, internal flash, among others (see the appendix for a full listing). When we encountered missing data, we coded additional product specifications through the official websites of all existing digital camera companies.
Controlling for these features is also important because the likelihood of introducing a product feature is a function of the industry background of the de alio producer. Benner and Tripsas (2012) show that producers’ conceptualization of the nature of a digital camera (e.g., as a substitute for film cameras or as a PC peripheral) affected the rate at which they introduced different product features. However, the imitation of successful features by other producers was rapid.
Accounting for differences in technical quality is complicated by the fact that users may provide reviews at any time after the product has been released. However, the release of other products with new features and improved performance will change the relative quality of the existing camera. To account for this, we include three measures. First, we include the release date of the camera, measured in weeks since the first release date observed in the sample (May 13, 1996). Second, we construct a measure of technical pressure to capture the focal product’s distance from the technical frontier as defined by other products on the market. To construct this measure, we computed the percentile rank of the focal camera relative to all other models along the 34 technical dimensions. Higher rank means lower technical pressure. The overall technical pressure is the average of percentile rank across all 34 technical specifications relative to all other digital camera models launched in the market for the past 18 months. Finally, we include the date of the review, calculated as weeks since the release date.
An individual’s assessment of a product may be affected by the ratings of previous reviewers. For a review recorded at time t, we therefore include the average score, across all previous reviews for the same camera, along each of the five underlying dimensions. We also included a count of the number of reviews of the focal camera prior to the current review.
Beyond the user-generated reviews, the Digital Photography Review (DPR) also provides professional reviews and ratings for a portion of camera models. Between 1999 and 2010, the focal study period, DPR changed its rating schema several times (such as Recommended vs. Not Recommended; Gold Award, Silver Award, No Award; and most recently numeric scores). The only consistent measure about DPR’s opinion is whether it recommended the camera or not. We thus coded DPR recommended camera models as 1, and 0 otherwise.
Finally, we also include a count of the number of words in the focal review. Descriptive statistics are provided in Table 2.
Table 2. Summary Statistics.
Variable |
Mean |
Std. Dev. |
Overall score |
427.52 |
69.34 |
Construction |
430.71 |
81.16 |
Features |
431.06 |
76.38 |
Image quality |
420.52 |
92.56 |
Ease of use |
431.75 |
78.17 |
Value for money |
423.56 |
97.38 |
N prior reviews (00) |
0.39 |
0.46 |
Technical pressure |
0.74 |
0.08 |
Recommended by DPR |
0.48 |
0.50 |
Word count (00) |
1.47 |
1.65 |
Release week |
407.32 |
126.11 |
Weeks since release |
49.08 |
53.52 |
Film company |
0.73 |
0.45 |
Keywords |
1.58 |
3.00 |
Note: N = 26,996.
RESULTS
Table 3 presents ordinary least squares regression estimates of the effects of a producers de alio identity on the overall rating assigned to a camera by a reviewer. (As noted above, we have multiplied the dependent variable by 100.) In the first model, we see that cameras that have been rated more highly by previous reviewers receive higher scores, which may be due to this variable picking up unmeasured quality differences, or a social contagion process. As we would expect given the pace of innovation and new product introductions in the early years of consumer acceptance of digital cameras, ratings decline as the number of weeks since release increases, and as the product falls further behind the technical frontier with the introduction of competing products.
Table 3. The Effects of de alio Identity on User Ratings of Digital Cameras.
In the first model in Table 3, cameras released by producers from the film photography industry do not appear to be more highly rated than cameras produced by other de alio producers. However, this is because the ratings difference between film and non-film producers dissipates over time as the market moves out of its emergence phase. This is shown in the second model, where we interact the film company dummy variable with the release date. The negative interaction effect indicates that this effect declines over time. In the early history of the digital camera market, products released by film companies had a perceptual advantage over otherwise technically equivalent products released by non-camera companies. This film company advantage is consistent with their identities as traditional camera producers allowing consumers to more easily make sense of the new technology, with more positive valuations as a result. Producers from other industries likely activate different base domains, resulting in lower interpretive fluency and hence valuations.
One concern with these results is that the producers have pre-existing brands and capabilities that might influence consumer ratings and be correlated with our measure of de alio identity. In the third column of Table 3 we address this issue by including producer fixed effects in the model. While this means that we cannot estimate a main effect of being a film camera producer, we can examine the unequal time-varying effect through the interaction between the film camera dummy variable and release week. This coefficient remains negative and significant as before.
These results are in line with our theoretical arguments, which stressed that the categorical identities of de alio producers play a critical role in shaping consumer evaluation, especially in the emerging period of a recombinant innovation. One source of concern, however, is the potential relationship between ratings, market exit and prior industry. Firms with poor consumer ratings should be more likely to exit the digital camera market. This seems particularly likely to be true of non-film producers. For producers from the film photography industry, the growth of digital cameras goes hand in hand with a steep decline in the market for film cameras. As a result, failure in digital cameras leaves them with few options, which should increase their exit threshold. Producers of consumer electronics, or computer companies, by contrast, face less pressure to remain in the market. In short, low-rated non-film producers are more likely to exit than low-rated film producers. Such differential exit patterns could explain the decline in the ratings advantage of film companies. Comparisons of the Kaplan-Meier survival curves of film and non-film companies (not shown here) confirm that non-film companies did exit the market at a substantially higher rate than film companies.
We address this possibility in the fourth model in Table 3, where we restrict the sample to “survivors” — that is, producers present in the data for the entire observation period. Detecting exit from the market is difficult. Our approach is to code a firm as exiting the market if it had not released a product appearing on DPR for an extended period of time.7 The estimates in the final column of Table 3, which re-estimate the third model on the restricted sample, suggest that the differential survival cannot explain our results. (When we re-estimate the second model, that is, without fixed effects, on the sample of survivors, we find similar results.)
Our analyses in Table 3 use the overall ratings as the dependent variable. Table 4 allows us to examine whether the differential ratings of firms from different industries are driven by a particular dimension of evaluation. This does not appear to be the case. We see the basic pattern in Table 3 for four of the five dimensions. That is, initially consumers consider cameras by film companies as having better construction, more attractive features, superior image quality, and higher value for money. Yet the advantages along these four evaluation dimensions dissipate steadily over time. In the one exception, for ease of use the pattern is sustained but the coefficients are not statistically significant. Overall, Table 4 provides additional support for our main findings in Table 3.
Table 4. The Effects of de alio Identity on Ratings of Specific Dimensions of Camera Quality.
In Table 5 we address the possibility that consumers who are more knowledgeable about photography may be less influenced by the de alio identity of producers in the new market. We believe that this subgroup of more sophisticated consumers are not only more capable of developing their mapping and inference criteria based on technical merits, but also more capable of comprehending the target domain by combining properties from multiple base domain. In other words, the de alio identity would play a less important role for product evaluation by these consumers. In order to measure expertise, we coded the textual portion of the reviews provided along with each rating. Specifically, we tallied the appearance of a set of keywords that indicate specialized knowledge of photography. We assume that higher scores indicate that the reviewer is more knowledgeable about camera technology.8
Table 5. The Effects of de alio Identity on User Ratings, Net of User Expertise.
As is apparent in the first column of Table 5, consumers who use more keywords in their reviews tend to be more critical and give the product lower ratings. Moreover, the result in the second column indicates, as expected, that these consumers are less influenced by the de alio identity of the producer. This result holds when controlling for firm fixed effects, as exhibited by Model 3 in Table 5. It is worth noting that this moderating effect of expertise suggests a role of cognition in the valuation process consistent with our analogical learning model; this moderation would not be expected if the differences in ratings were due to “true” quality differences (i.e., rooted in capabilities).
In order to better understand the temporal pattern in the effects of de alio identity, in Table 6 we spilt the sample into two periods with roughly equal length: pre-2004 and post-2004 period, based on the date when a user review was published. Household penetration for digital cameras exceeded 50% for the first time in 2004, and the market increasingly shifted from first-time buyers to repeat buyers (Digital Photography Review, 2005). New product introductions also slowed, as seen in Fig. 1, and 2004 marked the first year that the number of consumer reviews dropped on DPR.9
Table 6. The Effects of de alio Identity on User Ratings, by Phases of Market Emergence.
Table 6 compares the difference of identity effect for the full sample, the survivor subsample, and for the full sample with knowledgeable consumer effect included. While two coefficients for Film Company are not statistically significant, the basic pattern conforms to our existing findings. Film companies are consistently better rated before 2004; however, they receive much less advantageous ratings after 2004. The knowledgeable consumer effect and its interaction with Film Company are also in line with our predictions.
CONCLUSION
In this article, we have advanced a theoretical framework for understanding the role of producer identities in the emergence of new product markets. The distinctive element of this theory has been to emphasize the role analogical learning processes play in the sensemaking of consumers (and other external audiences) when confronted with a novel product. Specifically, producer identities matter because different categorical identities activate different base domains, and are therefore more or less likely to facilitate fluent mapping from the implied base domain to the new product. The strategic implication of this process is that in emergent markets, producers whose identities (or actions) facilitate greater interpretive fluency will be advantaged. Our empirical examination of the role of producer categorical identities in the digital camera industry demonstrates that producers from the film camera industry were systematically advantaged in consumer ratings, net of the technical characteristics of their offerings.
It is worth noting that in the case of digital cameras, the role of categorical identities changes as the market matures. In our view, this highlights a key feature of the emergence phase of a new market, namely the absence of a well-defined interpretive framework for the new product. Once a more settled meaning of the category “digital camera” emerges, for example, consumers need not rely on analogical sensemaking to assess a product, in which case the activation of the base domain by the producer’s identity plays less of a role.
While digital cameras have transitioned to the status of a mature (and indeed, with the advent of smartphone cameras, rapidly declining) market, similarly emergence processes play out today thanks to ongoing recombinant activation. At the risk of rapidly dating this article, contemporaneous examples include such market categories as “wearables” (with both de novo entrants and de alio entrants from the fashion, watch, sports, and computer industries) and “the Internet of Things” (with start-ups as well as major de alio firms such as Intel and Google). Both of these settings are particularly interesting because they, unlike the digital camera industry, have a lot of de novo start-ups. In our view, a key challenge for start-ups in this space, given the analogical learning processes we have described, is to establish a clear identity from the outset. Both market labels (“wearables,” “IoT”) are sufficiently diffuse at this point that start-ups that adopt these labels may be at a disadvantage relative to de alio firms that happen to match well with the use cases that consumers discover for the new technologies. A critical issue for de novo entrants, in this sense, is likely whether such market matches exist for de alio firms.
Our discussion leaves unexamined the question of how the sensemaking process works out for de novo entrants to an emerging market, largely because there were almost no de novo entrants in the digital camera industry. Nonetheless, the analogical learning framework suggests some implications for de novo firms. Such entrants are, of course, not constrained by any pre-existing associations in the minds of consumers; in this sense, the firm has full “control” over the base domain that could be activated and the resulting fluency of interpretation. The start-ups marketing decisions and claims are thus likely to play a critical role. At the same time, start-ups in emerging markets may find the lack of constraint problematic in certain respects. In particular, the emergent nature of the market may make it unclear to the firm which base domain their marketing efforts should try to activate (and perhaps how). Much as start-ups experiment with different business models, they may respond to this by experimenting with different identity claims, creating confusion in the minds of consumers.
Finally, we believe that an important feature of these models of fluency in mapping and inference is that they allow for a conceptualization of the continuity or “distance” between an existing product category and a nascent category (Moreau, Lehmann, et al., 2001). As a result, the models serve as a foundation for predictive claims regarding which existing industries should make their incumbents relatively advantaged or disadvantaged – from an identity perspective – in an emerging product market. As an empirical matter, formulating such predictions requires a deep understanding of the cognitive schema associated with the different industries in question, as well as the features of the new product, and hence was beyond the scope of the current article. Nonetheless, we hope that future work can identify tractable techniques for assessing the distance between the established identities of market entrants and the features of the novel product.
NOTES
ACKNOWLEDGMENTS
We are grateful for comments and suggestions from seminar participants and colleagues at the Stanford Graduate School of Business, the National University of Singapore, and Peking University. Feedback from Henrich Greve and Marc-David Seidel is much appreciated. All remaining errors are our own. This research was supported by the Stanford Graduate School of Business and National Natural Science Foundation of China under Grant No. 71272032.
REFERENCES
Benner, M. J., & Tripsas, M. (2012). The influence of prior industry affiliation on framing in nascent industries: The evolution of digital cameras. Strategic Management Journal, 33(3), 277–302.
Carroll, G. R., Bigelow, L. S., Seidel, M.-D. L., & Tsai, L. B. (1996). The fates of de novo and de alio producers in the American automobile industry 1885–1981. Strategic Management Journal, 17(S1), 117–137.
Carroll, G. R., & Hannan, M. T. (2000). The demography of corporations and industries. Princeton, NJ: Princeton University Press.
Clark, K. B. (1985). The interaction of design hierarchies and market concepts in technological evolution. Research Policy, 14(5), 235–251.
Digital Photography Review. (2005). Digital camera sales hit all-time high. Retrieved from http://www.dpreview.com/news/2005/07/18/infotrend_dcsales
Fleming, L., & Sorenson, O. (2001). Technology as a complex adaptive system: Evidence from patent data. Research Policy, 30(7), 1019–1039.
Gentner, D. (1989). The mechanisms of analogical learning. In S. Vosniadou & A. Ortnoy (Eds.), Similarity and analogical reasoning (pp. 199–241). New York, NY: Cambridge University Press.
Gentner, D., & Smith, L. (2012). Analogical reasoning. In V. S. Ramachandran (Ed.), Encyclopedia of human behavior (2nd ed., pp. 130–136). Oxford: Elsevier.
Gregan-Paxton, J., Hibbard, J. D., Brunel, F. F., & Azar, P. (2002). “So that’s what that is”: Examining the impact of analogy on consumers’ knowledge development for really new products. Psychology & Marketing, 19(6), 533–550.
Gregan-Paxton, J., & John, D. R. (1997). Consumer learning by analogy: A model of internal knowledge transfer. Journal of Consumer Research, 24(3), 266–284.
Hannan, M. T., Pólos, L., & Carroll, G. R. (2007). Logics of organization theory: Audiences, codes, and ecologies. Princeton, NJ: Princeton University Press.
Hargadon, A. B., & Douglas, Y. (2001). When innovations meet institutions: Edison and the design of the electric light. Administrative Science Quarterly, 46(3), 476–501.
Helfat, C. E., & Lieberman, M. B. (2002). The birth of capabilities: Market entry and the importance of pre-history. Industrial and Corporate Change, 11(4), 725–760.
Henderson, R. M., & Clark, K. B. (1990). Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms. Administrative Science Quarterly, 35(1), 9–30.
Khessina, O. M., & Carroll, G. R. (2008). Product demography of de novo and de alio firms in the optical disk drive industry, 1983–1999. Organization Science, 19(1), 25–38.
Klepper, S., & Simons, K. L. (2000). Dominance by birthright: Entry of prior radio producers and competitive ramifications in the US television receiver industry. Strategic Management Journal, 21(10–11), 997–1016.
Moreau, C. P., Lehmann, D. R., & Markman, A. B. (2001). Entrenched knowledge structures and consumer response to new products. Journal of Marketing Research, 38(1), 14–29.
Moreau, C. P., Markman, A. B., & Lehmann, D. R. (2001). What is it? Categorization flexibility and consumers’ responses to really new products. Journal of Consumer Research, 27(4), 489–498.
Munir, K. A. (2005). The social construction of events: A study of institutional change in the photographic field. Organization Studies, 26(1), 93–112.
Negro, G., Koçak, Ö., & Hsu, G. (2010). Research on categories in the sociology of organizations. In G. Hsu, G. Negro, & Ö. Koçak (Eds.), Categories in markets: Origins and evolution (Vol. 31, pp. 3–35). Research in the Sociology of Organizations. Bingley, UK: Emerald Group Publishing Limited.
Nisbett, R., & Ross, L. (1980). Human inference: Strategies and shortcomings of human judgment. Englewood Cliffs, NJ: Prentice Hall.
Podolny, J. M. (1993). A status-based model of market competition. American Journal of Sociology, 98(4), 829–872.
Rindova, V. P., & Petkova, A. P. (2007). When is a new thing a good thing? Technological change, product form design, and perceptions of value for product innovations. Organization Science, 18(2), 217–232.
Rosa, J. A., Porac, J. F., Runser-Spanjol, J., & Saxon, M. S. (1999). Sociocognitive dynamics in a product market. The Journal of Marketing, 63(Special Issue), 64–77.
Schumpeter, J. A. (1950). Capitalism, socialism and democracy. New York, NY: Harper.
Sujan, M. (1985). Consumer knowledge: Effects on evaluation strategies mediating consumer judgments. Journal of Consumer Research, 12(1), 31–46.
Tripsas, M. (2009). Technology, identity, and inertia through the lens of the digital photography company. Organization Science, 20(2), 441–460.
Tripsas, M., & Gavetti, G. (2000). Capabilities, cognitions, and inertia: Evidence from digital imaging. Strategic Management Journal, 21(10–11), 1147–1161.
Tushman, M. L., & Anderson, P. (1986). Technological discontinuities and organizational environments. Administrative Science Quarterly, 31(3), 439–465.
Yamauchi, T., & Markman, A. B. (2000). Inference using categories. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(3), 776.
Zuckerman, E. W. (1999). The categorical imperative: Securities analysts and the illegitimacy discount. American Journal of Sociology, 104(5), 1398–1438.
APPENDIX
Table A1. Observed Technical Features of Digital Cameras.
