The Design Evolution of Digital Cameras and Its Implications

Zhang, Berger Academy of Management Conference, 2008, Anaheim, California

THE INFLUENCE OF TECHNOLOGY EVOLUTION ON TECHNOLOGY ADOPTION: A STUDY OF DIGITAL CAMERAS

Min Zhang

Boston University [email protected]

Paul D. Berger Bentley College and Boston University

[email protected]

Zhang, Berger The Influence of Technology Evolution on Technology Adoption

Abstract

In this study, we examine technology adoption by integrating it with technological evolution. We trace both the technology evolution S-curve, which represents changes in performance of key components over time, and the emergence of dominant designs, which represents changes in the architecture over time. We highlight the importance of differentiating three aspects of dominant designs: component, internal and external. We suggest that the emergence of external dominant designs may have an impact on innovation diffusion, particularly in markets with network effects and for technologies that depend on a larger system. Our study on digital cameras in the U.S. market indicates that the emergence of both internal and external dominant designs have a positive and significant impact on new camera sales. The technology evolution S-curve and internal, and external dominant designs are all captured by quarterly key performance data of digital cameras shipped from 1996 to 2005. The innovation diffusion S-curve is tracked by actual quarterly digital camera sales data from this same 1996 to 2005 time period.

Keywords: innovation diffusion, technology evolution, dominant designs, marketing, sales, digital cameras

2


INTRODUCTION

Since the early days of diffusion inquiries in agriculture (Ryan and Gross, 1943,

Griliches, 1957), research on innovation diffusion has grown tremendously (Rogers, 1976,

Rogers, 2003). Researchers in marketing (e.g., Bass, 1980, Norton and Bass, 1987, Mahajan and

Wind, 1988, Mahajan, Muller and Bass, 1990), and technology strategy (e.g., Gort and Klepper,

1982, Lilien and Yoon, 1990, Agarwal and Bayus, 2002) are especially interested in the diffusion

and time path of new consumer products, because adoption of most innovations involves the

purchase of new products. During this process, the market for the technology develops and

expands, firms that produce the technology get their products sold, and the technology becomes

adopted by consumers. Thus, diffusion is critical to the performance of firms.

A review of these diffusion models indicates that these models focus on the information

about the innovation, characteristics of the potential adopters, the relationship among the

adopters, and the initial choice of early adopters (Geroski, 2000). The characteristics of the

innovation itself have drawn little attention. Rogers pointed out that it was important to view

technology dynamically, but little work had been done in this area (Rogers, 2003). A few

previous studies shed some light on the impact of technological changes on diffusion, such as the

diffusion of hybrid corns (Griliches, 1957), the effect of dominant designs on sales (Anderson

and Tushman, 1990), the adoption and substitution of successive generations of high-technology

products (Norton and Bass, 1987), and the effects of firm entry on market evolution and sales

takeoff (Agarwal and Bayus, 2002). We build on this stream of research and advance our

understanding of the effects of technological changes on diffusion by studying the evolution of

digital cameras in the U.S. market. We hypothesize that both technology improvements and the

emergence of dominant designs have positive and significant impacts on diffusion. We suggest

3


that the emergence of an external dominant design may have an even greater influence on

diffusion in the presence of network effects and for products that belong to a larger system. Our

hypotheses are supported by the results of an empirical study of digital cameras. To represent the

technology evolution S-curve, we used key performance data of over 1000 types of digital

cameras shipped in the U.S. from 1996 to 2005. To represent market evolution, we used sales

data of digital cameras in the U.S. for the same time period.

The structure of the paper is as follows. In the next section, we explain the mechanisms

through which improvements in technology performance and the emergence of dominant designs

affect market evolution. Then we analyze our data on digital cameras to test our hypotheses.

Finally, we conclude with discussion and implications associated with our study.

INNOVATION DIFFUSION, TECHNOLOGY PERFORMANCE EVOLUTION AND THE EMERGENCE OF DOMINANT DESIGNS

“Diffusion is the process in which an innovation is communicated through certain

channels over time among members of a social system” (Rogers, 1995, p5). “Innovation is an

idea, practice or object that is perceived as new by an individual or another unit of adoption.”

(Rogers, 1995, p12). Innovation and technology are used as synonyms in diffusion research. We

focus on the adoption of consumer products, and measure diffusion by new product sales. The

core artifact in the diffusion process is innovation itself, and the attributes of the innovation are

often an ever-changing one. In this paper, we provide new insights about how technology

evolution affects the sales of new consumer products.

There are two prominent ways to view technology evolution: one is tracing the progress

of key performance indicators, which usually follows an S-curve (Foster, 1986); the other is

4


tracing the emergence of dominant designs (Dosi, 1982, Abernathy and Utterback, 1978). We

suggest that both are positively related to the diffusion of new products.

Figure 1 and Figure 2 illustrate the technology performance S-curve using resolution per

2004 dollar cost as the key performance indicator, and the cumulative sales S-curve for digital

cameras shipped in the U.S., from the first quarter of 1996 to the first quarter of 2005. However,

since digital cameras are still in the development stage, the curves have not reached the inflection

points yet (i.e., the “S” shape is not yet visible).

---------------------------------------- Insert Figure 1 about here

------------------------------------- -------------------------------------

Insert Figure 2 about here --------------------------------------

We choose digital cameras to illustrate our theory because digital cameras have

undergone significant technological change over time, and information on digital cameras is

abundant. All descriptions about digital cameras include the connection between the cameras and

the larger digital-imaging system. We also note that network effects exist in the digital camera

market.

Digital cameras capture images on electronic sensors and store them in a digital format. It

is a disruptive technology change from the traditional film-based camera technologies. Digital

cameras were initially targeted toward the niche market of professional users in the print and

press industries because of the high expense, and these users’ need for instant transmission of

images. Later, after prices dropped, individual amateur consumers became the majority adopters

of this technology.

5


The basic concepts for a digital camera emerged in 1963. When it was first

commercialized in the 1980s, its resolution level was significantly lower than that of the

traditional camera. From that time on, the digital camera product category, as noted earlier, has

undergone significant technological change and its widespread diffusion began to blossom in the

late 1990s. In 2005, the last year for which we have available data, both the digital camera

technology and its diffusion were still progressing at a fast pace. Therefore, the digital camera’s

development process provides an excellent background to research the relationship between

technology evolution and market evolution.

Digital cameras have some similarities to film-based cameras: in both types of cameras, a

lens focuses an image onto a recording medium. Once the recording medium receives the image,

the raw image information is processed to produce a permanent image. The major differences

between a digital camera and a traditional camera are the recording medium and the storage

medium. Conventional cameras use plastic film coated with light sensitive chemical emulsion,

while digital cameras use a light-sensitive electronic array—the sensors. With conventional

cameras, the recording medium, the film, becomes the permanent home of the original exposure.

With digital cameras, electronic signals generated when photons strike the recording chip are

moved away from the CCD (Charge Coupled Device) and processed electronically into a digital

file. This file is then stored elsewhere, on a memory chip in the camera itself, or on a removable

recording medium. Film-based cameras interact with the bigger imaging system through the film.

Digital cameras can interact with the external imaging system through many more options:

removable memory cards, cables, docking stations, etc.

6


Technology Improvement Positively Impacts Innovation Diffusion

The technology S-curve describes the path of technology improvements in performance

when more effort is put into developing the technology (Foster, 1986). It suggests that the

magnitude of a product’s performance improvements in a given time period due to a given

amount of engineering effort are likely to change along an S-curve as the technology matures. In

the early stages of development, the rate of progress in performance will be relatively low;

gradually it will accelerate, and as the technology matures, it will slow down again (Christensen,

2003). The technology S-curve can be used to guide firms in allocating their time and effort in

research and development at different stages of technology maturity, and switching to new

architectural technology. The S-curve is often measured by key a performance indicator, which

reflects the attribute of a key component (Christensen, 1992). In the case of computers, the key

component is the processor, and its key attribute is its speed. From the design point of view, the

technology S-curve reflects the progress in the key component. Of course, the key performance

indicator can change as the focus of research and development on the technology changes from

one dimension to another (Christensen, 1997). We contend that as technology improves along its

S-curve, it will influence its diffusion in the market place in a positive way, by affecting

information about the product - both the price and quality of the product, and the user’s

perception of the product.

Technology performance improvements affect information about a new product. The

epidemic model of diffusion points out that the information about the technology is the driver of

innovation diffusion (Geroski, 2000). When technology improves, information about the new

product becomes more appealing. Consumers become more likely to spread the news about the

product, which, in turn, leads to faster diffusion of the product.

7


Technology performance improvements can also affect perceived attributes of a new

product. Perceived attributes of innovations are considered to be one of the most important

factors influencing adoption decisions (Rogers, 2003). Relative advantage, which is the ratio of

the expected benefits to the costs of innovation, is one of the most important attributes of an

innovation. Technology performance improvements favorably influence relative advantage,

because they entail better performance at the same cost, or both better performance and lower

cost.

The unified technology acceptance model also endorses this idea. It is a static model

which measures users’ intentions to use a technology and/or actual usage of a technology, based

on perceptions of the attributes of the technology. This model argues that four major factors can

be used to predict an individual’s acceptance of a technology (Venkatesh, Morris, Davis and

Davis, 2003). These four factors are: performance expectancy, effort expectancy, social

influence and facilitating conditions. Performance expectancy of a technology is defined as “the

degree to which an individual believes that using a technology will help him or her to attain

gains in job performance.” Effort expectance is defined as “the degree of ease associated with the

use of a system” (Venkatesh, et al., 2003). As a technology progresses along the S-curve, it tends

to better meet the performance expectancy and effort expectancy, thus making it more probable

that the technology will be accepted.

Technology performance improvements can affect both the actual cost and the quality of

a new product. Scholars in the field of marketing advocate that new firm entries affect both

supply-side and demand-side factors by increasing quality and decreasing price, which,

accordingly, will affect the market evolution of new products (Agarwal and Bayus, 2002). When

a new product enters the market, sales are low because the product is still primitive. As new

8


firms enter the market, they not only drive prices down, but also improve distribution channels,

product features, and branding, in order to differentiate from early entrants (Agarwal and Bayus,

2002). Their 2002 study approximates performance improvements and price decreases by tracing

firm entries; we, instead, directly analyze the key performance indicators of a new product,

which, we believe, is a more accurate reflection of technology improvement. Therefore, we

propose:

Hypothesis 1: Technology performance improvements positively affect innovation diffusion for digital cameras.

Dominant Design’s Positive Impact on Innovation Diffusion

The concept of dominant designs

In the beginning of a new product category, the technology is in a ferment stage; there are a large

number of new firm entries and many design patterns that meet the needs of different segments

of customers and the technological uncertainty is great. As the technology evolves, certain

features will be incorporated, while others are abandoned. Gradually, a dominant design

emerges, which marks the end of the ferment stage and the beginning of the incremental stage,

during which technology will progress along the trajectory defined by the dominant design

(Abernathy and Utterback, 1978, Dosi, 1982, Utterback and Suarez, 1993, Klepper, 1996). “A

dominant design is a specific technological path, along an industry’s design hierarchy, which

establishes dominance among competing designs” (Utterback and Suarez, 1993). The emergence

of the dominant design is determined by technical, market and organizational factors, and has

important implications on the adoption of innovations and the survival of firms (Anderson and

Tushman, 1990, Tushman and Rosenkopf, 1992, Suarez and Utterback, 1995).

9


The term, “dominant designs” and “standards,” are used interchangeably in many

contexts. Usually, a dominant design embodies multiple standards. “In many cases and by

implication, a dominant design becomes the industry standard, or for complex assembled

products with many parts, embodies a collection of related standards.” (Suarez and Utterback,

1995).

Dominant designs generated by architectural innovation and component innovation have

different effects on the survival of firms (Christensen, Suarez and Utterback, 1998). Architecture

is the way in which the “components are integrated and linked into a coherent whole”

(Henderson and Clark, 1990), or the list of components of a system and their roles (Baldwin and

Clark, 2000). Architectural innovations change the way in which the components of a product

are linked together, while leaving the core design concepts and the basic knowledge underlying

the components untouched, and have far-reaching impacts on the competitiveness of established

firms (Henderson and Clark, 1990). Interface is a “detailed description of how the different

modules will interact, including how they will fit together, connect, communicate” (Baldwin and

Clark, 2000). Thus, architectural innovation is innovation at the interface. Previous studies have

looked at the role of interfaces on the success of firms (e.g., Cusumano and Gawer, 2002). Yet,

there has been no clear distinction between two very different types of interfaces, which we call

external architecture and internal architecture. While internal architecture governs the way in

which the components of the technology interact with each other, external architecture governs

the way in which the focal technology interacts with other technologies. Figure 3, illustrates this

idea: study of system A1 includes studying its components A11 and A12, how A11 and A12

interact with each other (internal architecture) and how A1 interacts with A2 and A3 (external

architecture).

10


-----------------------------------------

Insert Figure 3 about here ------------------------------------------

This distinction provides more clarity to our conceptualization concerning design,

architecture and innovation. It stresses what the focal system is in the analysis and will help us to

understand the roles played by different types of interfaces. We argue that dominant designs in

internal architecture (internal dominant design) and external architecture (external dominant

design) may impact diffusion through different channels. This distinction is more meaningful for

products that depend on complementary products within the larger system they belong to and in

markets where network effects are present.

We conjecture that the importance of external dominant design is positively related to the

degree to which the new product is dependent on the bigger system. If the connection of the new

product to its complementary products in the bigger system is important to its usage, then the

influence will be stronger. We also conjecture that the importance of external dominant design

on diffusion is contingent upon the strength of network effects. Network effects are a

consumption externality a user derives from consumption of a good when the number of other

consumers who purchase compatible items increases. It can be either a direct physical effect as in

the case of telephones and fax machines, or an indirect effect as in the case of computer

hardware and software, where consumers of a product derive more utility when the market for its

complementary products expands (Farrell and Saloner, 1985, Katz and Shapiro, 1985, Farrell and

Saloner, 1986). Network effects can precipitate and enlarge initial differences between different

technologies (Arthur, 1989). When two standards compete with each other, and one of them has

a small initial advantage, that advantage is going to be magnified into bigger differences and

eventually, one or a few designs will dominate the others. A dominant design can facilitate the

11


connection between a technology and its complementary products; therefore, the technology can

take further advantage of the diffusion of its complementary products through indirect network

effects. A dominant design can also facilitate the communication between different users of

compatible technologies through a direct network effect. In the case of digital cameras, cameras

can be linked to the larger system in a more consistent way and can take better advantage of the

indirect network effect enabled by diffusion of its complementary products. Of course, we

cannot statistically analyze these two conjectures, since we are focusing our study on only one

product. We will discuss this further in our section on limitations and directions for future

research.

The impact of dominant designs on innovation diffusion

The emergence of a dominant design positively impacts sales, and is a prerequisite to

mass adoption and volume production (Tushman and Anderson, 1990). In a study of cement,

glass and minicomputer industries, it was found that sales of all versions of a new technology

peaked after the emergence of a dominant design (Tushman and Anderson, 1990). Here, we

categorize the arguments they provided according to whether they address internal dominant

design or external dominant design and we also provide additional theoretical support. Basically,

we contend that the internal dominant designs directly impact the producers, and the external

dominant designs directly impacts the consumers. We propose that the formation of a dominant

design can impact diffusion through the following channels:

Emergence of an external dominant design can reduce the risks associated with adoption

when multiple competing designs exist. During the era of ferment, potential consumers have to

choose one design among multiple competing designs. If they are locked in to an external

12


architectural design that is not the dominant design, they will have compatibility problems with

connectivity to other systems if the designs are not compatible (Anderson and Tushman, 1990).

This problem does not exist for internal architecture, because internal architecture is concerned

only with connectivity within the product and does not deal with connectivity to other systems.

In the example of digital cameras, if a consumer buys a camera that uses only RS232C

(recommended standard-232C) as the connection standard between the camera and a computer,

the consumer can connect only to computers that use that interface standard. On the other hand, a

consumer does not care and is even not aware if the digital camera uses CCD or CMOS

(Complementary Metal Oxide Semiconductor) as the sensor, because that design does not affect

how the camera relates to complementary devices such as a computer or printer.

The dominant design of external architecture can positively affect diffusion by enhancing

ease of use of a new product. Diffusion theory indicates that complexity, the degree to which an

innovation is perceived to be difficult to use, is negatively related to adoption (Rogers, 2003).

On the other hand, the technology acceptance model also indicates that when a technology

product is perceived to be easy to use, it is more likely to be adopted (Davis, 1989). From

consumers’ perspective, dominant designs can reduce product class confusion (Anderson and

Tushman, 1990). When the number of types of interactions among a technology and other

technologies decreases, the consumers are likely to find it easier to use that technology.

Sometimes, to guard against uncertainty concerning which design will become dominant,

producers will combine multiple designs in one product, making it difficult for consumers to

manage the product. But once a dominant design is formed, this cautious approach on the

producer’s side is no longer necessary.

13


If we take the example of a digital camera, we can see that it belongs to the larger system

of digital imaging products and the even larger system of computer technologies. The pictures

taken by digital cameras need to be downloaded to computers to be manipulated, and they need

to be in a specific format in order to be processed by imaging processing software. Therefore, its

relationships with computers, printers, and imaging processing software are keys to its adoption.

When an external dominant design is formed, there emerges a standard way of interaction

between the camera, and the computer, the printer and the internet. There used to be many types

of interfaces between digital cameras and complementary products such as computers and

printers. In 1999, 30.14% of cameras used USB (Universal Serial Bus) ports, 67.18% of cameras

used serial ports, 7.31% of camera used parallel ports, and 6.85% of cameras used IEEE (A

standard established by Institute of Electrical and Electronics Engineers, Inc.) ports (numbers

add to more than 100%, since some cameras used multiple standards). Now, almost all cameras

use USB ports. It is much easier for users to use only a USB standard than using both a USB

standard and an IEEE standard. The use of JPEG (Joint Photographic Experts Group – the

original name of the committee that wrote the standard) as an image processing standard has a

similar effect, and when most cameras store images in JPEG, users will be able to manipulate,

exchange, and transfer images more easily.

Hypothesis 2: The emergence of a dominant design in external architecture positively affects new product sales for digital cameras. Dominant designs of internal architecture can impact diffusion by decreasing the cost of

a new product. Dominant designs of internal architecture influence primarily the producers of the

new product. The emergence of a dominant design is a prerequisite to volume production

(Anderson and Tushman, 1990). “Dominant designs permit firms to design standardized and

interchangeable parts and to optimize organizational processes for volume and efficiency

14


(Abernathy, 1978; Houndshell, 1984). They permit more stable and reliable relations with

suppliers and vendors and consumers (Anderson and Tushman, 1990). Before a dominant design

is formed, economies of scale will have little effect, because a large number of variants of a

product need to be produced by many entering firms. After a dominant design is formed, the

products incorporating the dominant designs can be produced in larger quantities, and thus prices

can drop accordingly (Suarez and Utterback, 1995). When costs decrease while performance

stays at the same level or even increases, the relative advantage will increase and the technology

becomes more attractive to potential adopters. Since there is less uncertainty about which type

of external architecture will become the dominant design, producers no longer have to combine

multiple designs in one product, and thus, the cost for producing that product will drop. Thus, we

propose:

Hypothesis 3: The emergence of an internal dominant design positively affects new product sales for digital cameras.

The positive effects may or may not be linear, depending on what point in the

development process we examine, because once the market penetration reaches a certain point,

no matter how technology improves, the market may react only slightly. While not formally

testing it, we suggest that the importance of external dominant design is contingent upon the

dependency of the new product on the larger system it belongs to, and the strength of the

network effect the new product exhibits.

METHOD

Data Sources

We have two types of data: one is digital camera attribute data, which reflect the

technology evolution of digital cameras; the other is digital camera sales data, which reflect the

15


market evolution of digital cameras. Camera attribute data are provided by Lyra Research,

which is a premier consulting firm in the imaging industry. Camera sales data are provided by

three sources: Lyra Research, Photo Marketing Association (PMA) and IDC (International Data

Cooperation), a global provider of market intelligence and consulting services for technologies.

We used the data provided by IDC to base our analysis on and validated our research with data

provided by the other two firms, and found the results stayed robust across the different data sets.

Lyra’s digital camera database was created in 1996 and updated with new information over the

past 11 years. Up to June 1, 2005, the database consisted of information on 1658 types of digital

cameras shipped in the United States and 1248 types of digital cameras shipped in other

countries, mostly Japan, Germany and United Kingdom. Although cameras are shipped in

different countries, the producers are the same group of international companies. Lyra Research

gets information on cameras from the manufacturers. Lyra makes an effort to include every

camera shipped, and we believe that the Lyra database covers at least 95% of camera products

shipped in the United States and at least 80% of the cameras shipped in other countries. The

camera information in the database include: shipment date, product name, initial price, image

resolution in pixels, weight in ounces, height, width and depth of the camera in inches, sensor

type, interface type, image compression type, whether removable storage is included, whether

LCD Viewer is included, etc. We filled in some missing data points by camera information

provided by various online camera databases, especially http://www.digicamhistory.com. We

also made corrections to some errors in terms of the units of weight, and size during our initial

data processing. We discarded data on cameras whose information was incomplete and discarded

all data on digital camcorders, since they tend to be relatively heavy and have low resolution

16

http://www.digicamhistory.com/


when used as cameras. The dataset we use in our analyses includes about 80% of the cameras in

the Lyra database.

The IDC sales data provided were acquired from retail firms in the industry

including major chains. From 1996 to 1998, they have only yearly data. Starting in 1999, they

collected quarterly sales data to reflect the fast-changing market conditions.

Measurements

Digital camera sales. To trace innovation diffusion, quarterly sales in units for cameras shipped

in the U.S. during the first quarter of 1999 to the first quarter of 2005 are used. For the years

1996, 1997 and 1998, only yearly sales figures were available, so we weighted those years’ data

based on the quarterly percentages for 1999 – 2005 to approximate those years’ quarterly sales

data. The diffusion S-curve of digital cameras is examined. To track the technology S-curve, we

used key performance indicators of digital cameras. A digital camera is a system product, in that

it is composed of many subsystems. The performance of a digital camera is the integrated output

of all subsystems, which can be reflected by key performance indicators.

Resolution per dollar. Most consumers agree that the most important indicator of technology

performance is resolution per dollar. Both median resolution per constant 2004 U.S. dollar and

minimum resolution per constant 2004 U.S. dollar were analyzed and it was found that they

provide similar information. Resolution change reflects technology change: more resolution

means clearer pictures. But the cost to achieve the same amount of resolution in different years is

very different. For example, the cost of a 1.5 mega pixel camera in 1996 was about $12000, in

2000 it cost only about $500, and in 2005, a 4 mega pixel camera costs only about $200. So,

resolution per dollar can reflect technology performance change more accurately than resolution.

17


It also reflects the relative advantage advocated by Rogers (2003) more precisely than resolution

alone.

Camera weight. The second factor is camera weight in ounces. We expect weight to be

negatively related to sales, because consumers generally prefer lighter cameras to heavier ones.

Camera size. The third is camera size in cubic inches, the product of height, length and width.

We expect the size to be negatively related to sales, because consumers generally prefer smaller

cameras.

We also trace technology evolution by the formation of dominant designs. We analyze

three types of dominant designs: a dominant design for internal architecture, a dominant design

for external architecture and a full dominant design. A previous study in 2002 pointed out that

for digital cameras, the sensor, the LCD display, the computer interface, and the removable

storage combine to form the dominant design features of a camera (Zelton, 2002). Beside these

elements, we introduce one more aspect: the flash lighting.

Dominant design of internal architecture. We consider the internal dominant design for digital

cameras to include the following aspects: using a CCD as the image sensor, possessing an LCD

viewer for image viewing and instant review of the image captured, and possessing an internal

flash for lightening. An image sensor is an electronic device that can transform light signals into

electronic signals. It is the most important and expensive component of a digital camera and

from the beginning, the CCD has been the prevalent sensor. The other major alternative sensor is

the CMOS. The LCD Viewer is the screen that displays both images to be captured and images

already captured. The built-in flash can help to capture clear images in a dark background. These

three components are incorporated in almost all digital cameras.

18


Dominant design of external architecture. We consider the external architecture of digital

cameras to include the following aspects: using JPEG as the file compression standard, using

USB as the interface standard between the camera and computer, and the inclusion of removable

storage. A digital camera is a component in a bigger system - a networked computer system. It is

important to ensure the connection between the digital camera and the bigger system. The

storage medium is not just about storage; it also represents the link between the camera and the

outside. It can be used to transfer images from the camera to the computers and the printers. USB

is an external peripheral interface standard for communication between a computer and external

peripherals over an inexpensive cable using serial transmission. JPEG is a standardized image

compression mechanism. JPEG compression can make image files smaller and make it easier for

transmitting files across networks and for archiving libraries of images.

The formation of a dominant design is a process; the key elements of the dominant design

can emerge separately in different products, then there starts to be products that embody all the

elements of the dominant design, and eventually all the elements of the dominant design are

incorporated by nearly all products (Christensen, et al., 1998). Consistent with the previous work

of Christensen, Suarez and Utterback, we trace the percentage of cameras incorporating these

dominant design concepts to measure the formation of the dominant design.

Descriptive Analysis: Tracing Technology Evolution and Innovation Diffusion

The innovation diffusion and technology evolution S-curves were displayed in Figures 2

and 3, utilizing the curves of median resolution per 2004 dollar and cumulative sales in units for

digital cameras. We put the two curves together in Figure 4, along with two other

aforementioned performance measures, camera size and camera weight. It can be seen from

Figure 4 that the technology curve for resolution per 2004 dollar and the diffusion curve closely

19


move together over time. Correlation analysis indicates that the two series are highly correlated.

However, it is difficult to deduce which curve, if either, drives the other.

It can also been seen from Figure 4 that weight and size indices do not correlate nearly as

closely with sales as does resolution per 2004 dollar. This may indicate that weigh and size are

not as important to diffusion as resolution per dollar. Indeed, when predicting market evolution,

it is important to identify the most important attribute.

------------------------------------------- Insert Figure 4 about here

------------------------------------------

We consider a dominant design to be formed when the first products that incorporate all the key

features of the design appear on the market, and we found that for digital cameras, an internal

dominant design was formed in the second quarter (Q2), 1996 and both an external dominant

design and a full dominant design were formed in Q2, 1998. Thus, the dominant design for

internal architecture was formed far ahead of the dominant design for external architecture. Thus,

for digital cameras, the evolution of the full dominant design was primarily constrained by the

external architecture. Figure 5 shows the dominant design formation process. First, the dominant

design for internal architecture was formed in the second quarter of 1996; then the dominant

design of external architecture was formed in the second quarter of 1998.

------------------------------------------

Insert figure 5 about here ----------------------------------------

Regression Analysis: The Impact of Technology Evolution on Innovation Diffusion

Based on the descriptive data analysis in the previous section, we employ regression

analysis to estimate the impact of technology evolution on market evolution. We believe that a

20


log linear model can reflect the relationship between technology evolution and market evolution

more accurately than a linear model. As the performance of the technology improves, more sales

can be generated; however, we expect the positive effect to diminish as technology progresses

further. We experimented with both linear and log-linear models and found that the log-linear

models fit the data better and, also, an auto-correlation problem encountered with linear models

was eliminated when we used log-linear models.

Table 1 summarizes the hypotheses, constructs, their operational definitions and the data

sources of our study/analysis. Camera weight and size are not included in the models because

they become insignificant once resolution per 2004 dollar is included in the model, indicating

resolution per 2004 dollar is more important than weight and size and it is sufficient to include

solely resolution per 2004 dollar as the key performance indicator.

----------------------------- Insert Table 1 about here -----------------------------

We suspect that there may be a lag between the performance of cameras in one quarter

and the sales of those cameras. Thus, we experimented with technology performance leading one

quarter, two quarters, three quarters, and four quarters ahead of sales; yet, we found that the

model with technology performance with the same quarter as sales fits the data best. This result

suggests that the technical performance of digital cameras has an instant, or at least “very fast,”

impact on sales.

RESULTS

Table 2 shows the summary statistics for, and correlations among, our variables:

---------------------------- Insert Table 2 about here

21


----------------------------

Table 3 summarizes the results of our regression analyses. Durbin-Watson values are all

close to 2, indicating that autocorrelation is not a concern. In each model (regression), the

dependent variable is (quarterly) Sales, and the “quarter” variable is coded as a linear trend

variable, ranging from 0 to 36. Since our hypotheses are clearly directional, all p-values reflect

one-tail tests.

------------------------------ Insert Table 3 about here ------------------------------

It can be seen that our hypotheses are supported by the regression models. Model 1

supports hypothesis 1, showing that a technology performance increase, in the form of higher

resolution per 2004 dollar, has a significant positive effect on diffusion. Model 2, goes one step

further by adding the variable “quarter”, to show that resolution per 2004 dollar is significant

above and beyond the time effect. But since “quarter” and “resolution per dollar” are somewhat

correlated, when “quarter” and “resolution per dollar” are in the same model, sufficient

collinearity occurs to render the effect of time (quarter) insignificant (p >.25).

Model 3 supports hypothesis 2, showing that emergence of an external dominant design

has a positive and significant effect on diffusion, over and above the impact of time. Model 4

supports hypothesis 3, showing that the emergence of an internal dominant design has a positive

and significant effect on diffusion, over and above the impact of time. All models have a high

value of R2 (each > .90).

All three factors (resolution, external dominant design and internal dominant design) are

significant in their individual models; thus, all our hypotheses are supported by our analyses.

22


Although we cannot claim that technology evolution causes market evolution, we can safely

conclude that both dominant design and technology evolution are significantly positively

associated with the diffusion of digital cameras. A remaining question that would be reasonable

to ask would be whether the three variables (resolution per 2004 dollar, internal dominant design,

external dominant design) are really “one and the same” variable. In Table 3, the results of model

5 indicates that, indeed, this “one and the same” is not the case. With all three key variables,

along with quarter, in the model, all three key variables are significant, indicating that each of the

three variables, above and beyond the other two (and above and beyond “quarter”), contribute to

the value of sales.

CONCLUSION AND DISCUSSION

Through an empirical study of the technology changes and sales of digital cameras in the

U.S., we have demonstrated that technology performance improvements and internal and

external dominant designs have positive and significant effects on the sales of digital cameras.

Digital cameras belong to a larger system of digital photography and a market with network

effects. We suggest that external architecture is more important for more complex products that

depend on a larger system than for simpler products. According to the taxonomy of Tushman and

Rosenkopf, there are 4 types of technologies in terms of their complexity: non-assembled

products; simple assembled products, closed systems and open systems (Tushman and

Rosenkopf, 1992). Open systems are composed of technologies that are linked to each other

through interface technologies. So, the effect of dominant design for external architecture on

diffusion will likely be stronger for open systems than for simpler systems. We suggest further

that the influence of external dominant design is more important for products that exhibit

23


network effects than for products that do not. Network effects will magnify the impact of

external dominant designs on diffusion.

This paper has the following contributions: It advances innovation diffusion research by

exploring how changes in technology affect the change in diffusion for a product that belongs to

a larger system where network effects are present; it traces technology evolution by both the

technology S-curve and the dominant designs and thus, it provides a fuller picture of the

technology evolution process; it contributes to innovation theory and dominant design theory by

distinguishing internal dominant design and external dominant design, and has demonstrated the

linkage between the different types of dominant designs and innovation diffusion. It is possible

that the distinction can have other impacts as well, such as the survival of firms. Finally, for

practitioners, this study provides a potential new way to forecast sales growth and, based on the

technology performance increase of their products, firms can predict future sales with more

confidence and make more informed decisions.

The following are possible limitations on our study and suggest further study: The

relationship between technology evolution and diffusion is more complex than what we have

explored. The causality may be reversed; it may be that the market growth attracted firms to

invest in technology for that market. There may also be a dual process: product diffusion and

technology evolution influence each other. It is also possible that during different stages of

development, the causality between market evolution and technology evolution changes: in the

beginning, the technology pushes the market; after the market takes off, the market pulls the

technology. For different types of technology the magnitude of influence and direction of

causality between technology evolution and diffusion may be different. The technology

improvement may increase market growth through different channels for different users: For

24


higher-end cameras, the driver may be performance; for consumer mid- to lower-end cameras,

the driver may be price; for special purpose cameras: medical, agriculture, etc., the driver may be

special features; for cameras designed for fun use, the driver may be decrease in size and price.

In addition, we have looked only at the early stage of the digital camera market and

technology development, with the inflection point in neither the technology S-curve nor the

diffusion S-curve having been reached. We may be able to uncover even richer information if we

could look at the entire S-curve. More sophisticated research designs can be made and more

rigorous methods can be employed to uncover more details concerning the relationship between

the dual evolution processes of market and technology. Performance indicators are

multidimensional, and in different stages of technological progress the key indicator may be

different (Christensen, 1997). We have looked in detail only at one indicator, resolution per 2004

dollar. When one indicator is no longer a concern, some other indicators may become key.

Future studies can look at the effect in the transition of key performance indicators on diffusion.

Finally, we repeat that we have examined only one product--digital cameras. If one were

to study a multitude of products, one would be able to statistically test for interaction effects that

we have postulated—particularly interaction effects between each type of dominant design and

both the degree to which the product is dependent on a larger system, and the strength of network

effects.

ACKNOWLEDGEMENTS

It has also benefited from feedbacks form Professor Fernando Suarez, Professor Jeff

Furman and Professor George Wyner, and the guidance of Professor Venkatraman, Professor

John Henderson and Professor Mark Gaynor at Boston University, School of Management. The

25


authors are also grateful to the support from the industry during data collection process,

including the support from Lyra Research, IDC and Photo Marking Association. The authors

especially want to thank Peter Zelten at Polaroid, who based his thesis at MIT on digital imaging,

for his graceful support.

REFERENCES

Abernathy, W. J. and J. M. Utterback, Patterns of Industrial Innovation, 1978, Cambridge, MIT Press Agarwal, R. and B. L. Bayus, The market Evolution and Sales Takeoff of Product Innovations. Management Science, 48, 8, 2002, 1024 Anderson, P. and M. L. Tushman, Technological Discontinuities and Dominant Designs: A Cyclical Model of Technology Change. Administrative Science Quarterly, 35, 1990, 604-633 Arthur, W. B., Competing Technologies, Increasing Returns, and Lock-in By Historical Events. The Economic Journal, 99, 1989, 116-132 Baldwin, C. and K. Clark, Design Rules: The Power of Modularity, 2000, Cambridge, MA, MIT Press Bass, F. M., Product Policy. The Journal of Business, 53, 3, 1980, IIS51 Christensen, C. M., Exploring the Limits of the Technology S-Curve. Part I: Component Technologies. Production and Operations Management, 1, 1992, 334-366 Christensen, C. M., Patterns in the Evolution of Product Competition. European Management Journal, 15, 1997, 117-127 Christensen, C. M., The Innovator's Dilemma, 2003, New York, Harper Business Essentials Christensen, C. M., F. F. Suarez and J. M. Utterback, Strategies for Survival in Fast-Changing Industries. Management Science, 44, 12, 1998, 207-220 Davis, F. D., Perceived Usefulness, Perceived Ease of Use and User Acceptance of Information Technology. MIS Quarterly, 13, 3 (September), 1989, 319-340 Dosi, G., Technological paradigms and technological trajectories. A suggested integration of the determinants and directions of technical change. Research Policy, 11, 1982, 147-172

26


Farrell, J. and G. Saloner, Standarization, compatibility, and innovation. Rand Journal of Economics, 16, 1, 1985, 70-83 Farrell, J. and G. Saloner, Installed base and compatibility: Innovation, product preannoucements and predation. American Economic Review, 76, 1986, 940-954 Foster, R., Innovation, The Attacker's Advantage, 1986 Geroski, P. A., Models of technology diffusion. Research Policy, 29, 2000, 603-625 Golder, P. N. and G. J. Tellis, Will It Ever Fly? Modeling The Takeoff of Really New Consumer Durables. Marketing Science, 16, 3, 1997, 256-270 Gort, M. and S. Klepper, Time Paths in the Diffusion of Product Innovations. The Economic Journal, 92, 367, 1982, 630-653 Griliches, Z., Hybrid Corn: An Exploration in the Economics of Technological Change. Econometrica, 25, 4, 1957, 501-522 Henderson, R. M. and K. Clark, Architectural Innovation: The Reconfiguration of Existing Product Technologies and the Failure of Established Firms, 1990 Katz, M. and C. Shapiro, Network externalities, competition and compatibility. American Economic Review, 75, 1985, 424-440 Klepper, S., Entry, exit, growth, and innovation over the product life cycle. The American Economic Review, 86, 3, 1996, 562 Lilien, G. L. and E. Yoon, The timing of competitive market entry: an exploratory study of new industrial products. Management Science, 36, 5, 1990, 568-585 Mahajan, V., E. Muller and F. M. Bass, New Product Diffusion Models In Marketing: A Review And Direction for Research. Journal of Marketing, 54, 1, 1990, 1 Mahajan, V. and Y. Wind, New Product Forecasting Models - Directions for Research and Implementation. International Journal of Forecasting, 4, 3, 1988, 341-358 Norton, J. A. and F. M. Bass, A Diffusion Theory Model of Adoption and Substitution for Successive Generations of High-Technology Products. Management Science, 33, 9, 1987, 1069 Rogers, E. M., New Product Adoption and Diffusion. Journal of Consumer Research, 2, 4, 1976, 290-301 Rogers, E. M., Diffusion of Innovations, 2003, New York, Free Press Ryan, B. and N. C. Gross, The Diffusion of Hybrid Seed Corn in Two Iowa Communities, 1943

27


Suarez, F. F. and J. M. Utterback, Dominant designs and the survival of firms. Strategic Management Journal, 16, 6, 1995, 415 Tushman, M. and L. Rosenkopf, Organizational determinants of technological change: towards a sociology of technological evolution. Research in Organizational Behavior, 14, 1992, 311-347 Tushman, M. L. and L. Rosenkopf, Organizational Determinants of Technological Change: Towards a Sociology of Technological Evolution. Research in Organizational Behavior, 14, 1992, 311-347 Utterback, J. M. and F. F. Suarez, Innovation, competition, and industry structure. Research Policy, 22, 1, 1993, 1 Venkatesh, V., M. G. Morris, G. B. Davis and F. D. Davis, User Acceptance of Information Technology: Toward a Unified View. MIS Quarterly, 27, 3, 2003, 425-478 Zelton, P., Digital Photography and the Dynamics of Technology Innovation, 2002

FIGURE 1 Median Resolution Per 2004 Dollar for Digital Cameras Shipped in the U.S.

28


0

5000

10000

15000

20000

25000

0 10 20 30

quarters (quarter 1, 1996 - quarter 1, 2005)

med

ian

reso

lutio

n pe

r 200

4 do

llar

40

median resolution per 2004 dollar

FIGURE 2 Cumulative Sales for Digital Cameras Shipped in the U.S.

29


0

10000000

20000000

30000000

40000000

50000000

60000000

70000000

80000000

0 5 10 15 20 25 30 35 40

quarters (Quarter 1, 1996 -Quarter 1, 2005)

cum

ulat

ive

sale

s in

uni

ts

cumulative sales in units

FIGURE 3 Internal and External Architectures

System A3 System A2

System A12 System A11

System A1

FIGURE 4

Evolution of Digital Camera Attributes

30


Index of picture resolution, camera size, camera weight and cumulative sales

0

20

40

60

80

100

0 10 20 30quarters(Q1 1996-Q12005)

Inde

x

40

Index of Cumulative Sales Index of Median Resolution Per Dollar

index median size index median w eight

FIGURE 5 Tracing the Emergence of Dominant Designs for Digital Cameras

-20

0

20

40

60

80

100

0 5 10 15 20 25 30 35 40

Quarters (Q1 1996-Q1 205)Per

cent

age

of p

rodu

cts

havi

ng a

set o

f des

ign

feat

ures

external dominant design internal dominant design full domianant design

TABLE 1

31


Summary of Hypotheses, Constructs, Their Operational Definitions and the Data Sources

Hypotheses Constructs Operational definitions Sources Hypothesis 1. Technology performance improvement positively affects innovation diffusion. Innovation Diffusion

Natural log form of quarterly digital camera shipments in units in U.S.

IDC quarterly sales data

Technology performance improvement

Natural log of resolution per 2004 dollar

Lyra Research camera data

Hypothesis 2. The formation of a dominant design in external architecture positively affects diffusion

Emergence of external dominant design

Natural log of percentage of digital cameras that use JPEG, SUB, removable storage


Hypothesis 3. The formation of a dominant design in internal architecture positively affects innovation diffusion.

Emergence of internal dominant design

Natural log of percentage of digital cameras that has CCD, LCD and flash


TABLE 2 Summary Statistics and Correlations

Summary Statistics Correlations Min Max Mean s.d. 1 2 3 4 5 1. Sales 10.88 16.04 13.69 1.42 1.00 0.96 0.60 0.91 0.96 2. Resolution 6.11 9.70 7.94 1.14 0.96 1.00 0.53 0.90 0.99 3. Internal dominant design 0.00 4.61 4.12 1.05 0.60 0.53 1.00 0.54 0.52 4. External dominant design 0.00 4.61 2.99 1.90 0.91 0.90 0.54 1.00 0.88 5. Quarter 0.00 9.00 4.19 2.66 0.96 0.99 0.52 0.88 1.00

TABLE 3 Results of Regression Analyses (t-statistics in parentheses)

32


33

Independent variables Model 1 Model 2 Model 3 Model 4 Model 5

Resolution per 2004 dollar 1.20 (2.13)* 1.00 (2.78)* .65(1.91)*

External dominant design .24 (3.39)* .16(2.32)*

Internal dominant design .19 (2.56)* .15(2.26)* Quarter 0.09 (0.56) .36 (7.24)* .47 (16.17)* .16(.78) R-squared 0.928 0.929 0.935 0.927 0.949 Adj R-squared 0.926 0.925 0.931 0.922 0.943 *p< 0.05 n=37

Date post:	09-Feb-2022
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

The Design Evolution of Digital Cameras and Its Implications

Documents