Technological Change and the Growing Importance of Patents for Telecom Operators∗

Looking for Trouble? Technological Change and the Growing Importance of Patents for Telecom Operators∗

Andrea Fosfuri† Christian Helmers‡ David Wehrheim§

October 9, 2014

ABSTRACT

We analyze the effect of a fundamental structural technological change in the telecommunication industry on telecom operators. The technological change pushed operators away from basic analog voice transmission to complex digital data transmission technology. We show that this recent change confronts operators with an enormous challenge because technologies for digital data are notorious for their complex patent landscape characterized by standard essential patents, patent thickets, litigation, and cross-licensing. We show that operators have fundamentally changed their patenting behavior, increasing filings on technologies for digital data almost fourfold over the past decade. Our results indicate that operators were most likely to patent in technologies most severely affected by the complex- ities of the patent landscape. While operators were so far on the sidelines on the global patent wars, our results suggest that they might soon see themselves embroiled in patent related disputes.

KEYWORDS: Telecommunication, operators, patents, digital data transmission

∗We thank Tim Pohlman and Stefan Wagner for providing data. The authors also thank representatives at Telef´onicafor insightful discussions. †Universita Bocconi; e-mail: [email protected] ‡Santa Clara University; e-mail: [email protected] §Universidad Carlos III de Madrid; e-mail: [email protected] 1 Introduction

How do innovative industries move from an equilibrium in which patents play no or only a minor role to an equilibrium in which patents are a crucial element that determines a company’s ability to compete in the product market? Boldrin and Levin (2013) for example suggest that as industries mature, incumbents switch to using and enforcing patents to protect their market shares by extracting rents from (potential) entrants. Hall and Ziedonis (2001) document the onset of the so-called ‘Patent Portfolio Races’ in the semiconductor industry during the 1990s. The dramatic increase in patent filings within the industry reflected an increased perception of the strategic value of patent rights. This shift was remarkable because the semiconductor industry had largely operated without recourse to patent protection until then (Cohen et al., 2000). Jell and Henkel (2014) document a similar sharp increase in strategic patenting in the German newspaper printing machine industry after the year 2000. The shift in patenting behavior was triggered by a single company’s move towards more aggressive patenting, similar to the U.S. semiconductor industry where a single firm (Texas Instruments) is credited with triggering the patent portfolio races. In this paper, we provide empirical evidence for a different channel through which an industry can move from an equilibrium in which patents play no or only a minor role to an equilibrium where strategic patenting is a crucial element in firms’ ability to compete in the market: technological change. We look at telecommunication operators for which historically patents have played no or only a minor role. We show how the fundamental technological change from analog voice transmission to digital data transmission technology has been pushing telecom operators into patent races. Digitization has fundamentally transformed the information and communication technology (ICT) industry more broadly. The technological change has been accompanied by a dramatic increase in the use of patents and escalating patent enforcement which have resulted in outright ‘Patent Wars’ (Financial Times, October 17, 2011). Several interrelated factors explain this shift towards patenting in ICT: First, there has been a global patent explosion (Kortum and Lerner, 1999; Fink et al., 2013), perhaps most notably in China (Eberhardt et al., 2011). The number of patents in ICT granted by the USPTO has nearly doubled between 2000 and 2006 (from nearly 234,000 to 431,000).1 Second, there is an overall increase in the level of technological complexity and the convergence of different technological domains. The increased complexity and combination of different types of technologies result in a high degree of interdependency between different components of a given technology. This complexity has led to overlapping claims across patents which can result in patent thickets if patent ownership is diffuse (Shapiro, 2001). The more complex technologies are, the more prevalent are such thickets (von Graevenitz et al., 2011). Third, the need for interoperability and interconnectivity of different technologies and devices to combine them into single products and services creates demand for

1Granted patents are obtained from the latest version of the NBER patent database. We retrieved ICT related patents by extracting forty-four patent technology classes identiﬁed by Ozcan and Greenstein (2013).

2 technological standards that promote compatibility. While such standards, in prin- ciple, allow to link different products and technologies by different producers, such standards also involve patent rights, providing additional incentives to patent strate- gically and define standards accordingly. Similarly, firms seek to establish alternative mechanisms to structure the patent jungle. Patent pools and cross-licensing agreements are popular solutions, but they have the obvious drawback that in order to participate, firms need patents. The whole array of these complex issues – the fast increase in the size of companies’ patent portfolios, technological complexity and interdependencies, patent thickets, standard relevant patents, patent pools and cross-licensing agreements, IP litigation, and trolling non-practicing entities (NPEs) – has affected perhaps most prominently the market for smartphones and tablet computers. Telecommunication operators have been largely spared so far. However, there are signs that this might change soon.2 Telecom operators’ revenues from their traditional core services have started to dry up; standard business models based on revenue streams from traditional voice telephony and SMS have given way to alternative, internet-based services largely circumventing operators. While operators have been forced to compete through lower prices and more bandwidth for a given price, they saw other companies take advantage of their infrastructure to offer new digital products and services including smartphones, social networks, cloud services, online gaming, online financial transactions, online advertising, and many forms of e-commerce. Hence the combination of a faltering traditional business model, a universally shifting paradigm towards digital content, and the need to regain weight in an environment dominated by new players in the industry, has lead telecom operators to enter the playing field so far occupied by Apple, Microsoft, Google, Facebook, etc. This move is carried out on the back of large amounts of financial resources and the infrastructure built over the past decades. Probably the most visible recent move into this direction is provided by the Spanish Telefónicagroup which underwent profound restructuring in 2011 involving the creation of a new business unit: TelefónicaDigital (El Pais 05 September 2011). The new business unit of Telefónica focuses specifically on steering Telefónica’sglobal business towards the creation and marketing of digital services including internet services, cloud computing, videos, e-financial services, mobile advertising, and M2M (machine-to-machine) services. As part of this restructuring, Telefónicacreated an internal ‘R&D Patent Office’ to ensure the group creates a patent portfolio by moving from an environment in which patents played no or a minor role to one where patents are seen as an integral part of Telefónica’snew business model. The example of Telefónica’srestructuring offers a compelling illustration of the strategic re-positioning of operators by moving into digital markets and the resulting perceived need for strategic patent portfolios. In this paper, we document the major technological shift that has occurred in

2The European Patent EP1186189 (priority date 08.03.1999) illustrates what we might expect in the near future. This patent, entitled ‘Method of allocating access rights to a telecommunications channel to subscriber stations of a telecommunications network and subscriber station,’ was originally owned by the German manufacturer Bosch. However, IPcom, a NPE, acquired the patent from Bosch and uses it to sue HTC (a smartphone producer). In addition, IPcom also goes after telephone operators: British Vodafone and German T-mobile.

3 the telecommunication operator industry and investigate the likely consequences for operators. We show that operators have recently started to move into technologies related to digital data transmission. We also show that these technologies have been most directly affected by the increase in strategic patenting and the accompanying problems such as thickets as well as the move towards litigation. Our analysis shows that telecom operators increase their patenting activity most heavily in these technologies – despite the substantial potential for conflict and costs involved in strategic patenting. This offers evidence on how technological change in form of technological convergence and a corresponding increase in complexity can push an industry into patent races. The remainder is organized as follows. Section 2 provides an overview of the telecom industry and the technological change it has undergone over the past two decades. Section 3 describes the data used in our analysis and Section 4 discusses our empirical approach and shows our results. Section 5 offers some concluding remarks.

2 The telecommunication industry

2.1 Background In recent years, the telecommunication industry has been undergoing a radical trans- formation accompanied by the evolution from a supply oriented industry offering basic fixed line services to a demand oriented industry that offers mobile telecommunication services, and integrates IT and media services (Anderson and Mollyerd, 1997). In particular, the move towards Internet Protocol (IP) based services has substantially changed the telecommunication landscape and enriched the service portfolio of mobile telecommunication providers (Li and Whalley, 2002). Through new services, such as Voice over IP (VoIP), IP telephony, video portals, social com- munities or mobile payment systems operators are able to expand their customer base from selling services not only to subscribers (who pay for use of infrastructure) to potentially everyone that is connected to the internet. According to the European Telecommunications Network Operators Association (ETNO), the total number of mobile subscribers in Europe has grown from approximately 597 million in 2006 to 763 million in 2011.3 Similarly, the mobile subscriber penetration rate increased by 25.5 percentage points over the period, exceeding 127 percent (127 mobile subscribers over 100 inhabitants) averaged across European countries in 2011. The fixed line subscriber base in contrast dropped from 261 million in 2006 to 207 million in 2011 as more and more subscribers switched to VoIP using their broadband lines or mobile devices. In 2011, 61 million subscribers used VoIP-based services not taking into account those customers using over-the-top (OTT) solutions such as Sykpe, Facebook or Google Talk (ETNO, 2012). Figure 1 presents the development of the number of subscribers fixed, mobile and VoIP services. Another development that has influenced the telecommunication business is the convergence of telecommunication technologies with internet technologies that have

3Including Turkey, excluding Russia, Ukraine and Georgia.

4 Figure 1: Development of the total number of fixed telephone line, mobile cellular and VoIP line subscribers in Europe followed two separate trajectories until the mid 1980s. The first refers to the digitization of the telecommunication network starting in the early 1970s that enabled digital data transmission (Davies, 1991). The digital transmission and the related ISD standards enabled digital switching and supported digital air interfaces in mobile telecommunication systems. The other development refers to the internet which is based on the TCP/IP protocol that standardizes the rules of packaging, trans- mitting and receiving data over the internet. Whereas the traditional telecommunication technologies are based on circuit-switching, the internet-related technologies are based on packet-switching that provide higher flexibility and lower costs (Bohlin et al., 2000). However, the increased importance of the internet for telecommunication means that applications and services related to mobile telecommunication have to be com- patible with the TCP/IP protocol. This is evident in a range of standardization efforts around core data transmission technologies. For example, the European Telecommunications Standards Institute (ETSI) decided in early 1998 to adopt a new standard for so-called third generation mobile services, referred to as universal mobile telephone system (UMTS). The rising popularity of internet-based applications that rely on the infrastructure of telecommunication operators is evident in the increasing demand for data services. In the European Union (EU), the broadband penetration for mobile subscribers rose from 13 percent in 2008 to 54 percent in 2013. Particularly in Nordic Countries (i.e. Denmark, Sweden and Finland), the mobile broadband penetration, averaged across handheld devices and computer users, reached already 100 percent in 2013 (EC, 2013). Cisco Systems estimates that the total data volume in mobile networks will grow at a compounded annual

5 growth rate of 66 percent from 520 petabytes per month at the end of 2011 to 11.2 exabytes per month in 2017 (Cisco, 2013).

2.2 Structural changes Traditionally, the relatively stable environment that characterized the telecommunication industry over a long period consisted of three main technology layers. The equipment layer provides essential network elements, transmission and switch systems; the network layer provides infrastructure to connect sender and recipient; content and services were located in the third layer (Fransman, 2001; Li and Whal- ley, 2002). The network and the equipment layer were connected by air-interface standards such as GSM (Global System for Mobile Communication) and CDMA (Code Division Multiple Access) (Peppard and Rylander, 2006). Through their monopoly position before deregulation in the 1990s, telecommunication incumbents controlled the entire value chain across layers (Steinbock, 2003). However, the rapid technological developments that have been driving the above- mentioned convergence have added complexity. Fransman (2001) argues that due to the telecommunication industry’s increased complexity, it is now better characterized by the interplay of six diﬀerent layers ranging from equipment and software, network, navigation and middleware, applications including content as well as customers. This increased complexity and interdependence has lead a number of scholars to describe the evolution of the telecommunication industry as a shift from a value chain to a value network (Li and Whalley, 2002; Olla and Patel, 2002; Stein- bock, 2003; Peppard and Rylander, 2006; Tilson and Lyytinen, 2006; Funk, 2009). This has several implications for the mobile telecommunication industry. First, the reduction of barriers to entry has drawn new players from other industries into the market and given them direct access to customers. Today, content and services are likely to be developed and managed by highly specialized companies with latest technologies that support most advanced services (Peppard and Rylander, 2006). However, telecommunication operators do not control these newly created revenue streams and there is an ongoing debate about the provision of third- party applications using the infrastructure owned by operators (FT, 2012). Second, the convergence of telecommunication with internet-based technologies has given rise to new areas of competition in the “datacom space” related to data and voice services over IP based networks (Seaberg et al., 1997). Coupled with higher-capacity wireless networks, in particular those followed by the evolution from UMTS to LTE (Long Term Evolution), and an almost complete mobile broadband coverage,4 this has given large internet companies the opportunity to enter the market. Today, companies such as Yahoo, Microsoft, Facebook or Twitter are playing a dominant role in the telecommunication industry. While these new markets are characterized by aggressive competition (Fertig et al., 1999; Birch and Burnett- Kant, 2001; Li and Whalley, 2002), telecommunication incumbents are forced to

4The European Commission recently announced that the target of 100 percent basic broadband coverage across Europe was achieved ahead of schedule. By the end of 2012, 99.4 percent of EU households had access to basic ﬁxed or mobile broadband coverage (EC, 2013).

6 invest billions in spectrum license auctions5 and the maintenance of extensive pro- prietary infrastructures6 to maintain their key position in the corresponding mobile delivery channels. On the other hand, the fast growth of IP-based services allowed entrants to challenge the dominant position of operators. Internet companies have also been relying extensively on acquisitions (Chan-Olmsted and Jamison, 2001; Waverman and Trillas, 2002; Pennings et al., 2005) to offer the latest technologies and a broad range of technological alternatives to customers. A case in point is Microsoft’s acquisition of Skype in October 2011 for around USD 8.5 billion, representing Microsoft’s largest acquisition to date.7 At the time Skype had a relatively large user base of some 170 million customers and invested heavily in the acquisition of the intellectual property enhancing its peer-to-peer (P2P) network (Microsoft, 2011). The competitive advantage of companies such as Skype lies in their ability to provide services and applications on handheld mobile devices that had already reached market acceptance on the computer. As some of these services have become financially more lucrative than traditional telecommunication services like text messaging or voice telephony, they have become serious competition for operators. Declining stock prices, persistent high levels of debt (Isern and Rios, 2002), profit warnings (FT, 2012) and negative price trends8 are only some indicators that illustrate how telecommunication operators are struggling with the consequences of the fast changing environment, in particular the shift away from traditional voice and text messaging towards the provision of digital data services. To illustrate this, Figure 2 shows the long-term drop in traditional voice and text messaging operations for the Spanish mobile telecommunication industry.9 Adjusted for the number of customers, the average monthly mobile voice traffic decreased by 3.8 percent from 119.4 minutes in 2008 to 114.9 minutes in 2011. The decline in text messaging is even steeper with a drop of 33.6 percent between 2006 and 2011 (from 19.7 in 2008 to 13.1 in 2011). On the revenue side, Figure 3 shows overall revenue over time as well as the three main revenue streams: voice, data and text messaging. From 2008 to 2011, revenues from voice and text messaging operations declined from EUR 11.3 and 1.6 billion to EUR 8.2 and 0.7 billion, respectively. Besides an increase in competition, regulatory measures have also played a key part, with cuts to mobile termination rates (MTR) since 2009 contributing to declining

5For the 20 years non-tradeable UMTS licenses that were auctioned in Germany in 2000, operators paid the amount of EUR 50.8 billion. 6Investments in mobile networks represented 45.5 percent of the total capital expenditure in Europe in 2011 or EUR 20.7 billion, a 5.6 percent increase compared to 2010 as network operators need more and more capacity to meet customer demand for mobile data, investing in 3G and first 4G networks (ETNO, 2012). 7Skype was founded in 2003. The former European-based startup was acquired by eBay for approximately USD 2.5 billion in September 2005 and then acquired by an investment group led by Silver Lake in November 2009 (Microsoft, 2011). 8In the EU-27 countries, the average price for a mobile minute dropped from 16.3 cent in 2003 to 13.8 cent in 2008 (Péladeauet al., 2010). 9We chose the Spanish the telecommunication industry because the Spanish national regulatory authority CMT (Comisióndel Mercado de las Telecomunicaciones) offers rich data on the Spanish market. The data contain not only information about number of subscribers, minutes of traffic, and revenues, but also distinguishes between pre-pay and post-pay clients, and on-net and off-net traffic. Moreover, it contains data about revenues obtained from fixed monthly subscription fees (CMT, 2012).

7 Figure 2: Development of the SMS and voice traﬃc in Spain voice revenues for operators (Kongaut and Bohlin, 2012).10 In contrast, mobile data and internet revenues show an upward sloping trend over time and are gaining importance in relative terms: in 2009, data revenue was about 11.5 percent of mobile service revenues and surpassed text messaging revenues, accounting for 9.7 percent. In 2011, the provision of data and internet related services reached almost 20 percent. This clearly suggests that structural changes rather than cyclical factors are quickly transforming the industry.

2.3 Operators’ strategic choices To avoid “degenerating into a bit pipe,”11 mobile operators decided to revise their corporate strategies and re-position themselves in the market place to compensate for dwindling revenue from the traditional telecommunication business. Operators have recently begun to emulate big internet companies such as Google or Yahoo and to differentiate their offer towards “converged” products and services. Mobile operators have approached the strategic challenge in multiple ways. As mentioned above, perhaps the most drastic and clear-cut example for the strategic makeover of mobile operators is Telefónica. The Spanish company underwent

10However, it should be noted that cuts to MTR might also provide beneﬁts for operators in terms of reduced operating costs. 11“Bit pipe” describes a situation in which mobile telecommunication operators are only needed for data-transmission and billing processes. This would result in low margins as the operation of the network would most likely be highly competitive due to lack of product/service diﬀerentiation and excess capacities.

8 Figure 3: Development of Data, Voice and SMS service revenue in Spain a wholesale restructuring which resulted in a new global business unit in 2011: TelefónicaDigital. With headquarters in London and regional offices in Silicon Val- ley, SãoPaulo, Spain and Israel it was set up specifically to develop digital services. According to Telefónica,the unit is expected to create annual revenues of approximately EUR 5 billion by 2015 with an annual revenue growth rate of 20 percent across eight key sectors: financial services, M2M, advertising, eHealth, video & media, security, cloud service and applications (Telefónica,2012). TelefónicaDigital reflects not only the shift from analog to digital technologies, it also illustrates the move by operators towards new mobile value added services including Near Field Communication (NFC) (e.g. mobile payment systems), Machine-to-Machine Com- munication (M2M) (e.g. vehicle tracking tools or remote clinical care) and service offerings based on IP Multimedia Core Network Subsystem (IMS) for delivering IP multimedia services (e.g. multimedia content sharing during a voice call, video call and video sharing). Similar efforts to restructure fundamentally the business model of operator companies have been found by van Kranenburg and Hagedoorn (2008) for three other major European operators: Royal KPN NV, Deutsche Telekom AG and British Telecom PLC. van Kranenburg and Hagedoorn (2008) show that these companies have entered new technologies not only through the creation of new subsidiaries or business units as in the case of TelefónicaDigital, but also through strategic alliances and joint ventures. Pennings et al. (2005) confirm the increased importance of inter-firm partnerships for operators in entering digital technologies. The authors show that since the beginning of the 1990s operators have increasingly sought al-

9 liances with firms outside the telecommunications industry. This pattern is easily explained by their need to gain access to new and complementary capabilities and resources and to integrate them with their traditional business model. For example, TelefónicaDigital recently signed a partnership with Visa Europe to develop new products and services in areas such as mobile wallet or contactless payments. TelefónicaDigital has also acquired an equity stake in the music streaming provider Rhapsody International and a strategic alliance with Masternaut involved in telem- atics solutions. Apart from internal restructuring, alliances, and M&A, mobile operators have also recognized the importance of patents for digital data technologies. Hall and Ziedonis (2001) showed that patent filings took off in the U.S. semiconductor industry at the end of the 1980s and that this is explained by an increase in companies’ patenting propensity for strategic reasons rather than increased innovative activity. Patenting propensities in ICT displayed even more pronounced an increase during the same period. The patents-R&D ratio (patents per 1992 US$ million) increased from around .4 in 1987 to .9 in 2000 – for comparison, the increase in semiconduc- tors was from .3 to .6 over the same period (updated Figure 1 of Hall and Ziedonis (2001)). More recent data by Fink et al. (2013), who cover 13 OECD economies, show that patenting propensities in ICT continued to climb during the 2000s. Fink et al. (2013) also show that the relative increase in patent filings in ICT is in fact partly due to extremely fast growth in filings on digital communication technologies – filings increased in absolute terms by almost 15 percent per year between 1995- 2008, outstripping by far the R&D growth rate in that sector. Our analysis focus on this aspect of changes in operators’ behavior following the structural technological change that the industry has undergone in the past two decades.

3 Data

The data used in this study come from four main sources. First, we collected quarterly information on the mobile telecommunication industry in 15 Western European economies from the Business Monitor International Telecommunication Reports.12 These reports cover general and operator specific market data, operator profiles, company histories, financial information, as well as telecommunication network and service information. Apart from providing detailed data on the telecommunication industry, the reports also allowed us to identify the complex ownership structure of the European mobile telecommunication companies. The second data compo- nent provides additional information on companies’ ownership structure, in form of ownership links provided by Bureau Van Dijk’s (BVD) Amadeus database for the period between 2000 and 2008, and M&A data from SDC Platinum and Zephyr. The ownership data identifies business groups, while M&A data allows us not only to attribute changes in business groups over the period between 2000 and 2008 but also helps us reconstruct ownership links to affiliates that have dissolved prior to 2000, and fully integrated into the parent company. Third, our source of patent data is the EPO Worldwide Patent Statistical Database (edition April 2012). Fourth, firms’

12Countries included are Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ire- land, Italy, Netherlands, Portugal, Spain, Sweden, Switzerland and the UK.

10 financial data were extracted from OSIRIS provided by BVD. We combined the different datasets in the following way. We started by identify- ing the ultimate owners of all European telecommunication companies. Among the set of owners, we selected only those that are themselves telecommunication operators and that make more than half of their revenue from operations in Europe, using the Business International Telecommunication Reports and the operators’ annual reports. We then traced each operator’s history to identify possible name changes and restructuring. In sum, we identified a total of 10 telecommunication operators that are of potential interest for our study.13 Next, we retrieved all wholly owned active subsidiaries of these 10 parent firms between 2000 and 2008 from Amadeus. We then matched these firms to the SDC M&A and Zephyr databases to determine whether any of these affiliates have been directly acquired by the parent company or its subsidiaries between 2000 and 2008. Finally, we identified companies that have been directly acquired by a sample firm or by one of the affiliates of a sample firm prior to 2000 that did not exist as a separate company between 2000 and 2008. In a last step, we matched all of the company names to the patent data. For matching, we standardized assignee and company names and relied on string matching.14 After some manual checks and corrections, we matched 42,672 patent applications to 127 patent holders (i.e. assignees). Of all identified patent applications, 61 percent are filed under the name of the parent company, 11 percent under the name of one of the affiliates of our sample firms, and 28 percent are incoming patent application via a merger or an acquisition. In order to identify telecommunication related patents, we used the concordance table provided by Palmberg and Martikainen (2006). Based on the IPC classes, the authors link four-digit technology classes into technology categories. Table 1 distinguishes between the traditional telecommunication technologies, internet related telecommunication technologies and several other relevant areas related to next generation applications and network technologies.

4 Analysis

In this section, we first describe a few stylized facts about changes in patenting behavior by telecom operators. Next, we analyze factors that have influenced the choice by operators to obtain patents in specific technologies. Finally, we investigate the different modes of patent acquisition.

13Operators included are Deutsche Telekom (Germany), Elisa (Finland), Orange (France), KPN (Netherlands), Telecom Italia (Italy), Telefónica(Spain), Telenor (Norway), TeliaSonera (Sweden), Swisscom (Switzerland), and Vodafone (UK). 14Compared to approximate string matching (ASM), the advantage of this conservative method lies in a high level of precision at cost of some loss of completeness. As a first step, company and assignee names were standardized following the steps proposed by Magerman et al. (2006). The main standardization procedures involved in this study cover the following categories: character cleaning; punctuation cleaning (pre-parsing); legal form indication treatment; spelling variation standardization; condensing; umlaut standardization. In a second step, we matched the sample firms with the list of the harmonized patent assignee names.

11 Technology Categories Four-digit IPC classes

Traditional telecom Transmission H04B, H01Q, H01P, H04J, G01R Switching H04Q, H01H Voice applications and equipment H04M, H94R, G10L

Internet related telecom Data and internet applications G06F, H04L, G06N Encrypting and security H04K User authentication and access control G09F

Applications Pictoral communication H04N Positioning G01S Games A63F Electronic payment G07G Mechanical technologies B23K, B29C, G06N, H05K, H01B, H01R, H02B, H02G Codecs and algorithms H03M, H03L Machine to machine G08C Photography G03B

Others Impedance Networks H03H Pulse Technique H03K Broadcast Communication H04H Wireless Communication Networks H04W Recognition of Data G06K Image Data Processing G06T

Table 1: Concordance table between IPC classes and ICT relevant technology categories

12 45000 35%

30% 40000

25%

35000 20%

15% 30000

Number of patent ofpatent applications Number 10%

25000 Shareof internet related telecomtechnologies 5%

20000 0% 2000 2001 2002 2003 2004 2005 2006 2007 2008

Share internet related telecom Patent stock

Figure 4: Operator patent stock size

4.1 Changes in patenting activity Figure 4 shows the evolution of the patent stocks of the mobile telecommunication operators between 2000 and 2008, respectively. The operator patent stocks increase at a relatively constant rate since 2000. When we look at the distribution of patenting activities across technological fields, we can clearly see a shift towards digital technologies. Filings related to “Data and internet related applications” which are internet-related telecommunication technologies have grown fastest, followed by “Wireless Communication Networks.” Traditional telecommunication technologies such as “Transmission”, “Switching”, and “Voice application and equipment” have also grown over time but at much slower rate. This shows that internet-related communication technologies have become more important than traditional telecommunication technologies. In fact, patent applications in traditional telecommunication technologies have remained relatively stable accounting for around 28 percent of our operators’ entire patent portfolio over the sample period between 2000 and 2008. On the other hand, the share of internet-related telecommunication technologies increased significantly from 17 percent to 32 percent representing now the largest technological category. Another interesting observation is the relatively low number of patent applications related to the various technologies relevant to smartphones. Only patent applications in “Pictoral Communication” have increased significantly accounting for 6 percent of the patent stocks in 2008. Other fields such as “Electronic Payment”,

13 Data and internet applications

Wireless Communication Networks

Voice applications and equipment

Transmission

Pictoral communication

Switching

Recognition Of Data

Image Data Processing

Codecs and algorithms

Positioning

Broadcast Communication

Mechanical technologies

Machine to machine

Encrypting and security

User authentication and access control

Games

Electronic payment

Pulse Technique

Impedance Networks

Photography

0 2000 4000 6000 8000 10000 12000 14000 16000

2008 2000

Figure 5: Ranked distribution of patent applications by technological ﬁelds

“Positioning” or “Photography” have remained at low levels by 2008. In order to measure the concentration of patenting across technologies, we com- pute the Herdindahl index.15 We find that the Herfindahl index increased from 0.08 to 0.14 (a 75 percent increase). There was a substantial increase in the concentration of patenting activities around a few technology classes between 2000 and 2008. Looking at the distribution of patents across technologies in Figure 5, the increase in concentration is driven by an increased focus on technologies related to digital data transmission. On closer inspection of the four-digit IPC classes that are underlying the technological field “Data and internet applications,” we see significant differences between the three classes G06F, H04L and G06N. Most noteworthy is the dramatic increase in patent applications in class H04L (‘transmission of digital information’): Out of 9,237 new patent applications in “Data and internet applications,” subclass H04L contributes 81 percent to that increase. As a consequence, patent applications in H04L represent 26 percent of total patent applications in 2008. The increased importance of H04L applies across all operators. H04L contains technologies for the transmission of digital information. A core application of this technology is the Internet. Thus, the significant increase of patent filings in this technological field shows the increased importance of convergent technologies and demonstrates the move of operators towards new technological domains related to digital data.

15The Herfindahl index measures the concentration of patenting activities across technological fields. A high value of the Herfindahl index indicates high concentration of patenting activities.

14 60%

50%

40%

30%

digit digit IPC classes -

20% Shareof four 10%

0% 2000 2001 2002 2003 2004 2005 2006 2007 2008

Share one IPC class Share two IPC classes Share three and more IPC classes

Figure 6: Distribution of patent applications in H04L (‘transmission of digital information’) by the numbers of distinct three-digit IPC classes

Since increased technological complexity and convergence may require the combination of different technologies, more patents might include the H04L class in addition to a number of other technologies. To investigate this, Figure 6 shows the share of patents with one, two, and three IPC subclasses per patent application for all firms that have patents in class H04L. The figure shows that the share of patents with only one subclass has increased substantially over time and accounts for almost 60 percent in 2008. This means that companies are genuinely shifting towards technologies directly related to digital data transmission instead of broadening existing technologies to also cover digital data transmission technologies.

4.2 What explains technology choices? The previous section has shown a large increase in absolute and relative terms of patents on digital data transmission held by telecom operators. In this section we show that operators have been forced to increase their patent holdings in this technology area despite of – or precisely because of – the complex patent landscape that characterizes digital technologies. In fact, we show that counter-intuitively operators have increased their patent holdings especially in technologies that hold large potential for conﬂict and costly litigation. This points toward patent portfolios reﬂecting a strategic need in technologies characterized by a complex patent landscape.

15 The existing literature has identiﬁed a number of sources of conﬂict that have been found mostly in digital data transmission technologies:

• Standard essential patents (SEPs): Although technology standards are pervasive across industries (Baron et al., 2013), patenting of standard related technologies has been most prevalent in ICT. Standard essential patents are of particular value to companies because infringement can be inferred from the use of the standard.16 Although there are obvious beneﬁts to technology standards (above all compatibility and interoperability), the role of SEPs has been more controversial. For example, Blind et al. (2011) oﬀer survey evidence that shows that participants of standard setting organizations consider SEPs to make it harder to reach consensus and slow the standardization process. Companies that do not own SEPs indicated that they encounter problems ne- gotiating licensing conditions. Kang and Bekkers (2013) argue that SEPs are often the outcome of strategic insertion of patents into the standard despite little technological value of these patents to the standard. Perhaps more im- portantly, SEPs can create market power that goes beyond the patent but that derives from the standard.17 Even (F)RAND licensing obligations often do not solve the issue as frequently disputes about the actual (F)RAND royalty rate erupt. Moreover, royalty stacking might also occur if ownership of SEPs on a given standard is split across multiple patent owners. To measure the existence of SEPs across technologies, we use the number of SEPs by IPC subclass.18

• Patent thickets: Patent thickets arise when different patent owners hold patents with overlapping claims (Shapiro, 2001). Such overlaps can create problems as companies may block each other’s use of the technology protected by the overlapping set of patent claims. The more complex technologies are (in the sense defined by Cohen et al. (2000)), the more prevalent are such thickets (von Graevenitz et al., 2011). Patent thickets are associated with increased transaction costs for companies. For example, Cockburn and Mac- Garvie (2011) show for the software market that patent density (which can be regarded as a proxy for thickets) within narrowly defined markets is associated with less product market entry. They also find that entry depends on the size of a company’s own patent portfolio – companies with larger patent portfolios pre-entry are more likely to eventually enter the market. Hall et al. (2013) come to similar conclusions, although they look at entry at the technology level. They find a negative association between entry into technologies and

16According to ETSI (2011: 6), “essential” means that it is not possible on technical (but not commercial) grounds, taking into account normal technical practice and the state of the art generally available at the time of standardization, to make, sell, lease, otherwise dispose of, repair, use or operate equipment or methods which comply with a standard without infringing that IPR. 17As explained by Richard Posner in his judgment in Apple vs. Motorola, “The purpose of the FRAND requirements [...] is to conﬁne the patentees’ royalty demand to the value conferred by the patent itself as distinct from the additional value – the hold-up value – conferred by the patent’s being designated as standard-essential.” (Judge Posner, 2012: 18). 18We obtained the data from Baron et al. (2013), who collect the data from almost 6,300 diﬀerent ICT standards.

16 thickets density. These studies show that markets and technologies characterized by patent thickets generate transaction costs for companies that make entry more diﬃcult. To capture the prevalence of patent thickets, we use the ‘triple counts’ measure developed by von Graevenitz et al. (2011) at the IPC subclass level.

• NPEs: NPEs usually acquire patents with the intention to monetize them, often under the threat of litigation or directly through litigation. In most cases, they do not engage in innovative activity or sell products in the product market (Hagiu and Yoffie, 2013). NPEs pose particular challenges to producing companies. They are difficult if not impossible to counter-sue precisely because they do not undertake any productive activity. They also often ambush companies by initiating legal action when an injunction would be particularly costly to the producing firm (exploiting hold-up). Bessen and Meurer (2014) present survey data that suggests that larger companies spend on average around US$1.5 million in legal expenses on a law suit against a NPE and more than US$7 million in licensing fees. There has been an astronomic rise in lawsuits brought by NPEs in the U.S. – the number of cases jumped from 144 in 2001 to 3,134 in 2013 involving 3,716 producing companies.19 Using U.S. data for 2000-2007, Allison et al. (2010) show that software and telecommunications patents are generally litigated more often than patents in other technology fields and especially so by NPEs. We take the presence of NPEs by IPC subclass into account by constructing a measure of NPE presence. We identified patent holding companies and large patent aggregators based on patent holder litigant data provided by Cotropia et al. (2012).20 We extended these data and checked for name variations by using additional information on NPE assertions from other sources, such as RPX and Patent Freedom. We then matched these NPE names with patent data.

• Litigation: Litigation is often the end point of a patent dispute. It is well- known that only a tiny fraction of patent disputes ends in court (Greenhalgh et al., 2010) as court cases engender often very substantial legal expenses and expose a business to enormous uncertainty. Despite this, there has been a recent strong increase in patent litigation speciﬁcally on information and communication technologies. To measure the risk of litigation for patents by IPC subclass, we obtained data on patent litigation from Cremers et al. (2013). The data covers all patent cases in Germany, the UK, France, and the Netherlands over the period 2000-2008.21 We allocated patent cases to IPC subclasses based on the patents involved in a lawsuit and created litigation case counts by IPC subclass and year (where year means the year a law suit was initiated).

19See patentfreedom. 20We manually cleaned the data to eliminate large patent aggregators that are part of a producing entity. 21Since our focus is on European telecom operators, data on litigation in Europe is preferable over U.S. litigation data.

17 • Licensing: Firms that are active in technology fields that are characterized by SEPs, patent thickets, and litigation often resort to licensing to resolve costly (legal) disputes and the possibility of injunctions. Siebert and von Graevenitz (2011) show for the semi-conductor industry that licensing can help resolve disputes in the presence of patent thickets. But their evidence also indicates that the more fragmented patent rights are, the lower likelihood of licensing. Cockburn et al. (2010) also suggest that firms facing patent thickets have a higher propensity to engage in in-licensing, which supports the view that licensing can mitigate the defragmentation of rights. Hence, technology fields in which technology transactions in the form of licensing take place might be able to mitigate at least to some extent the problems mentioned above. We extracted licensing data from SDC Platinum and ktMINE and matched the data to BvD’s Amadeus database to allocate licenses to NAICS industry classes. We used an IPC-NAICS correspondence table provided by Lybbert and Zolas (2013) to transform license counts by industry into license counts by IPC subclass.

Figure 7 shows the combination of SEPs, patents held by NPEs, and thickets density by IPC subclass. The figure shows that a few IPC subclasses stand out. Subclass H04L is in the top right corner, which means it scores high in terms of SEPs, NPE patents, and the size of the bubble shows that patent thickets are most prevalent within this subclass. As shown above, it is precisely subclass H04L where operators have increased their patent holdings the most. This is noteworthy because as Figure 7 demonstrates, this is the subclass that clearly stands out in terms of char- acteristics that are known to be highly problematic. The figure also suggests that a number of other subclasses H04B (transmission), H04W (wireless communication networks), H04M (telephonic communication), and H04N (pictorial communication) within class H04 (electric communication) are problematic.22 Outside of H04, only subclass G06F (electrical digital data processing) stands out. A closer look at the data underlying Figure 7 confirms that the operators in our sample do not contribute in any significant way to the density of SEPs or thickets, that is, they enter technology fields characterized by a strong presence of SEPs, thickets, and NPE activity. It is important to emphasize that operators moved into these digital technologies due to the exogenous technological change that has occurred in their industry (see Section 2.2 above). Hence, the increase in their patent holdings in related technologies (H04 and G06F) coincides with entry into these technologies rather than a simple change in patenting behavior. In other words, in order to enter digital technologies, operators were forced to engage in patenting in technology fields most heavily affected by the complex patent landscape outlined above. This point is made more formally by the regression results shown in Table 2. The data for the regressions consists of 30,821 patent filings by our sample of operator companies over the period 2000-2008. The unit of observation is the IPC subclass-year where we include all IPC subclasses in which at least one patent was filed during our sample period (a total of 401 distinct subclasses). Our main interest is in the coefficients on the litigation, SEP, NPE, licensing and thickets measures.

22The concentration of SEPs in class H04 has also been found by Blind et al. (2011).

18 2000

1800

1600 G06F

1400 H04L

1200

1000

800

600 H04N Average number of triples

H04W 400 H04B 0-100 triples

Average numberAverage of standard essential patents H04M 101-1000 triples 200 G10L H04J > 1000 triples 0 0 500 1000 1500 2000 2500

Average patent portfolio size of NPEs

Figure 7: Patent landscape by IPC subclasses (2000-2008)

19 The specification also includes operator dummy variables (France Telecom (Orange) is the omitted category), a dummy for IPC subclass H04L and the log or total assets and R&D intensity of operators. Table A-1 in the Appendix contains descriptive statistics. The results are striking: there is a positive association between the number of patents filed within IPC subclasses and the different measures of the potential for conflict associated with the patent landscape. One would expect companies to avoid entering patent classes that are characterized by the problems discussed above. For example Hall et al. (2013) find that companies are less likely to file patents in classes characterized by high density of patent thickets. Our results suggest that the oppo- site holds for telecom operators. Note that the coefficient on the licensing variable is negative and statistically significant. This suggests that IPC subclasses characterized by more licensing activity see fewer filings by operators. This suggests that patent filings and licensing act as substitutes in this context. When we include the H04L dummy in column (III), we find a large, positive coefficient. This underscores the descriptive findings above.

20 Table 2: Determinants of patent ﬁlings by IPC subclass (negative binomial regression)

(I) (II) (III)

Ln(# litigation cases) 0.551 0.468 (0.353) (0.354) Ln(# standard essential patents) 2.803∗∗∗ 2.713∗∗∗ (0.245) (0.245) Ln(NPEs patent stock) 2.228∗∗∗ 2.205∗∗∗ (0.277) (0.277) Ln(# licensing agreemeents (adj.)) -5.181∗∗∗ -4.968∗∗∗ (1.624) (1.623) Ln(# triples) 0.810∗∗∗ 0.753∗∗∗ (0.232) (0.232) Dummy for H04L 4.854∗∗ (1.956) Ln(Assets) -2.138 -0.895 -0.891 (4.141) (2.200) (2.206) Ln(R&D intensity) 1.602 1.029 1.028 (2.630) (0.984) (0.978) Mannesmann 37.91∗∗∗ 95.93∗∗∗ 94.04∗∗∗ (11.60) (18.00) (17.71) Deutsche Telekom -11.61∗∗∗ -5.513∗∗∗ -5.480∗∗∗ (1.929) (0.729) (0.725) Elisa -17.11∗∗∗ -10.68∗∗∗ -10.64∗∗∗ (2.583) (1.335) (1.360) KPN -12.86∗∗∗ -7.608∗∗∗ -7.580∗∗∗ (2.803) (1.385) (1.389) Swisscom -16.99∗∗∗ -9.920∗∗∗ -9.901∗∗∗ (2.064) (1.384) (1.387) Telef´onica -14.27∗∗∗ -7.535∗∗∗ -7.500∗∗∗ (1.345) (0.698) (0.695) Telenor -14.91∗∗∗ -9.584∗∗∗ -9.554∗∗∗ (2.641) (1.218) (1.228) Telecom Italia -9.976∗∗∗ -5.405∗∗∗ -5.386∗∗∗ (2.141) (0.812) (0.806) TeliaSonera -6.493 -6.308∗∗∗ -6.290∗∗∗ (5.508) (2.221) (2.237) Vodafone -7.663∗∗ -5.304∗∗∗ -5.218∗∗∗ (3.423) (1.390) (1.394)

Year Dummies Yes Yes Yes Observations 2205 2205 2205 Wald χ2 363.28∗∗∗ 1800.37∗∗∗ 1891.15∗∗∗

Notes: Negative binomial regression. Marginal effects reported. Heteroskedastic-consistent (robust) standard errors in parentheses. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01. Operator dummy variables benchmark: Orange. Dependent variable: number of patent applications. Unit of observation is IPC subclass-year and the sample consists of all IPC subclasses with at least one patent application over period 2000–2008. Summary statistics of variables are listed in Table A in the Appendix. Each regression includes an intercept as well as year-fixed effects and operator dummies.

In Table 3 we split the sample period into before and after the mobile internet boom in 2005. The main focus is on the coefficients on the H04L dummy variable. The marginal effect doubles in the 2005-2008 period, indicating a strengthening of patenting efforts in this subclass.

21 Table 3: Determinants of patent ﬁlings by IPC subclass – 2001-2004 and 2005-2008 (negative binomial regression)

Period 1: 2001-2004 Period 2: 2005-2008 “Before” “After”

Ln(# litigation cases) 0.0224 -0.0859∗∗ (0.040) (0.036)

Ln(# standard essential patents) 0.245∗∗∗ 0.246∗∗∗ (0.022) (0.027)

Ln(NPEs patent stock) 0.211∗∗∗ 0.161∗∗∗ (0.023) (0.031)

Ln(# licensing agreemeents (adj.)) -0.473∗∗∗ -0.512∗∗ (0.134) (0.229)

Ln(# triples) 0.0793∗∗∗ 0.126∗∗∗ (0.021) (0.025)

Dummy for H04L 0.276∗ 0.546∗∗∗ (0.158) (0.167)

Ln(Assets) 0.676∗∗∗ 0.506 (0.197) (0.519)

Ln(R&D intensity) 0.133 -0.0877 (0.119) (0.134)

Firm Dummies Yes Yes Observations 1106 773 Robust Wald χ2 test 71.27∗∗∗

Notes: Negative binomial regression. Heteroskedastic-consistent (robust) standard errors in parentheses. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01. Dependent variable: number of patent applications. Separate results for 2 sub-periods 2001–2004 (‘before the mobile internet boom’) and 2005–2008 (‘after the mobile internet boom’). The unit of observation is IPC subclass-year and the sample consists of all IPC subclasses with at least one patent application over period 2001–2008. Summary statistics of variables are listed in Table A in the Appendix. Each regression includes an intercept as well as year-fixed effects and firm dummies.

These results support the main proposition of our analysis: the exogeneous shift in telecommunication technology forced telecommunication operators to adapt their business towards digital technologies and hence pushed them towards patenting in technologies that are characterized by a complex patent landscape which is likely to confront operators with substantial challenges in the (near) future.

4.3 Mode of acquisition A related question is how operators have amassed patents in digital technologies. More in general, companies can develop the technology internally and then patent it, buy patents through arms’ length contracts, or acquire entities that hold relevant patent portfolios (Aurora et al., 2001).23 Table 4 shows the results from a

23Table A-2 in the Appendix gives an overview of the parent company as assignment mode for the 20 largest IPC4 classes.

22 multinomial logit where we compare market-based sources of patents with in-house development.24 We ﬁnd the coeﬃcients on the H04L dummy to be negative for modes ‘subsidiary’ and ‘acquisition.’ This means that patents in H04L and in fact in digital technologies more generally (including G06F) originate mostly directly from operators.

Table 4: Multinomial regression models (with robust standard errors) for the assignment mode choice (patent level)

(I) (II)

Mode: Parent vs. Subsidiary

Dummy for H04L -0.114∗∗ (0.045)

Ln(Assets) -1.302∗∗∗ -1.301∗∗∗ (0.143) (0.143)

Ln(R&D intensity) 0.983∗∗∗ 0.984∗∗∗ (0.076) (0.076)

Constant 2.250∗∗∗ 2.268∗∗∗ (0.677) (0.677)

Mode: Parent vs. Acquisition

Dummy for H04L -0.685∗∗∗ (0.076)

Ln(Assets) 7.333∗∗∗ 7.285∗∗∗ (0.440) (0.435)

Ln(R&D intensity) -1.941∗∗∗ -1.856∗∗∗ (0.204) (0.204)

Constant -41.583∗∗∗ -41.202∗∗∗ (2.284) (2.254)

Year Dummies Yes Yes Firm Dummies Yes Yes Observations 32232 32232 Pseudo R2 0.394 0.396 Wald χ2 589705.33∗∗∗ 591289.51∗∗∗

Notes: Multinomial regressions. Heteroskedastic-consistent (robust) standard errors in parentheses. ∗ p < 0.10, ∗∗ p < 0.05, ∗∗∗ p < 0.01. The unit of observation is the patent application and the sample consists of all patent applications over the period 2000–2008. Each regression includes an intercept as well as year-fixed effects and firm dummies.

This shows that operators have started patenting themselves in the critical technology classes (IPC subclasses H04 and G06), supporting the view that patenting strategies of telecom operators have changed. Operators are now heavily engaged in patenting in those technology classes most heavily aﬀected by the various problems discussed in Section 4.1 above. 24Due to the large number of patents acquired by Vodafone from Mannesmann in 2000, we exclude patents assigned to Mannesmann from our regression analysis in this section.

23 5 Conclusion

The telecom operator industry has been directly affected by a major technological change in ICT that was largely driven by outsiders, but which threatened to erode the traditional business model of telecom operators. Operators were slug- gish to respond to this sweeping technological change and the resulting competitive challenges. One particular challenge was the heavy use of strategic patenting in digital data communication. This created a sudden need for operators to adapt their patenting strategies and to build patent portfolios on the new technologies. There- fore, the largely exogenously imposed technological change in the telecom business not only pushed operators to move into a new technological direction, but it also pushed the operator industry away from an equilibrium in which patents played only a minor role for firms’ ability to compete in the market to an equilibrium in which patents are a necessary precondition for participation in the market. In fact, our analysis shows that operators move precisely into technology classes that have been most heavily affected by issues such as patent thickets, NPEs, and standard essential patents. While largely descriptive in nature, our study offers at least three lessons. First, sectors might switch to aggressive patenting behavior as a result of exogenous technological change. Interestingly enough, in the case of telecommunication operators, technological change is not the outcome of innovation within the sector; instead, it has occurred because of technological convergence and blurring boundaries across sectors. This is an important consideration for scholars who investigate the rela- tionship between industry dynamics and patenting. Second, we find an interesting result which contradicts the conventional wisdom in the patents and entry literature. We show that telecom operators enter precisely those technology classes that are characterized by a more complex patent landscape and hence higher transaction costs (including a higher risk of patent-related disputes). Finally, our research has practical implications for telecom operators. It is clearly too early to tell what the effect of the increased importance of patents on the industry is going to be. That said, in the short run, it is very likely to create substantial costs for operators, both in terms of obtaining patents as well as transaction costs involved in managing the complex patent landscape in digital data. As a result, the move towards digital technologies might see more legal patent experts employed by telecom operators, which reflects the strategic nature of their increased patenting activity. We leave an evaluation of the net social effect of such a change for future research.

24 References

J. R. Allison, M. A. Lemley, and J. Walker. Patent quality and settlement among repeat patent litigants. The Georgetown Law Review, 99:677, 2010.

P. Anderson and B.G. Mollyerd. Telecommunication services in context distribution consequences of technological change and convergence. Journal of Service Management, 8(5):453–473, 1997.

A. Aurora, A., Fosfuri, and A. Gambardella. Markets for technology: The economics of innovation and corporate strategy. MIT Press, 2001.

J. Baron, Y. Meniere, and T. Pohlmann. Standards, consortia and innovation. International Journal of Industrial Organization, forthcoming, 2013.

J. Bessen and M. J. Meurer. The direct costs from npe disputes. Cornell Law Review, 99:387, 2014.

D.J. Birch and E. Burnett-Kant. Unbundling the unbundled. The McKinsey Quar- terly, 4:103–111, 2001.

K. Blind, R. Bekkers, Y. Dietrich, E. Iversen, F. K¨ohler,B. M¨uller,T. Pohlmann, S. Smeets, and J. Verweijen. Study on the interplay between standards and intellectual property rights (iprs). Report, European Commission Tender No ENTR/09/015 (OJEU S136 of 18/07/2009), 2011.

E. Bohlin, K. Brodin, A. Lundgren, and B. Torngren. Convergence on Communica- tions and Beyond. Amsterdam: North Holland, 2000.

M. Boldrin and D. K. Levin. The case against patents. Journal of Economic Per- spectives, 27:322, 2013.

S. Chan-Olmsted and M. Jamison. Rivalry through alliances:: Competitive strategy in the global telecommunications market. European Management Journal, 19(3): 317 – 331, 2001. ISSN 0263-2373.

Cisco. Cisco visual networking index: Global mobile data traﬃc forecast update, 2012-2017, February 2013. URL http://www.cisco.com/en/US/solutions/ collateral/ns341/ns525/ns537/ns705/ns827/white_paper_c11-520862. html. accessed December 2013.

CMT. Informe annual 2012, 2012. URL http://cmtdata.cmt.es/cmtdata/jsp/ inf_anual.jsp?tipo=1. accessed December 2013.

I.M. Cockburn and M.J. MacGarvie. Entry and patenting in the software industry. Management Science, 57:915933, 2011.

I.M. Cockburn, M.J. MacGarvie, and E. M¨uller. Patent thickets, licensing and innovative performance. Industrial and Corporate Change, 19:899–925, 2010.

25 W.M. Cohen, R.R. Nelson, and J.P. Walsh. Protecting their intellectual assets: Ap- propriability conditions and why u.s. manufacturing ﬁrms patent (or not). NBER Working Paper No. 7552, 2000.

C. Cotropia, J. P. Kesan, and D. L. Schwartz. Patent holder and litigation dataset, 2012. URL http://npedata.com/. last updated Jan. 15, 2014. K. Cremers, M. Ernicke, F. Gaessler, D. Harhoﬀ, C. Helmers, L. McDonagh, P. Schliessler, and N. van Zeebroeck. Patent Litigation in Europe. Center for European Economic Research – Discussion Paper No. 13-072, 2013.

A.C. Davies. The Digital Divide: A Political Economy of the Restructuring of Telecommunications. PhD thesis, University of Sussex, Brighton, 1991. PhD Thesis, University of Sussex, Brighton.

M. Eberhardt, C. Helmers, and Z. Yu. Is the dragon learning to ﬂy? an analysis of the chinese patent explosion. Working paper, 2011.

EC. Digital agenda scoreboard, 2013. URL http://ec.europa.eu/ digital-agenda/sites/digital-agenda/files/DAE%20SCOREBOARD%202013% 20-%20SWD%202013%20217%20FINAL.pdf. accessed December 2013.

ETNO. Third annual economic report, November 2012. URL http://www.etno.be/ datas/publications/economic-reports/etno-economic-report-2012.pdf. accessed December 2013.

D. Fertig, C. Prince, and D. Walrod. What kind of telco is the fairest of them all? The McKinsey Quarterly, 4:144–148, 1999.

C. Fink, M. Khan, and H. Zhou. Exploring the worldwide patent surge. WIPO Economic Research Working Paper No. 12, 2013.

M. Fransman. Analysing the evolution of industry: The relevance of the telecommunications industry. Economics of Innovation and New Technology, 10(2-3): 109–140, 2001.

FT. Drop in texting heralds industry shift, January 2012. URL http://www.ft. com/cms/s/0/3d976108-3f77-11e1-ad6a-00144feab49a.html. accessed De- cember 2013.

J. L. Funk. The emerging value network in the mobile phone industry:the case of japan and its implications for the rest of the world. Telecommunication Policy, 33(1-2):4–18, 2009.

C. Greenhalgh, J. Philips, R. Pitkethly, M. Rogers, and J. Tomalin. Intellectual property enforcement in smaller uk ﬁrms. Report, Strategy Advisory Board for Intellectual Property Policy (SABIP), 2010.

A. Hagiu and D. B. Yoﬃe. The new patent intermediaries: Platforms, defensive aggregators, and super-aggregators. Journal of Economic Perspectives, 27(1):45– 66, 2013.

26 B. Hall, C. Helmers G. von Graevenitz, and C. Rosazza-Bondibene. Technology entry in the presence of patent thickets. Working Paper, 2013.

B.H. Hall and R. H. Ziedonis. The patent paradox revisited: an empirical study of patenting in the us semiconductor industry, 1979-1995. RAND Journal of Economics, 32(1):101–128, 2001.

J. Isern and M.I. Rios. Facing discinnection: Hard choices for europes telcos. The McKinsey Quarterly, 1:82–90, 2002.

F. Jell and J. Henkel. Patent portfolio races in concentrated markets for technology. Available at SSRN: http://ssrn.com/abstract=1673106, 2014.

B. Kang and R. Bekkers. Just-in-time inventions and the development of standards: How ﬁrms use opportunistic strategies to obtain standard-essential patents (seps). Eindhoven Centre for Innovation Studies Working Paper 13.01, 2013.

C. Kongaut and E. Bohlin. Impacts of mobile termination rates (mtrs) on retail prices: The implication for regulators. In 23rd European Regional Conference of the International Telecommunication Society, Vienna, Austria, 1-4 July 2012, 2012. URL http://www.econstor.eu/bitstream/10419/60348/1/720230160. pdf.

S. Kortum and J. Lerner. What is behind the recent surge in patenting? Research Policy, 28:1–22, 1999.

F. Li and J. Whalley. Deconstruction of the telecommunications industry: from value chains to value networks. Telecommunication Policy, 26(9-10):451–472, 2002.

T. J. Lybbert and N. J. Zolas. Getting patents and economic data to speak to each other: An ‘algorithmic links with probabilities’ approach for joint analyses of patenting and economic activity. Research Policy, 2013.

Microsoft. Microsoft oﬃcially welcomes skype, October 2011. URL http://www. microsoft.com/en-us/news/press/2011/oct11/10-13skypepr.aspx. accessed December 2013.

P. Olla and N.V. Patel. A value chain model for mobile data service providers. Telecommunications Policy, 26(9-10):551 – 571, 2002. ISSN 0308-5961.

Y. Ozcan and S. Greenstein. Composition of innovative activity in ict equipment r&d. Loy. U. Chi. LJ, 45:479–479, 2013.

C. Palmberg and O. Martikainen. Diversiﬁcation in response to ict convergence indigenous capabilities versus r&d alliances of the ﬁnnish telecom industry. Info, 8(4):67–84, 2006.

P. P´eladeau,K. Sabbagh, B. El-Darwiche, and R. Friedrich. Riding the data tsunami a paradigm shift for telecom operators, June 2010. URL http://www.booz.com/ media/file/Riding_the_Data_Tsunami.pdf. accesses December 2013.

27 J. Pennings, H.L. van Kranenburg, and J. Hagedoorn. Past, present and future of the telecommunications industry. Research Memoranda 015, Maastricht : METEOR, Maastricht Research School of Economics of Technology and Organization, 2005.

J. Peppard and A. Rylander. From value chain to value network:: Insights for mobile operators. European Management Journal, 24(2-3):128–141, 2006.

J.G. Seaberg, J. Hawn, G.E. Dinceler, C.C. Eugster, and N. Rao. Attackers versus incumbents: The battle for value in an ip-networked world. The McKinsey Quarterly, 4:138–153, 1997.

C. Shapiro. Navigating the patent thicket: Cross licenses, patent pools, and standard setting. In A. Jaﬀe, J. Lerner, and S. Stern, editors, Innovation policy and the economy. NBER, MIT Press, Cambridge, MA, 2001.

R. Siebert and G. von Graevenitz. Are licensing agreements appropriate instruments to cut through the patent thicket? Working Paper, 2011.

D. Steinbock. Globalization of wireless value system: from geographic to strategic advantages. Telecommunications Policy, 27(3-4):207–235, 2003.

Telef´onica.What is ‘telef´onicadigital’?, 2012. URL http://www.telefonica.com/ en/digital/html/about_telefonica_digital/home.shtml. accessed Decem- ber 2013.

D. Tilson and K. Lyytinen. The 3g transition: Changes in the us wireless industry. Telecommunications Policy, 30(10-11):569 – 586, 2006. ISSN 0308-5961.

H.L. van Kranenburg and J. Hagedoorn. Strategic focus of incumbents in the european telecommunications industry: The cases of bt, deutsche telekom and kpn. Telecommunications Policy, 32(2):116 – 130, 2008. ISSN 0308-5961.

G. von Graevenitz, S. Wagner, and D. Harhoﬀ. How to measure patent thickets – a novel approach. Economics Letters, 111:14, 2011.

L. Waverman and F. Trillas. Corporate control and industry structure in global communications: an introduction. Telecommunications Policy, 26(5-6):219 – 224, 2002. ISSN 0308-5961.

28 APPENDIX A Appendix Tables

Observations IPC4 Mean StdDev Min Max

# patent applications 2205 401 22.99 71.47 1.00 1787.00

# litigation cases 2205 401 13.26 30.13 0.00 229.00

# standard essential patents 2205 401 195.60 511.74 0.00 3631.00

# licensing agreemeents (adjusted) 2205 401 3.71 7.38 0.00 35.96

NPEs patent stock 2205 401 314.55 470.85 0.00 2435.00

# triples 2205 401 224.14 510.28 0.00 2553.00

Assets (in ebn) 2205 401 121.65 84.66 1.73 278.25

R&D intensity 2205 401 0.75 0.81 0.00 6.69

Table A-1: Descriptive Statistics

i Four-digit IPC class Number of patent applications Share assigned to parent

H04L 7462 0.84 H04W 4606 0.67 G06F 2799 0.84 H04M 2658 0.72 H04N 1529 0.76 H04B 1338 0.86 H04Q 927 0.69 G06Q 924 0.71 G10L 799 0.95 G06K 438 0.83 H04J 415 0.97 G07F 382 0.71 G06T 320 0.91 H03M 273 0.94 G01S 251 0.75 G08G 216 0.72 B41J 210 0.87 H04H 193 0.72 G07C 145 0.83 H01Q 144 0.82

Table A-2: Parent as assignment mode per four-digit IPC class (2001-2008)