3D Printing: Media Hype or Disruptive Innovation?

(1)

3D Printing: Media Hype or Disruptive Innovation?

From niche to mainstream

Master Thesis Msc. Marketing (MM)

Name Student: Ivo Furda Student ID number: s1854356

Student email: i.j.furda@student.rug.nl Date Thesis: 22/06/2015

Name Supervisor: prof. dr. P.C. Verhoef

(2)

1

(3)

2 ABSTRACT

The main purpose of this paper is to get a better grasp on the trajectory path of 3D Printing. By

use of extant literature coverage, the theoretical part of the paper aims to clarify the processes new

technologies, respectively innovations, need to go through in order to become mainstream. The

quantitative part of the paper zooms in on eleven purely 3D Printing companies. By use of Google

search queries, for the years 2010-2015, a preliminary attempt is conducted in order to explain

stock price movements of these eleven 3DP companies over the last couple of years. Contrary, to

the well-known theory of stocks being a random-walk some additive, significant, explanations are

found for the stock price movements of the companies within our sample.

(4)

3

Introduction

Additive manufacturing (AM), more colloquially known as 3D printing (3DP), first emerged in 1987 with stereo lithography from 3D Systems, a process that solidifies thin layers of ultraviolet light‐sensitive liquid polymer using a laser (Wohlers & Gornet, 2011). In late 1987, 3D Systems shipped its first beta units to customer sites in the U.S., followed by production systems in April 1988. These were the first commercial additive‐manufacturing system installations in the world.

In general, the term 3DP covers a host of processes and technologies that offer a full spectrum of possibilities for the production of parts and products in different materials. It should be stated that 3DP does not comprise one single technology but entails a rather broad subset of different ones. Essentially, what all of the processes and technologies have in common is the manner in which production is carried out - layer by layer - in an AM process.

In the past few years, this method of additive manufacturing has taken investors, scientists, tech geeks and the rest by storm. Descriptions w.r.t. possibilities range from exciting current uses in medical, manufacturing and other industries to futuristic ideas — such as using 3D printers on asteroids and the moon to create parts for spacecraft and lunar bases. On the 12

^th

of May 2013, at the annual State of the Union

¹

, President Barack Obama called 3DP a technology that has the

“potential to revolutionize the way we make almost everything”. 3DP is even dubbed as the third industrial revolution by the Economist

²

magazine. Some even argue that 3DP could up-end the last two centuries of approaches to design and manufacturing with profound geopolitical, economic, social, demographic, environmental, and security implications.

Today, approximately 40 manufacturers sell the 3D printers most commonly used in businesses, and over 200 startups

³

worldwide are developing and selling consumer-oriented 3D

1 The White House (2013) Remarks by the President in the State of the Union Address,

https://www.whitehouse.gov/the-press-office/2013/02/12/remarks-president-state-union-address (Accessed: 1st of May 2015)

2 The Economist (2012) The third industrial revolution, http://www.economist.com/node/21553017 (Accessed: 1st of May 2015)

3 Gartner (2014) Gartner Says Consumer 3D Printing Is More Than Five Years Away http://www.gartner.com/newsroom/id/2825417 (Accessed: 1st of May 2015)

(9)

8 printers. As analysis of AM still is scarce and has predominantly focused on production cost or other firm level aspects (e.g., Mellor et al. 2014; Petrovic et al. 2011; Ruffo and Hague 2006), this paper focuses on the locus of innovation and progression of 3DP. The main purpose of this paper is to get a better grasp on the trajectory path of 3DP.

By use of theories, the first two chapters aim to clarify the processes new technologies, respectively innovations, need to go through. As 3DP can be classified as a ‘potentially’ disruptive technology, Chapter 1 zooms in on the theory of disruptive technology (DT). The term DT was first introduced by Bower and Christensen in their seminal journal article (Bower & Christensen, 1995), it posited that disruption occurred when an initially inferior technology introduced by a new entrant improved to meet the needs of the mass market. Extant literature and debate has shown that defining a technology’s “disruptiveness” is not a standard line of process.

Whether “today’s” 3DP already can be classified as a true disruptive technology therefore cannot easily be answered. As 3DP is not a single technology, it can be argued that the answer to this specific question, to some extent centers more around definitions than around applicabilities.

As Information Technology (IT) is such an important aspect of the speed, magnitude and applicability of 3DP, some emphasis has been put on the practical implications IT intrinsically offers. As Campbell (2011) stated: “The Internet first eliminated distance as a factor in moving information and now AM eliminates it for the material world.”

Chapter 2 chronicles around another theory, which clarifies the market’s expectations among a “single” technology over time, respectively the Gartner Hype Cycle (GHC). From 1995 onwards, using consensus, Gartner analysts started to position technologies based on hype. The GHC provides a snapshot of the position of technologies along a predictable pattern of enthusiasm, disillusionment and eventual realism. It highlights technologies that are the focus of attention because of particularly high levels of hype.

As 3DP entails a broader subset of technologies and different technologies do not move at

the same speed through the hype curve (O’Leary, 2008), when we zoom in, within every phase of

the Hype Cycle 3DP applications and techniques show up. Each for a different market, many using

different technologies, but all overlapping in the sense that fabrication will be done layer-by-layer.

(10)

9 Since 2012-2013 Gartner’s, distinct, vision w.r.t. 3DP became that fundamentally, there are two markets here, the consumer and enterprise market. Both driven by different uses and requirements.

Chapter 3 aims at sketching a better understanding of speed, market size and broader implications of 3DP. As many challenges remain ahead for us as a society if this technology would be fully adopted, some controversies are being outlined to the reader. One should think in term of the borders between the private and public spheres, the ratio between security and freedom and so on. The remainder of Chapter 3 will zoom in on the size of the market, the trajectory path of 3DP, and the major market players of Computer Aided Design software.

The quantitative part, Chapter 4, 5 and 6, of the paper zooms in on explaining stock price movements of 3DP companies. One should think of 3D printer Manufacturers and suppliers of Computer Aided Design (CAD) Software. Chapter 4 explains the hypotheses. Chapter 5 embarks on explaining the model, selection criteria and data collection. The selection criteria are being applied in order to find and classify ‘purely’ 3DP companies. Eleven companies fulfill these criteria. An overview of the sample can be found in Figure 7 within Chapter 5.

By use of a regression model, it is being investigated whether explanatory power for stock price movements of these eleven companies can be found. As Choi and Varian (2012) have shown that data from Google Trends can be linked to current values of various economic indicators, including automobile sales, unemployment claims, travel destination planning and consumer confidence, Google Trends search queries are a vital part of our regression model. As Google search frequency data is a vital part of the analysis, clarification is given in Chapter 4 regarding the quality of this data.

The first hypothesis chronicles around the thought that increased popularity, increased search volume, is reflected within the stock prices of the companies which are affected by this.

The same positive relation is assumed w.r.t. the second hypothesis between the stock price

movements of these companies and the amount of Google search queries which best describes the

sector, in this case “3D Printing”. The third hypothesis assumes a positive correlation between

these two variables, over time. As an example, if people are becoming more interested in “3D

Printing”, it is assumed that the popularity of “3D Systems” would subsequently be on the rise as

well.

(11)

10 Besides the hypotheses, emphasis has been put, in the first part of Chapter 6, on depicting trends and explanations, w.r.t. revenue and income streams, of the individual company and for the sample as a whole. A more detailed description of the sample, data collection and hypotheses can be found in Chapter 4 and 5. Chapter 6 reports the results of the study. The final chapter, Chapter 7, deals with discussing extensions and adoptions for future research and concludes on the results found within the qualitative part (Chapter 1 – 3) as well as the quantitative part (Chapter 4 – 6) of this study.

(12)

11

1. Innovation

Innovation is often described as the predominant source of economic growth. This finding is called

“Solow’s surprise” by Easterly (2001) and is listed by King and Rebelo (1999) as one of the

“stylised facts” about economic growth. Productivity growth is the key economic indicator of innovation. Although innovation contributes only a modest portion of growth, this is vital to long- term gains in the standards of living of a society. Economic growth can take place without innovation through replication of established technologies (Jorgenson et al, 2010). Without innovation but only replication, output will increase in proportion to capital and labour inputs.

By contrast the successful introduction of new products and new or altered processes, systems, and business models generates growth of output that exceeds the growth of capital and labour inputs (Jorgenson et al, 2014). This results in growth in productivity or output per unit of input. Innovation can be considered far more challenging than replication of already existing technologies. The diffusion of successful innovation requires mammoth financial commitments.

Another explanation can be that it is not always easy for inventors to appropriate the benefits of their innovations. Products or technologies which are easy to replicate once invented might explain the modest portion of economic growth contributed by innovations.

1.1 Information Technology

During the last century the information technology industry has shown remarkable growth. More advanced technological capabilities and usage due to greater global interconnectivity make it possible for companies to overcome difficulties in transferring information across international borders (Todd & Javalgi, 2007).

Castells (1999) even argues that the information and communication technology is the essential tool for economic development and material well-being in our age; it conditions power, knowledge and creativity; it is, for the time being, unevenly distributed within countries and between countries; and it requires, for the full realization of its developmental value, an inter- related system of flexible organizations and information-oriented institutions.

With the establishment of the Internet, geographical boundaries have become blurry. With

new information and communication technology, the network is, at the same time, centralized and

(13)

12 decentralized. It can be co-ordinated without a centre. Instead of instructions, we have interactions.

Much higher levels of complexity can be handled without major disruption (Castells, 1999).

It is often argued that information technology (IT) is a “special” technology in the sense that it affects a multitude of sectors and economic activities, and most importantly makes other sectors more productive (Kretschmer, 2012). Regarding the long-term impact of IT on a society there are different views. One can argue that the development of information technology (IT) is one of series of positive, but temporary shocks. The competing perspective is that IT produces a fundamental change in the economy leading to a permanent improvement in growth perspectives. Kretschmer argues that a narrow definition of IT investment would not capture the true impact of IT on the economy. IT is often considered a general purpose technology (GPT). The term GPT, has seen extensive use in recent treatments of the role of technology in economic growth, and is usually reserved for changes that transform both household life and the ways in which firms conduct business. Steam, electricity, internal combustion, and information technology (IT) are often classified as GPTs for this reason. They affect the whole economy (Jovanovic and Rousseau, 2005).

As with IT, 3DP is already a proven ‘general purpose’ technology that is being used for an enormous range of applications, such as fabricating spare and new parts for planes, trains and automobiles and thousands of items in between. It has huge environmental benefits, including substantial reduction in resources consumed in production, manufacturing products only on demand, and ‘just in time production’ of goods at or near where they are consumed. 3DP is likely to dramatically change business models, shift production location, shrink supply chains, and alter the global economic order, potentially degrading the importance of the Asian export manufacturing platforms and revitalizing the US innovation engine and the US economy (Garrett, 2014).

1.2 Technological Change

Technological change is critically important to firms for several reasons. First, it has the potential

to obsolete assets, labor, and intellectual capital of incumbents in the market. Second, it can create

entirely new markets, with new products, new customers, and exploding demand. Third,

technological evolution enables firms to target new segments within a market with improved

products (Sood and Tellis, 2010).

(14)

13 The drivers of technological change has been a topic of intense research and debate in the strategy literature. An early attempt to understand this phenomenon was by Foster (1986). He posited the theory of S-curves, which suggested that technologies evolve along successive S- curves; incumbents fail if they miss to switch to a new technology that passes the incumbent’s technology in performance. Other researchers built on the theory of punctuated equilibrium (Gould and Eldredge 1977) to propose a demand-side explanation for the phenomenon of disruption (Levinthal 1998, Adner 2002, Adner and Zemsky 2003, Mokyr 1990). They suggested that disruption occurs when a new technology that starts in one domain moves to a new domain with potentially higher demand and additional resources. Christensen (1997) proposed the theory of disruptive innovations.

1.3 Disruptive Innovation

“A disruptive innovation (DI) is an innovation that transforms the complicated, expensive services and products into things that are so simple and affordable that you and I can use them” (Christensen, 2002).

The term disruptive technology (DT) was first introduced by Bower and Christensen in their seminal journal article (Bower & Christensen, 1995), it posited that disruption occurred when an initially inferior technology introduced by a new entrant improved to meet the needs of the mass market. DT has been further developed in Christensen’s book The Innovator’s Dilemma (Christensen, 1997). The general terminology was changed to disruptive innovation (DI) in Christensen’s second book, co – authored with Michael E. Raynor, The Innovator’s Solution (Christensen & Raynor, 2003). Often DT and DI are used synonymously in the literature.

Additionally, Christensen found that people were misinterpreting the concept of DT and

were using it incorrectly, so the change was also to facilitate clarity of understanding (McDougall,

2014). The importance of DI theory is well recognised in the extant literature, with broad coverage

in business publications and numerous citations in a wide range of subject areas, such as marketing,

strategy and technology management (Danneels, 2004, p. 246). The theory of DI builds upon a

foundation of technological innovation studies presented in Figure 1 (Yu and Hang, 2009).

(15)

14 Figure 1. Timeline of DI theory

The meaning in Christensen’s work is simplicity and affordability. According to Christensen, most of the times at the beginning of an industry, the services or the products that are available are so complicated and expensive that the only people who can participate are people with a lot of money (Richardson, 2010).

A DI is an innovation that presents a novel and exceptionally different value proposition

than what is available in the current market. Products that are considered to be DIs generally

underperform in comparison to established products in mainstream markets and are commonly

stripped down, smaller, more convenient, have less features and cost less money (Christensen,

1997). While they lack many of the features that high value, mainstream customers desire, they

often appeal to fringe (and often new) customer groups who are not being satisfied by the

mainstream market’s offerings. The characteristics of disruptive businesses, at least in their initial

stages, can include: lower gross margins, smaller target markets, and simpler products and services

that may not appear as attractive as existing solutions when compared against traditional

performance metrics.

(16)

15 DIs improve products and services in ways that the market does not anticipate (Grant, 2010) and over time, through continual improvements and refinement, they often result in the removal of entrenched industry incumbents causing a disruption of the established mainstream market. In order to accomplish these levels of implementability one should see “a disruptive innovation” rather as a process than as a procedure. This process starts with products or services which begin as a simple application and then move up market, eventually displacing established competitors (Robles, 2015).

1.4 Dynamics of Competition

For simplicity of exposition, I would like to address the theoretical foundation of the theory of DI a bit more. Assume the market has two technologies (dominant and new), two dimensions (primary and secondary), and two segments: a mainstream and a niche. Figure 2 illustrates the dynamics of competition between the dominant technology and the new technology on the primary and secondary dimensions in one market.

Figure 2. Theory of Disruptive Innovations

Both segments have similar needs but differ in their preferences: the mainstream segment favors

the primary dimension, whereas the niche segment favors the secondary dimension, as shown by

(17)

16 their locations in Figure 2. However, both dimensions are both objective and vector—i.e., more is better. At time t1, the dominant technology is strong on the primary dimension but weak on the secondary dimension, whereas the reverse holds for the new technology. Given this preference distribution, at time t1, the mainstream segment prefers the dominant technology, whereas the niche segment prefers the new technology.

Both technologies improve on the primary dimension over time. At time t2, the dominant technology exceeds the needs of the mainstream segment on this dimension. However, the new technology improves sufficiently on the primary dimension so as to appeal to the mainstream segment, because it now meets its needs on both the primary and secondary dimensions. Thus, at time t2, demand of both segments shifts from the dominant technology to the new technology.

Christensen refers to this event as disruption. The niche segment plays the role of providing a demand for the new technology while it improves in performance on the primary dimension and meets the needs of the mainstream segment. Note that for this analysis, it is sufficient to assume segments with fixed preferences, as does Christensen (1997), so long as technologies improve over time.

1.5 Criticism

However, as with all theories, researchers have pointed out some weaknesses in the theory. First, researchers claim that the central thesis about a DI causing disruption appears to be tautological (Cohan 2000, Danneels 2004, Markides 2006) in the sense that Christensen’s writings alone suggest that the term could take on different meanings (Tellis 2006). The major issue is the use of the same term to describe both the causative agent (DI) and the effect (disruption). Second, the theory is ambiguous as to which domain of disruption the theory applies (Danneels 2004, Markides 2006).

However, despite the importance of disruptive innovations, there has been relatively little academic research on the characteristics (Danneels, 2004). The dearth of such research may be because there is neither an appropriate measure for the disruptiveness of innovations. Nor has prior research assessed the discriminant and convergent validity of the disruptiveness

characteristic relative to two other well-known innovation characteristics. One of these is

radicalness, which is technology based—that is, the extent to which an innovation advances the

(18)

17 performance frontier faster than the existing technological trajectory (Gatignon et al., 2002). The other characteristic is competency based—that is, the extent to which innovations build upon and reinforce, rather than destroy, existing competencies (Tushman and Anderson, 1986).

Christensen counters this point by arguing that rarely does an early researcher manage to describe a novel concept in an unambiguous way and that through a process of evolution, later researchers refine and develop the concept and remove the ambiguity (Christensen, 2006).

Govindarajan & Kopalle (2006) provide a more holistic view that accounts for the multi-faceted nature of the DI construct.

“A disruptive innovation introduces a different set of features, performance, and price attributes relative to the existing product, an unattractive combination for mainstream customers at the time of product introduction because of inferior performance on the attributes these customers value and / or a high price – although a different customer segment may value the new attributes.

Subsequent developments over time, however, raise the new product’s attributes to a level sufficient to satisfy mainstream customers, thus attracting more of the mainstream market.”

(Govindarajan & Kopalle, 2006).

Sood and Tellis (2011) even claim that Christensen’s theory of DI suffers from circular

definitions, inadequate empirical evidence, and lack of a predictive model. Some shortcomings which Sood and Tellis touch upon are that in many points in time, competing technologies coexist. In some cases, continue to survive and coexist with the new technology by finding a niche. Sood and Tellis noted, the literature often conflicts as to the actual definition of a

disruptive innovation and that clarification on its attributes was required to prevent ambiguity in future research.

It is true that some technologies do die, but many continue to survive even after being

disrupted. Second, some technologies experience disruption in one domain but not in another

domain. Third, there is a fascinating dynamic of emergence of new secondary dimensions of

performance. A new technology almost always introduces a new dimension of importance even

while competing with old technologies on the primary dimension. These secondary dimensions

appeal to various niche segments. However, in all cases, the competition for the mainstream

segment was still on the primary dimension of performance, which continued to improve

substantially over time. Fourth and foremost, contrary to current belief, Sood and Tellis (2011)

(19)

18 observe multiple disruptions or crossings between paths of technological performance. This

pattern occurs when technology disruption by a new technology is not permanent, because a

technology that has been surpassed in performance regains technological leadership. Sood and

Tellis find a total of four cases of multiple technology disruptions: two in computer memory and

two in electrical lighting. Thus disruption is not permanent as extant theory suggests.

(20)

19

2. Technologies positioned by hype

Determining when to adopt an emerging technology is a critical decision. If an enterprise launches its efforts too soon, it will suffer unnecessarily through the painful and expensive lessons associated with deploying an immature technology. If it delays action for too long, it runs the even greater risk of being left behind by competitors that have succeeded in making the technology work to their advantage (Fenn, 1995).

The Gartner Hype Cycle (GHC) is named for the IT research and advisory firm, Gartner, Inc. The hype curve was introduced by Gartner in 1995, and is used to characterize a typical progression of an emerging technology to its eventual position in a market or a domain.

According to their Web site: Hype Cycles provide a graphic representation of the

maturity and adoption of technologies and applications, and how they are potentially relevant to solving real business problems and exploiting new opportunities. GHC provides a snapshot of the position of technologies along a predictable pattern of enthusiasm, disillusionment and eventual realism. It highlights technologies that are the focus of attention because of particularly high levels of hype, or those that may not be broadly acknowledged, but that Gartner believes have the potential for significant impact. Using consensus, Gartner analysts position technologies based on hype.

Although Gartner has become an icon to their clients, academics apparently have paid limited attention in their research to the company, with few exceptions, such as the Artificial Intelligence and the Electronic Transactions sections. Although Gartner’s research likely dominates the practice side of technology, it has received at most limited attention from academics (O’Leary, 2008).

2.1 From trigger to plateau

The GHC model adds another dimension to technology life cycle models: it characterizes the

typical progression of an emerging technology from user and media overenthusiasm through a

period of disillusionment to an eventual understanding of the technology's relevance and role in a

market or domain (Linden and Fenn, 2003). Different technologies do not move at the same speed

through the hype curve (O’Leary, 2008).

(21)

20 Jackie Fenn, the originator of the Hype Cycle and co-author of the 2008 book Mastering the Hype Cycle says the pattern, happens over and over and over—so much that you must wonder how capable companies, adopting highly touted innovations, so often fail to understand what is happening. Why do so many organizations seem to rush lemminglike to an innovation, only to abandon it when it falls short of initial expectations? (Lajoie, 2014). See Appendix (Figure 1) for a graphical explanation of the GHC model.

As with most innovation models initially the model starts with an ‘Innovation Trigger’, this is where a technology is conceptualized. Following on from this comes the ‘Peak of Inflated Expectations’ when the technology is implemented and we hear much about both successes and failures. Media hype and expectations can be huge at this point. Consequently, Gartner predicts a

‘Trough of Disillusionment’ will follow when flaws lead to disappointment. As the technology’s potential becomes more broadly understood and realistic we embark along the ‘Slope of Enlightenment’. The final stage is the ‘Plateau of Productivity’ when the technology becomes widely implemented and its applications within the marketplace are more stable. These stages are best depicted by the figure below.

Figure 3. Stages within the GHC

(22)

21 In a fast-paced business world, executives often feel that they are being forced to adopt new technologies and practices at an ever-increasing rate. However, fundamental advances in technology are still taking over a decade — sometimes up to 30 years or more — to traverse the Hype Cycle from initial prototypes to mainstream adoption.

2.2 3DP and the GHC

"First, determine the material, performance and quality requirements of the finished items first; second, determine the best 3D printing technology; and third, select the right 3D printer” ~ Peter Basiliere

3DP was first highlighted in Gartner research as a technology to watch in 2006; reached the ‘Peak of Inflated Expectations’ in 2012-2013; and will likely take another five to ten years to fully realize its disruptive potential in industries such as design, retail, manufacturing, supply chain and construction. According to Gartner, consumer adoption will be outpaced by business and medical applications that have more compelling use cases in the short term. Figures 2-4 in the Appendix clearly depict how 3DP, in particularly Consumer 3DP, moved over the peak of inflated expectations within the GHC.

During the last years, 3DP and its uses continue to evolve rapidly in response to hype, greater visibility and, more importantly, demand. It is a technology of great interest to the general media, with demonstrations on science shows, on gadget websites and in other areas. — the hype leads many people to think the technology is some years away when it is available now and is affordable to most organisations (Rivera and Goastuff, 2013).

As Peter Basiliere, research director at Gartner, puts it;“3D printing is a technology accelerating to mainstream adoption.” Naturally, some technologies are maturing faster than others and will be widely available in just a few years. In fact, some are already in general use. An example is 3D printing for prototyping, which has been the mainstay of the 3D printing industry since its inception. Basiliere sees 3DP as a tool for empowerment, already enabling life-changing parts and products to be built in struggling countries, helping rebuild crisis-hit areas and leading to the democratization of manufacturing.

Although it is true that at the early stage there are some similarities between the consumer

and enterprise market as organizations are beginning to employ "consumer" devices in order to

learn about 3D printing's potential benefits with minimal risk and capital investment.

(23)

22 Fundamentally, the two markets are driven by different uses and requirements and must be

evaluated separately. First, the enterprise 3D printing market is very different from the consumer

market. Second, even though some 3D printers are priced at a few hundred dollars, the price is too

high for mainstream consumers at this time, despite broad awareness of the technology and

considerable media interest. Third, 3D printing is not one technology but entails a rather broad

subset of different ones. The different technologies each have pros and cons, and printers work

with varying build sizes and materials. This means organizations must begin with the end products

in mind. Figure 4 and 5, on the next page, show a clarification of how all these different

technologies and future appliciations w.r.t. 3DP can be placed within the GHC.

(24)

23 Figure 4. Hype Cycle for 3D Printing, 2014

Figure 5. Subset of 3DP technologies/developments with respect to the GHC

Position within GHC Technologies

On the Rise  Intellectual Property Protection (3DP)

 Macro 3D Printing

 3D Bioprinting Systems

 Classroom 3D Printing

 3D Printing and Supply Chain

 3D Printing for Oil and Gas

 Retail 3D Printing

 Industrial 3D Printing

At the Peak  3D Printing of Medical Devices

 Consumer 3D Printing

 3D Printing in Manufacturing Operations Climbing the Slope  3D Print Creation Software

 Enterprise 3D Printing

 3D Printing Service Bureaus

 3D Scanners

Entering the Plateau  3D Printing for Prototyping

(25)

24

3. Adding rather than subtracting

The definition of 3D printing (3DP) is somewhat inexact. While some experts in the field would restrict 3DP to units with inkjet-based print heads that create an object on a layer-by-layer basis, others would apply this term to office or consumer versions of rapid prototyping machines that are relatively low-cost and easy to use (Casey, 2009). The word ‘rapid’ relates to the ease of making a copy of an object due to the simplicity of writing a computer program that controls the object’s shape. The term ‘prototyping’ refers to this process as being too slow for use in mass production (Berman, 2012). Essentially, what all of the processes and technologies have in common is the manner in which production is carried out - layer by layer - in an additive manufacturing (AM) process.

According to Berman (2012) there are two important aspects which distinguish 3DP from other rapid prototyping technologies. The first distinction is cost. The second major difference between these technologies is that 3D printers seamlessly integrate with computer-aided design (CAD) software and other digital files like magnetic resonance imaging.

3.1 Subtractive vs. additive

Manufacturing comes from the French word for “made by hand.” This etymological origin is no longer appropriate to describe the state of today’s modern manufacturing technologies (Campbell, 2011).

Before AM was possible, final products were dependent on the capabilities of the tools used in the subtractive manufacturing processes. One should think of casting, forming, molding, and machining. This seemingly small distinction—adding rather than subtracting—means everything.

3DP, as a new form of manufacturing, will cause various changes to social balance and, consequently, to the way we perceive space (Pierrakakis, 2014). Some researchers even argue that traditional manufacturers will face a production paradigm as the digital age forces companies to adopt to these new technologies.

The beauty of this technology is that it does not need a factory to be deployed. Small items can be made by a machine like a desktop printer in the corner of an office, a shop, or even a house.

Bigger items (bicycle frames, panels for cars, aircraft parts, etc.) need a larger machine and a bit

more space. Therefore, industrial space may no longer exist in the currently known form. The

producer-consumer model is threatened the same way the client-server model was dethroned by

(26)

25 peer-to-peer protocols. One way or another, the reality of this technology’s potential to disrupt traditional design and manufacturing processes becomes more and more realistic.

3.2 Minimum efficient scale

AM promises to reduce—more so over time—the minimum efficient scale that gave rise to large modern industrial production facilities, lowering barriers to entry into manufacturing. The relationship between capital and scale is captured using the concept of minimum efficient scale—

the point at which the average cost of each unit of production is minimized. Where minimum efficient scale is high (i.e., where there are large capital costs required to initiate production) the number of production facilities will be small.

Figure 6. Additive vs. conventional

AM impacts the economics of production by reducing minimum efficient scale. In some cases, AM may allow consumers to satisfy their individual needs without the significant labour or capital investments that might have previously been required. Research supports this conclusion.

Multiple economic studies illustrate that minimum efficient scale for AM can be achieved at low

unit volumes—as low as one. This cost performance contrasts with that of traditional

(27)

26 manufacturing methods that face higher initial costs for tooling and setup (e.g., Allen et al. 2006;

Ruffo et al. 2006).

3.3 Geometrical Freedom

As 3D printing is an enabling technology that encourages and drives innovation with unprecedented design freedom, while being a tool-less process that reduces prohibitive costs and lead times. Components can be designed specifically to avoid assembly requirements, with intricate geometry and complex features created at no extra cost.

Naturally, every object produced in a 3D printer will not be the result of the individual’s own creativity and ingenuity. Sometimes the object will be one downloaded and printed from another person’s original design. In most cases the object will simply be a copy of an existing commercial product. This copy could come from at least two sources. The first source would be the Internet. CAD files are easily copied and distributed online. Once an individual creates the plan for an object and uploads that plan, it is essentially available to the world. Designs, not products, would move around the world as digital files to be printed anywhere by any printer that can meet the design parameters. The second source would be a 3D scanner. A 3D scanner has the capability to create a CAD file by scanning a 3D object. An individual with a 3D scanner would be able to scan a physical object, transfer the resulting file to a 3D printer, and reproduce it (See Appendix, Figure 5).

3.4 Controversies

AM encapsulates several controversies and raises the stakes even on ontological issues such as democracy, self-realization in the work place, the borders between the private and public spheres, the ratio between security and freedom and so on. With consumer-level availability of 3DP potential controversies become even more profound. As a consumer can now create a design and then make that design available to the whole world in the form of a digital file. It then becomes easy for anyone to replicate the creator’s work, either by sending the digital file to a third party or by printing the design on personal hardware.

Some US States have already started the process of regulating the 3DP of guns, while some

of the pioneers in 3DP dictate the need for regulation in raw material (namely gunpowder) instead

of printing designs or results of 3D printing per se. Another parameter, slightly harder to detect, is

(28)

27 the repudiation of liability; namely when a 3D printed object causes injury to an individual, who is it to blame, the design, the 3D printer, the raw material, the manufacturer of the printer or the owner? What would be the results of a faulty but highly utilized design? (Doherty, 2011). Besides safety issues copyright infringement seems to be almost impossible to avoid, as a substantial number of publicly available designs are decorations, games, or pop culture references.

3.5 Trajectory

"Consumer 3D printing is around five to 10 years away from mainstream adoption," Peter Basiliere, research vice president at Gartner.

Within the report: “As Seen On TV: 3D Printing Ready For Primetime”, published by investment bank Citi, the adoption of earlier popular household electronics in Japan is taken as a proxy for the trajectory of 3DP (See Appendix, Figure 6). The typical time to achieve 5% penetration of households was 3-5 years, with the 10% milestone taking an additional 2-4 years and full penetration requiring about 20-25 years.

With respect to adoption of Consumer 3DP, Citi analysts consider three scenarios. The most positive scenario being near 100% adoption by households over time much like refrigerators, TVs, microwave, etc. The trajectory of these products was typically much steeper with nearly 5%

penetration of households in fewer than 3 years and 20%+ adoption within 10 years. Probably, the adoption will be closer to devices such as the fax machine and video camcorders which had a specific value proposition to a large demographic but not the entire population. Adoption for these types of products typically inched up over time hitting ~10% after ten years, eventually reaching

~60-70% of households over the product life.

A McKinsey Global Institute study

⁴

goes on to predict that by 2025, up to 10 percent of all consumer products could be 3D printable. Cohen, Sargeant & Somers from McKinsey &

Company, attribute these numbers to the advancements in 3D printing technology, stating “these advances have brought the technology to a tipping point – it appears ready to emerge from its niche

4 McKinsey Global Institute (2014), 3-D printing takes shape http://www.mckinsey.com/insights/manufacturing/3- d_printing_takes_shape (Accessed: 1st of May 2015)

(29)

28 status and become a viable alternative to conventional manufacturing processes in an increasing number of applications. According to analysts of Wells Fargo

⁵

, 3-D printing nowadays has achieved approximately eight percent of its global market potential.

A worst-case scenario shows how adoption would look if 3D printers turn out to be a passing fad like mini disk and HD DVD players. These types of products usually showed modest adoption in early years before eventually petering off and showing only limited production runs to niche users.

3.6 Market size

Forecasts for growth of the AM market by research analysts range

⁶

from $7 billion by 2020 (Paul Coster of JP Morgan) to bull market scenarios as high as $21.3 billion by 2020 (Ben Uglow of Morgan Stanley). The global additive manufacturing market, reached sales of $3.0 billion in 2013, on annualized growth of 35 percent over sales of $2.3 billion in 2012. AM industry growth over the last 25 years has been 25.4 percent, and 29 percent in the last three years (Cotteleer & Joyce, 2014). Figure 7, in the Appendix, shows the growth of the market.

3.7 Computer Aided Design software

As you cannot 3D print something without the 3D design file – this is where Computer Aided Design (CAD) software comes in. As CAD is one part of the whole Digital Product Development activity within the Product Lifecycle Management processes, the market of CAD-software is highly integrated with 3DP. Jon Peddie Research

⁷

, a market watcher that tracks the design software market, estimated the CAD software market to be an $8 billion market, in 2014, with 5.15 million annual users. The market share breakdown is shown in the Appendix (Figure 8). Autodesk leads

5 Forbes (2014) Roundup of 3D Printing Market Forecasts and Estimates, 2014

http://www.forbes.com/sites/louiscolumbus/2014/08/09/roundup-of-3d-printing-market-forecasts-and-estimates- 2014/ (Accessed: 1st of May 2015)

6 Social Dashboard (2015) 2015 Roundup of 3D Printing Market Forecasts and Estimates https://socialdashboard.com/news/2015-roundup-of-3d-printing-market-forecasts-and-estimates (Accessed: 1st of May 2015)

7 Jon Peddie Research (2015) The CAD market returns to growth; reaches $8 billion in 2014 http://jonpeddie.com/publications/cad_report/ (Accessed: 1st of May 2015)

(30)

29 the pack at 29%. The rest of the pie is divided by Dassault Systemes (22%), Siemens PLM Software (11%), and PTC (9%).

Another estimate which keeps track of all the different CAD software is the G2 Crowd rate

⁸

. The G2 Crowd rate represents the democratic voice of real software users, rather than the subjective opinion of one analyst. The data used for their estimations is algorithmically based on data sourced from product reviews shared by G2 Crowd users and data aggregated from online sources and social networks. A more defined explanation can be found at the website of G2 Crowd.

Both surveys clearly show that Autodesk is being an industry leader within this sector, closely followed by Dassault Systemes (See Appendix, Figure 8 & 9)

8 G2 Crowd Compare Best CAD Software https://www.g2crowd.com/categories/cad (Accessed: 1st of May 2015)

(31)

30

4. Hypotheses

The three hypotheses being tested, within the quantitative part of this paper, are possible to test due to the availability of Google Trends

⁹

data. The hypotheses aim to find relations between stock price movements of 3DP stocks and specific search terms conducted online. In this Chapter I will first explain why Google Trends data can be seen as highly reliable. Second, recent

academic insights derived by the use of Google Trends data will be covered. Third, I will provide reasons of thought and foundations to each of the three hypotheses being tested.

Internet users commonly use a search engine to collect information, and Google continues to be the favorite. As of today, May 2015, Google accounts for 68.5% of all desktop search queries performed in the world

¹⁰

. The search volume reported by Google is thus likely to be representative of the internet search behavior of the general population. Second, and more critically, search is a revealed attention measure: if you search for a stock in Google, you are undoubtedly paying attention to it. Therefore, aggregate search frequency in Google is a direct and unambiguous measure of attention (Da et al., 2011).

At their core, financial trading data sets reflect the myriad of decisions taken by market participants. Preis, Moat & Stanley (2013) suggest that massive new data sources resulting from human interaction with the Internet may offer a new perspective on the behavior of market participants in periods of large market movements. Choi and Varian (2012) have shown that data from Google Trends can be linked to current values of various economic indicators, including automobile sales, unemployment claims, travel destination planning and consumer confidence.

Ginsberg et al. (2009) similarly find that search data for 45 terms related to influenza predicted flu outbreaks 1 to 2 weeks before Centers for Disease Control and Prevention reports. The authors conclude that, “harnessing the collective intelligence of millions of users, Google web

9 Google Trends https://www.google.nl/trends/

10 Net Marketshare, Desktop Search Engine Market Share

https://www.netmarketshare.com/search-engine-market-share.aspx?qprid=4&qpcustomd=0 (Accessed: 1st of June 2015)

(32)

31 search logs can provide one of the most timely, broad-reaching influenza monitoring systems available today” (p. 1014).

The research of Preis et al. goes deeper in a sense that Google Trends data does not only reflect aspects of the current state of the economy, but may also provide some insight into future trends in the behavior of economic actors. By analyzing changes in Google query volumes for search terms related to finance, Preis et al. find patterns that may be interpreted as ‘‘early warning signs’’ of stock market moves. Using historic data from the period between January 2004 and February 2011, the researchers detect increases in Google search volumes for keywords relating to financial markets before stock market falls. Their findings are consistent with the intriguing proposal that notable drops in the financial market are preceded by periods of investor concern. In such periods, investors may search for more information about the market, before eventually deciding to buy or sell. Their results suggest that these warning signs in search volume data could have been exploited in the construction of highly profitable trading strategies.

4.1 Hypothesis 1

The first hypothesis chronicles around the thought that increased search volumes should be reflected within the stock prices of these companies. The rather logical assumption being made here is that if the amount of search queries increases, so would the buying intensions for purchasing the stock, the stock therefore would subsequently rise in value. As an example, if the popularity of searches for “3D-Systems” increases over time it is expected that the stock price of 3D-Systems moves accordingly with this increased popularity of search queries.

As a proxy for popularity of the specific company, Google Trends data for the “company name”, ‘3D Systems’ in this case, is used. A positive relation between the volume of “company name” search queries and stock price movements of this same company is assumed here. This leads to the following hypothesis:

H1: A positive relation is expected between the stock price movements of a company and the

volume of online search queries of this same company over the same period of time.

(33)

32 4.2 Hypothesis 2

As all companies of our sample, see Chapter 5, operate within the same sector, respectively 3D Printing, the same positive relation is assumed between the stock price movements of these companies and the amount of search queries which best describe this sector, in this case “3D Printing”.

The assumption being made here is that when a positive trend evolves around a sector, this would be reflected in stock price movements of companies operating within this sector. The second hypothesis therefore becomes:

H2: A positive relation is expected between the stock price movements of a company and the volume of online search queries of the “trend”, over the same period in time, where the search term “trend” best depicts the sector in which the company operates.

4.3 Hypothesis 3

The third hypothesis looks into how these search queries, “company name” and “the trend” are associated to each other. A positive correlation between the search volume of the “trend”, and the search volume for the individual company, “company name”, over the same period of time, is expected here.

As an example, if people are becoming more interested in “3D Printing”, it is assumed that the popularity of “3D Systems” would subsequently be on the rise as well. The third and final hypothesis therefore is:

H3: “Company name” search queries are positively correlated to “trend” search queries, where the search term “trend” best depicts the sector in which the company operates.

Besides testing these three hypotheses, emphasis within the Results section, Chapter 6, has been

put on depicting trends and explanations, w.r.t. revenue and income streams, of the individual

company and for the sample as a whole.

(34)

33

5. Data

As 3DP received wide media coverage last couple of years and according to Gartner analysts reached the ‘Peak of Inflated Expectations’ within the GHC within 2012-2013, a preliminary attempt of testing whether a relation exists between online search popularity and stock price movements of 3DP companies is conducted.

The efficient markets theory (EMT) of financial economics states that the price of an asset reflects all relevant information that is available about the intrinsic value of the asset. It is generally believed that securities markets are extremely efficient in reflecting information about individual stocks and about the stock market as a whole. As a financial security represents a claim on future cash flows, and thus the intrinsic value is the present value of the cash flows the owner of the security expects to receive, stocks therefore basically reflect the public’s thoughts about potential profits of the stock. As a response variable the percentual differences from quarter-to-quarter of 3DP stocks is chosen.

The EMT is closely associated with the idea of a “random walk,” which is a term loosely used in the finance literature to characterize a price series where all subsequent price changes represent random departures from previous prices. It is rather interesting to see if we can find some explanations for the movements of 3DP stocks over the last five years or that statistically has to be concluded that we cannot explain any of the movements of these 3DP stocks, and a “random walk”

explanation is more likely.

5.1 Selection Criteria

The selection criteria for the sample are based on the The STOXX Global 3D Printing Pureplay index. The STOXX Global 3D Printing Pureplay index aims to select the top 30 companies highly related to the 3D Printing sector (>10% of revenues) Their criteria for selecting a fund are:

- More than ten percent of revenue must be generated from the 3-D printing sector - Minimum three-month average daily trading volume of 250,000 Euros ($333,675) - Minimum free-float market cap of 80 million Euros (about $107 million)

- Company must be listed in a country that is classified as a developed market according to STOXX’s country classification model

A list of the sample, companies which fulfilled these criteria, can be found below, Figure 7.

(35)

34 Figure 7. Companies within the sample

Figures in USD Ticker CN Search term Data?

Market Cap.

(May ‘15) Revenue ‘14 Net income ‘14 Dassault Systemes SA (DASTY) Dassault Systemes 2010 18.1 bln. 2.3 bln. 291 mln.

Autodesk Inc. (ADSK) Autodesk Inc 2010 13.11 bln. 2.5 bln. 82 mln.

ANSYS (ANSS) Ansys 2010 8.1 bln. 936 mln. 255 mln.

PTC Inc. (PTC) PTC Inc 2010 4.7 bln. 1.4 bln. 151 mln.

3D Systems Corporation (DDD) 3D Systems 2010 2.5 bln. 653 mln. 12 mln.

Stratasys Ltd. (SSYS) Stratasys 2010 1.8 bln. 726 mln. (124) mln.

Proto Labs Inc. (PRLB) Proto Labs Mar-12 1.8 bln. 210 mln. 41.6 mln.

Faro Technologies Inc. (FARO) Faro Technologies 2010 729.9 mln 342 mln. 33.7 mln.

Voxeljet AG (VJET) Voxeljet Oct-13 147.7 mln. 16 mln. (4.3) mln.

Cimatron Ltd. (CIMT) Cimatron Ltd. 2010 94.4 mln. 34 mln. 2.9 mln.

ExOne Company

(XONE) ExOne Feb-13 178.5 mln. 44 mln. (21.8) mln.

5.2 The Model

The regression model being tested has as a dependent variable the percentual movements with which the company’s stock increased or subsequently decreased between each quarter. The first two independent variables are conducted by use of the Google Trends website. The latter two independent variables are taken out of quarterly and annually reports, found on the websites of the companies of our sample. A time frame of five years is chosen (2010 – 2015), meaning that t=1 would imply the quarterly differences of the first 3 months of 2010.

S : β0 + β1 (CN) + β2 (DDD) + β3 (R) + β4 (NI) + ε

S = stock price of the specific company

CN = Company name of the specific company searched within Google Trends*

DDD = ‘3D Printing’ searched within Google Trends*

R = Revenue of the specific company NI = Net income of the specific company

*The numbers on the graphs on the Google Trends website reflect how many searches have been done for a particular term, relative to the total number of searches done on Google over time. They do not represent absolute search volume numbers, because the data is normalized and presented on a scale from 0-100. A downward trending line means that a search term's popularity is decreasing. It doesn't mean that the absolute, or total, number of searches for that term is decreasing. Each point on the graph is divided by the highest point and multiplied by 100. When we (Google Inc.) don't have enough data, 0 is shown.

(36)

35 As a quick reminder, the hypotheses being tested are;

H1: β1 is positive H2: β2 is positive

H3: (CN) and (DDD) are positively correlated to each other.

The three hypotheses focus mainly on the relation of stock price movements w.r.t search queries, as increased revenue and profitability often can be seen as good estimators of stock price movements, it has to be stated though that in general a positive relation between all beta’s w.r.t.

the regressand is expected.

For all variables, w.r.t. the regression, quarterly percentual differences are taken. All variables are standardized to 1. Meaning that a 0.07 positive difference from period t=1 to t=2 would mean that during these 3 months the underlying increased by 7 percent.

By reflecting how popular a trend or specific company is over time we know a lot more about the popularity of the search term. The specific search terms for each individual company can be found in the third column of Figure 7. As a broad assumption there is chosen for adding these variables while one could reason that as a company or trend is more searched for one could expect to see this increased popularity being reflected in the stock prices of these companies.

Variable (DDD) has been added as a proxy for the general popularity of 3DP. The variable (CN) has been added to see this popularity on a company level.

5.3 Data collection

Data regarding the first two variables is collected from the Google Trends website. Data regarding

stock prices from the Yahoo Finance website, where the “Adj. Close” stock price is used in order

to correct for the stock splits and or dividend pay-outs, which happened by some companies, over

the given period. Data with respect to the variables revenue and net income is taken from the

company’s quarterly and annually reports. As a guideline for consistency, all figures are taken

from the Consolidated Statement of Operations of the specific company.

(37)

36

6. Results

First we dive deeper into the revenues and net incomes of the eleven companies of the sample. A detailed list of the revenue and income streams can be found in the Appendix, Figure 10 and 11.

Figure 8 below shows how the revenue streams evolve quarter-to-quarter over the last five years. The revenues and incomes of the bigger companies of our dataset are more stable over the years. As expected Dassault and Autodesk account for a majority of all revenues. The companies of our sample made 9.1 bln. USD within 2014, where Dassault and Autodesk together accounted for 4.8 bln. USD of these revenues.

Figure 8. Quarterly revenue streams

Although these two companies dominate the market, medium-sized companies such as 3D Systems and Stratasys showed enormous growth in their revenues last couple of four years. From 2010 up to 2014, 3D Systems managed to triple their revenues, from 159.8 mln. USD up to 653.5 mln. USD over the year 2014. Stratasys even topped this by generating 117.2 mln. USD over 2010, managing a revenue stream of 725.7 mln. USD within 2014. These growth figures are unseen by the major, more mature, players within this market. Medium-sized companies within our sample show rapid growth, but the smaller companies, such as Cimatron, Exone and Voxeljet, struggle heavily to grow their revenues (Appendix, Figure 12 and 13). Figure 9 shows how the income

-400 100 600 1100 1600 2100 2600

Revenue per quarter (mln. USD)

Time (2010 - 2015)

Quarterly revenue streams

DASTY ADSK ANSS PTC DDD SSYS PRLB FARO VJET CIMT XONE

(38)

37 streams evolved last couple of years. What speaks to mind is the volatility for the sample as a whole. For 2010 – 2012 we see a steady upward trend in profitability, while showing a major drop in attaining this growth for 2013. Over the year 2013, the profitability seems to increase again, showing a major drop, and thereby a similar pattern, as for the year of 2014.

Figure 9. Quarterly net income streams

This makes us to believe that the sector, the sample, behaves rather cyclical. Periods of rather rapid profitability growth are followed by a significant set-back in profits, in order to maintain and seek their former growth path again. Of course these trends can be partly due to strategic choices of the management of these companies, but still the rather cyclical characteristics of profitability do show us signs of how aggressive the battle for market share and profits is within this market. To see how revenue and income streams of individual companies performed on a yearly basis see Figure 12 &

13 within the Appendix.

6.1 Panel data

Panel data models examine group (individual-specific) effects, time effects, or both. These effects are either fixed effect or random effect. A fixed effect model (FE) examines if intercepts vary across groups or time periods, whereas a random effect model (RE) explores differences in error

-130 -80 -30 20 70 120 170 220 270 320

Net income per quarter (mln. USD)

Time (2010 - 2015)

Quarterly net income streams

DASTY ADSK ANSS PTC DDD SSYS PRLB FARO VJET CIMT XONE

(39)

38 variances. A one-way model includes only one set of dummy variables (e.g., firm), while a twoway- model considers two sets of dummy variables (e.g., firm and year). A one-way model is being used here, as we have no time-invariant variables, as every variable behaves rather dynamic over time.

As recent literature heavily discusses the disadvantages (Bell & Jones, 2015) of each of the longitudinal models, FE and RE, chosen is to compare different types of models for our data, respectively pooled OLS, FE and RE are being compared. The descriptive statistics of our panel dataset can be found in the Appendix, Figure 15. Figure 10 below shows the different models.

Chosen is to run robust models, in order to account for heteroskedasticity within our variables.

Note as dependent variable here is chosen the stock price movement of the eleven 3DP companies within our sample, these returns are normal stock returns.

Figure 10. Robust hierarchical models, Pooled OLS, FE and RE

(1) (2) (3)

VARIABLES OLS Fixed Effects Random Effects

CN 0.0176 0.0255 0.0176

(0.0236) (0.0195) (0.0214)

DDD 0.121** 0.0999* 0.121**

(0.0503) (0.0529) (0.0530)

R 0.176 0.238 0.176

(0.180) (0.166) (0.158)

NI 6.95e-05 -0.00204 6.95e-05

(0.00167) (0.00146) (0.00135)

Constant 0.0307 0.0315* 0.0307***

(0.0201) (0.0162) (0.00638)

Observations 189 189 189

Prob. > F Prob. > chi2 R-squared

0.1507 0.032

0.1099 0.040

0.0728

Number of INC 11 11

Company FE YES YES

Robust standard errors in parentheses

* p<0.01, p<0.05, * p<0.1

3D Printing: Media Hype or Disruptive Innovation?