Shedding light on the black hole

The roll-out of broadband access networks by

private operators

Marieke L. Fijnvandraat

Shedding light on the black hole

The roll-out of broadband access networks by

private operators

Proefschrift

Ter verkrijging van de graad van doctor

aan de Technische Universiteit Delft,

op gezag van Rector Magnificus prof.dr.ir. J.T Fokkema,

voorzitter van het College voor Promoties,

in het openbaar te verdedigen op 16 januari 2008 om 10:00

door Maria Leonie Fijnvandraat

Ingenieur in de technische bestuurskunde

geboren te Leeuwarden

Dit proefschrift is goedgekeurd door de promotor:

Prof.dr. M.L. Mueller

Copromotor: Dr. W.A.G.A. Bouwman

Samenstelling promotiecommissie:

Rector Magnificus, Voorzitter

Prof.dr. M. L. Mueller, Technische Universiteit Delft, promotor

Dr. W.A.G.A. Bouwman, Technische Universiteit Delft,

Copromotor

Prof.dr.ir. W.A.H. Thissen, Technische Universiteit Delft

Prof.dr.ir. N.H.G. Baken, Technische Universiteit Delft

Prof. dr. ir. L.J.M. Nieuwenhuis, Universiteit Twente

Dr. K. Stordahl, MsC, Telenor Networks

Dr. B.Preissl, Wirtschaftsdienst ZBW

Prof.ir. W. Dik, Technische Universiteit Delft,

Reservelid

ISBN 978-90-79878-03-1

Printing: Gildeprint Drukkerijen, Enschede

Published and distributed by: Next Generation Infrastructures Foundation

P.O. Box 5015, 2600 GA Delft, the Netherlands

T: +31 15 278 2564

F: +31 15 278 2563

E: info@nginfra.nl

I: www.nginfra.nl

This research was funded by the Next Generation Infrastructures Foundation

programme.

Keywords: broadband technology, access networks, strategic decision-making,

evolutionary paths, multidisciplinary, risk and uncertainty.

Cover photo: The Circinus Dwarf Galaxy (a supermassive black hole, 13 million

lightyears away from Earth) Credit: NASA/AURA/STSCI

Copyright © 2008. M.L. Fijnvandraat. All rights reserved. No part of this

publication may be reproduced, stored in a retrieval system, or transmitted in any

form of by any means, electronic, mechanical, photocopying, recording, or

otherwise, without the prior permission in writing from the copyright owner.

Printed in the Netherlands

In memoriam Prof.dr.René Wagenaar 1954-2007

Voor mijn lieve oma

***

Acknowledgement Many people have contributed to the realisation of my dissertation. The first person I want to mention is my copromotor. Harry, we worked together for almost a decade now! Fortunately, we still have to write some papers, so we don’t have to end our working relationship abruptly. I want to thank you for all your suggestions, good ideas, and your help to get my dissertation finished. There were moments I saw the broadband market evolving so quickly that I had the feeling of being caught up by it, which, sometimes, made me questioning myself why I was actually doing research in this field. During these moments, you always convinced me of the added value of my work and really stimulated me to go on with the job! You moreover always stimulated me to challenge myself to aim at a higher quality level. I moreover want to thank you for always being there for me and for your interest and support on a more personal level. I will most definitely use that fountain-pen! I also want to thank my professor Milton Mueller. Although we have been working together for only a short time, you really helped me fitting all the pieces of my complex work together and helped me to create a coherent story. Working in the ICT section of our faculty during all those years always has been a pleasure to me! The nice atmosphere, the openness, the willingness to discuss all kinds of subjects and all the advices regarding my PhD made that I really enjoyed working there. During the years of working on my research I also experienced much support from friends and family. Mum, dad, thanks for always believing in me no matter what. I also want to thank my brother Arnoud, for always hearing me out, cheering me up and for showing me in person that ‘promotie dips’, as you experienced yourself during your own PhD, can be overcome and will eventually lead to good results! I furthermore want to thank all my close friends for being there, especially during the last half year of my PhD. Geertje, it was so nice to have you as a good friend at the faculty, working on your PhD as well, close by for chats, lunch and coffee breaks! And then some words for my two nimfjes, Dille and Karianne. Dillemeisje, although we are very different in some ways, we are so much alike and I want to thank you for all the laughs we enjoyed together (other people a little bit less but we don’t care) but also for your support during some difficult periods. I am really curious about the way you will appear on the day of my defense, considering the fact that your first reaction to my question whether you would like to be my paranimf was: “O, my god, but what do I have to wear then?!” I am sure you will look fine although I will not – as you hoped for- buy that expensive new suit for you!

And last, but most definitely not least, my close friend Karianne. Kari, when I had to write down all the good and bad moments we shared, this would result in a book much thicker than this dissertation! You always were a great support and good partner in crime to talk about PhD frustrations but also a close friend to celebrate all PhD highlights like accepted papers, finished chapters and Committee approval. Although our research subjects were so closely related, we managed to visit a conference together only once (and this was when you already had finished your PhD) and we never managed to write a paper together! As a final attempt I now added this as a recommendation for further research. Who knows, maybe somehow, sometime, we will find the time to write this article together but we most likely will just talk about it during one of our evenings in Lust or Square. Thnx, LS!! The good thing of several years of hard working during a PhD project is the feeling of working towards the moment that you finally have managed to do it. Fitting all the pieces together, overcoming difficult moments, and then, noticing that everything more and more becomes clear. Now, writing down the last sentences of my dissertation, a feeling of relaxation, relieve and self-conquest comes over me. On the other hand, however, a feeling of melancholy comes over me as well at this moment, because it also means the end of an era of working towards this final, so much desired result. It is closely comparable to the feeling you experience when coming closer to the end of a good, thick novel that has captivated you for weeks or even months, which, after having read the last pages, makes you leafing back to experience this beautiful end once again… But now the work is finally done. I have written my last words and I will start my new ‘novel’. Marieke Fijnvandraat December 2008

Table of contents Chapter 1 Introduction.................................................................................................17

1.1 Background and relevance ................................................................................17 1.2 Research objective, research questions ............................................................25 1.3 Thesis outline ....................................................................................................30 1.4 References ........................................................................................................32

Chapter 2 Domain description ....................................................................................35

2.1 Three perspectives on broadband in the local loop ...........................................35 2.2 The Dutch broadband market ............................................................................39

2.2.1 Supply side ...................................................................................................39 2.2.2 Demand side.................................................................................................50 2.2.3 Conclusion ....................................................................................................51

2.3 Broadband technology.......................................................................................52 2.3.1 Copper-based broadband technologies ........................................................54 2.3.2 Cable-based broadband technologies ..........................................................57 2.3.3 Fiber-based technologies..............................................................................60 2.3.4 Fixed –wireless broadband technologies ......................................................62 2.3.5 Analysis ........................................................................................................68 2.3.6 Evolutionary paths ........................................................................................71 2.3.7 Discussion and conclusion............................................................................77

2.4 The policy & regulation perspective: stimulating and regulating broadband roll-out ..........................................................................................................................78

2.4.1 Introduction ...................................................................................................78 2.4.2 The 2002 Telecommunications Act...............................................................81 2.4.3 A second way of public intervention in the broadband market: national and European stimulation programmes .............................................................................88 2.4.4 Action plans to stimulate broadband roll-out .................................................88

2.5 Conclusion.........................................................................................................96 2.6 References ........................................................................................................99

Chapter 3 Meta-analysis –the development of a conceptual model for broadband roll-out ..........................................................................................................107

3.1 Methodology ....................................................................................................107 3.1.1 Meta-Analysis .............................................................................................107 3.1.2 Content Analysis .........................................................................................109 3.1.3 Network analysis.........................................................................................112

3.2 Results: Broadband roll-out- a black hole........................................................113 3.2.1 Towards a conceptual model for broadband roll-out ...................................115

3.3 Conclusion.......................................................................................................119 3.4 References ......................................................................................................121

Chapter 4 Infrastructural investments: dealing with uncertainty and irreversibility.. ..........................................................................................................123

4.1 Financial methods to assess business value of investments...........................124 4.1.1 Discounted cash flow methods ...................................................................124 4.1.2 Ratio methods.............................................................................................125

4.2 From financial options to ‘real options’ ............................................................126 4.3 Types of real options .......................................................................................126 4.4 Risk and uncertainty ........................................................................................128

4.4.1 Uncertainty..................................................................................................128 4.4.2 Risk.............................................................................................................132 4.4.3 A typology framework for risk & uncertainty................................................133

4.5 Reducing risk and uncertainty .........................................................................135 4.6 Conclusion.......................................................................................................136 4.7 References ......................................................................................................138

Chapter 5 Strategic decision-making in a complex, dynamic and uncertain environment ..........................................................................................................143

5.1 The way towards a decision: the decision-making process .............................148 5.2 Describing the broadband market and explaining the realisation of decisions: the Garbage Can .................................................................................................................150

5.2.1 Telecom companies as organised anarchies..............................................150 5.2.2 Decision-making within the garbage ban ....................................................151

5.3 Logical Incrementalism....................................................................................152 5.4 Conclusions .....................................................................................................154 5.5 References ......................................................................................................156

Chapter 6 Exploring the rationale behind broadband roll-out- Qualitative analysis .. ..........................................................................................................159

6.1 Method.............................................................................................................160 6.2 The content-related side of broadband roll-out: the impact of technology, market and policy.......................................................................................................................164

6.2.1 Technology .................................................................................................164 6.2.2 Market.........................................................................................................167 6.2.3 Policy and regulation...................................................................................170

6.3 Risk & uncertainty............................................................................................173 6.3.1 Sources of risk and uncertainty...................................................................173 6.3.2 Effects of risk & uncertainty ........................................................................180 6.3.3 Dealing with risk & uncertainty ....................................................................182

6.4 The decision-making process on broadband roll-out .......................................184 6.4.1 Who’s in charge? ........................................................................................184 6.4.2 The duration of decision-making and roll-out ..............................................186 6.4.3 Cooperation in the value chain ...................................................................189 6.4.4 Financial assessment methods...................................................................191 6.4.5 Evolutionary or revolutionary upgrades.......................................................193

6.5 Applying decision-making theory to broadband local loop upgrades ...............197 6.5.1 The Garbage Can: criteria for organisational anarchies..............................198 6.5.2 The Garbage Can; applicability of streams and decision windows .............201 6.5.3 Logical Incrementalism: decisions are made in small steps to keep options open for feedback and adjustments ..........................................................................204

6.6 Concept validation as a first step in model validation ......................................207 6.6.1 The four-phase base model ........................................................................207 6.6.2 Phase 1: Decision-making process to upgrade/roll-out broadband.............209 6.6.3 Phase 2: Network upgrade/roll-out..............................................................217 6.6.4 Phase 3: End-user demand & adoption ......................................................219 6.6.5 Phase 4: Revenues.....................................................................................221 6.6.6 Analysis ......................................................................................................222

6.7 Conclusions .....................................................................................................225 6.8 References ......................................................................................................230

Chapter 7 Towards an explanatory parsimonious model- quantitative analysis..233

7.1 Exploring the field: national pre-test ................................................................234 7.1.1 Method........................................................................................................234 7.1.2 Development of scales................................................................................235

7.2 International survey .........................................................................................238 7.2.1 Method........................................................................................................238 7.2.2 Construction of scales.................................................................................241

7.3 Exploring broadband roll-out: descriptive results .............................................243

7.3.1 The content-related part: Factors influencing the decision to upgrade or roll-out broadband...........................................................................................................243 7.3.2 Risk and uncertainty ...................................................................................251 7.3.3 The process-related side of network upgrades ...........................................254

7.4 Towards a parsimonious explanatory model for broadband roll-out ................266 7.5 Limitations and discussion...............................................................................274 7.6 Conclusions .....................................................................................................275 7.7 References ......................................................................................................278

Chapter 8 Conclusion Shedding light on the black hole .....................................279

8.1 Introduction......................................................................................................279 8.2 Overview of the research.................................................................................279 8.3 Answering the research questions...................................................................285

8.3.1 The impact of technological, market-related and regulatory factors on broadband roll-out .....................................................................................................285 8.3.2 Exploring available multidisciplinary frameworks on broadband roll-out .....290 8.3.3 Assessing investment options under risk and uncertainty...........................291 8.3.4 Dealing with risk and uncertainty in investment-related decisions ..............293 8.3.5 The course of the decision-making process................................................297 8.3.6 Towards a parsimonious model for broadband-related decision-making ....299 8.3.7 Incremental decision-making is a logic response to uncertain interrelation between factors from different domains ....................................................................301

8.4 Limitations of this research and recommendations for further research ..........303 8.5 Research Contributions ...................................................................................306

8.5.1 Implications for scientists ............................................................................306 8.5.2 Implications for strategic decision-makers ..................................................308 8.5.3 Implications for policy-makers.....................................................................309

8.6 References ......................................................................................................313 Annexes ....................................................................................................................317 Annex A Numeric Tables Factor analysis quantitative results Pre-test ..............319 Annex B Numeric tables descriptive results quantitative analysis main study..321 Annex C Numeric tables confirmatory factor analysis & structural equation modelling ....................................................................................................................329 Annex D Questionnaire............................................................................................335 Annex E Articles and papers included in the Meta-analysis ................................351 Nederlandse samenvatting ..............................................................................................355

17

Chapter 1 Introduction

1.1 Background and relevance

The Netherlands are among the leading countries in the world when it comes to DSL and Cable connections (TNO, 2007). According to OECD studies, in June 2007, the Netherlands occupied a second position among OECD countries in terms of broadband penetration, including ADSL and cable subscriptions. Of the entire Dutch population, 59% have ADSL connections and 41% cable connections (TNO, 2007). Although market reports refer to these connections as ‘broadband’, in this study they are not considered real ‘broadband’ connections. Before proceeding, we provide the definition of the term broadband used in this study.

Defining broadband The term ‘broadband’ is a relative rather than a fixed concept. It is based on the actual availability of bandwidth at a certain place at a certain time. Because the available bandwidth changes over time, broadband means different things at different times. In addition, different countries use different definitions (CPB, 2005). For these reasons, the term ‘broadband’ represents a relatively vague concept that can be defined from technological viewpoint and from an economic perspective. Technically, the term broadband usually refers to a specific amount of bandwidth that can be transmitted over a network towards an end-user. Over the last ten years, transmission speeds have increased considerably, which means that the definition of broadband has changed. From an economic point of view, it is possible to take a different perspective and see broadband as an investment that can be used to create added value. Investments in broadband technology include investments in fixed assets like routers, modems, cables and software. Transporting information from one point to another creates added value through services like teleworking, e-learning and e-health. Thus, from an economic perspective, broadband can be seen as a way of generating economic profit (CPB, 2005). However, in order to establish a fixed point of reference, it makes sense to use some sort of working definition related to bandwidth. For this reason we define the concept of broadband as follows:

A connection of at least 10 Mbps symmetrical bandwidth in the local loop

A connection of this quality is assumed to be suitable for today’s available streaming video and audio and high quality online data-sharing applications, which at the moment are among the major emerging applications. Examples of popular applications are YouTube and Rapidshare, which rated 3rd and 12th worldwide in terms of traffic in July 2008. YouTube is a worldwide website that allows people to upload, tag and share videos, while Rapidshare enables users to upload up to 100 MB files for sharing.

18

Although transmission speeds in access networks have increased significantly in recent years, a symmetrical data connection of 10 Mbps is not yet available in the Dutch residential market. In 2007, the predominant download speed in the Netherlands was 1 to 2 Mbps (European Commission, 2008). Although most people have upgraded their Internet connection since 2003 (Dialogic, 2005), research among broadband users (Dialogic, 2007; Vermaas, 2007) shows that customers in the Netherlands do not use high-speed broadband capacities, which means that the Netherlands lag behind when it comes to the roll-out of real broadband technology like optical fiber networks. This is considered a problem by national as well as European governments, because broadband roll-out is expected to be important in economic as well as social terms. In eEurope 2002, the European Commission stated that “Broadband enabled communication, in

combination with convergence of technologies and services will bring social as well

as economic benefits. It will contribute to e-inclusion, cohesion and cultural

diversity” (European Commission, 2002, p.8). Below, we discuss both the economic and social relevance of broadband in greater detail.

Economic relevance of broadband “It is widely recognised that the benefits of broadband in terms of lower transportation costs can be large for an economy in terms of productivity” (CPB, 2005, p.9). “The telecommunication industry, and in particular the broadband market, is changing rapidly, providing customers with more options to exchange sound, video and data. These communication streams are transmitted along networks that become faster, better and cheaper” (CPB, 2005, p.9). Several market studies (a.o. Gartner, 2002; Criterion Economics, 2003; CEBR, 2003; The Allen Consulting group, 2003; ACIL Tasman, 2004)1 underline the positive relationship between broadband and economic performance. The sources of the economic benefits of broadband are both direct and indirect in nature (Telecommunication Industry Association, 2003, p.6; Price Waterhouse Coopers, 2004). Increased productivity, job creation, increases in wages and efficiency in time and money, the creation of new or spin-off industries and an increase in the demand for computer broadband equipment are significant benefits of the use of broadband technologies. Also, the Netherlands Bureau for Economic Policy Analysis (CPB) states that broadband can stimulate productivity through improvements in production processes, reductions in transaction costs and innovations like new service applications via broadband infrastructures (i.e. content) (CPB, 2005).

1 http://www.scotland.gov.uk/Publications/2007/01/09153006/6

19

In addition to these direct effects, the economic benefits of broadband can also be attributed to indirect factors like increased e-commerce, reductions in commuting, increased consumption of entertainment and savings in healthcare (Macklin, 2002), as well as increased efficiency in the distribution of goods, services and information. Social relevance of broadband Broadband is also very relevant from a social point of view. It can be used to improve the quality of a country’s education and health services, bring government closer to society, and provide jobs and prosperity (Firth & Mellor, 2005). Furthermore, broadband creates a favourable environment for private investment and for the creation of new jobs, which in turn will help boost productivity and modernise public services (European Commission, 2002). Because broadband telecommunication is seen as a source of social and economic gains, the European Union (EU) and other regions are encouraging the deployment of a secure broadband infrastructure (CPB, 2005, p.9). The common objective shared by various EU member states is to accelerate broadband deployment. In specific terms, the EU will support the upgrade and efficiency of technology for optical fiber access networks (European Commission, 2002). In i2010, the eEurope report succeeding eEurope 2005, the development of broadband networks and services is again given a high priority. The main objective for i2010 is defined as “(a) Single European Information Space offering affordable and secure high bandwidth communications, rich and diverse content and digital services” (European Commission, 2005, p.5).

Private investments in broadband local loop roll-out remain below expectations

In spite of the positive expectations with regard to the social and financial benefits of broadband, private investments in fiber local loop networks and broadband applications are still limited, due to the fact that realising broadband roll-out involves high-risk infrastructural investments. First of all, this has to do with the general characteristics of major infrastructural projects, which make the roll-out of broadband networks extremely costly to develop and require high initial investments (de Vlaam en de Jong, 2002). In addition to the generic characteristics of infrastructures, access networks also have specific characteristics. Both aspects are discussed below in greater detail. Specific characteristics of broadband infrastructure Although infrastructures are officially designed to satisfy specific social needs, in fact they shape social change in a much broader and more complex way. Telecommunication infrastructures enable us to communicate and exchange information without being troubled by distance. The roll-out of telecommunication networks is comparable to the

20

construction of railways, roads and electricity networks, which all share some of the same characteristics. Investments in infrastructures involve high levels of risk because they are extremely costly to develop and require high initial investments, which can only be earned back, if ever, after many years (De Vlaam en De Jong, 2002) because of high initial fixed costs and long construction times. Telecommunication networks have a number of the characteristics of natural monopolies, like indivisibility. The indivisibility of transport infrastructure facilities implies that it is often inefficient to have more than one local supplier of the service. Also, several facilities may be essential to reach customers and/or competitors and access complementary markets. As a result of these characteristics, new entrants experience high entry and exit thresholds (NZier, 2004). For these reasons, infrastructural investments are among the most complex, delicate and risky decisions. They are characterised by high and often unpredictable cost-of-ownership and their benefits are intangible in nature (Renkema, 2000). The access network: an extra dimension of complexity On top of generic characteristics of infrastructures, access networks have a number of specific characteristics. Other terms used for describing the access network are ‘last mile’, ‘first mile’ or ‘local loop’. The European Commission defines the local loop as “the physical circuit connecting the network termination point at the subscriber's premises to the main distribution frame or equivalent facility in the fixed public telephone network” (2002/19/EC, Art2 sub e). The network behind the local loop is known as the ‘backbone’ or ‘core network’. Core networks consist of high capacity optical fiber cable and are able to transmit high amounts of data at very high speeds. The local loop, on the other hand, consists of other, ‘traditional’, transmission media, like copper twisted pair and coaxial cable, which allow for much lower transmission speeds than optical fiber. For this reason, the local loop is the ‘bottle neck’ with regard to data transmission in the entire network. It can be imagined as a narrow section of a highway that still has to handle all the traffic, almost inevitably causing traffic jams. The same is true when it comes to for the transition from the core network to the access network. Downloading high amounts of data, for example streaming video, within a reasonable time-frame, demands high transmission speeds in the core network as well as large transmission capacity in the local loop. For that reason, the technology deployed in the local loop needs to be upgraded to technologies that provide higher upload and download speeds. The fact that the local loop connects the local switch to people’s individual homes means that all investments that are needed to upgrade this part of the network to allow for higher transmission speeds must be earned back from the money generated by service and access provisioning to individual households. The costs involved include the costs associated with digging, cables, network equipment and network maintenance. In cases where several network operators compete with each other, the revenues earned from access and services from the individual households

21

must be divided among the number of operators, which makes the available returns on investments even smaller. Critical mass and the chicken-egg problem In addition to these infrastructure-related factors, there are several other problems associated with investing in infrastructure. Infrastructural projects typically do not deliver benefits as such, but rather generate business value through, for example, facilitating innovations. In other words, most of the time social as well as economic benefits do not end up where investments are made (Expertgroep Breedband, 2002). This problem is clearly a case of the chicken or the egg: useful content will not be developed as long as the infrastructure (high-speed broadband networks) is not available, while suppliers will not invest in infrastructure when it is not clear whether or not there will be useful content. Moreover, these investments are characterised by far-reaching consequences in terms of business efficiency, effectiveness, competitive positioning and innovate capacity (Renkema, 2000, p.5).

The demand side In addition to the factors outlined above, difficulties with regard to infrastructural investments can be found in the demand side of the market. Operators cannot predict with certainty what their returns will be, because the adoption and usage of new technologies are not constant, but evolve over time. Patterns of technology adoption and diffusion are never fully predictable (Vermaas, 2007, p215). We will now look at end-user demand-related factors. Market demand is an important condition for continuous investment in access networks. End-users are, however, highly unpredictable and for that reason, it is extremely difficult to estimate future demand. Increased system complexity makes operators more dependent on equipment suppliers. This creates uncertainty for network providers about future revenues and makes it difficult to plan long-term investments in access networks. The Diffusion of Innovation theory (DoI) argues that the adoption of (new) media by later users depends on the presence of other users. Only where there are sufficient users new users will join in (Van de Wijngaert, 2001). The costs of adopting an innovation at the beginning of the adoption cycle outweigh the benefits for the adopters involved. This group of initial investors plays an important role in the development of a technology. The moment that as many initial investors have adopted the technology that it becomes also worthwhile for the other members of a social system to invest, the critical mass has been reached (Bouwman et al. 2005). The importance of reaching a critical mass was already recognised by the European Commission in 1993, which applied diffusion theory in its “White Paper on growth, competitiveness and employment: The challenges and ways forward into the 21st century” (COM(93)700, 05.12.93). It was written in order to stimulate broadband roll-out to realize the social and cultural benefits and calls for “the development of services and applications so as to attain a critical mass sufficient

22

for further infrastructural investments and for attracting new users adding to the value of a given network” (RAND Europe, 2003, p.3). Also Weening (2006) demonstrated that the combination of the ‘chicken-egg’ problem and operators not being able to reach the critical mass is problematic and causes the delay in the roll-out of fiber infrastructure in the Netherlands. In her study on Smart Cities, she concluded about the Dutch fiber pilot project ‘Kenniswijk’: “within Kenniswijk, services and end-user provisions have been realised, but there is barely any infrastructure” (Weening, 2006, p.99). The chicken-egg problem could not be solved. Problems were mainly related to difficulties concerning infrastructure roll-out and insufficient critical mass” (Weening, 2006, p.99-100) .

Additional complexity is caused by a multi-actor, multi-disciplinary environment

In addition to insufficient demand and the complexity caused by the characteristics of infrastructure in general and more in particular by the access network, additional complexity for network operators is caused by the many actors and their individual objectives and interests and the multidisciplinary environment in which they operate. Below, we provide insight into these additional complexities in greater detail.

Multi-actor environment Different actors have different views on how best to approach the development towards next generation broadband networks. Network providers, equipment suppliers, service operators, wholesale and residential end-users can all be identified as market players in this respect. The current telecom market is a highly competitive one. The situation is made extra complex by the interdependency between the actors involved. All the players in the market influence one another: on the one hand they depend on each other, while on the other hand they have power over each other, resulting in a complex multi-actor network. Greater insight is needed in the interrelationships between these parties and the way they influence the local loop upgrade strategy of network operators. Moreover, the market for broadband is changing quickly and continuously. Typically, traditional network operators have turned into triple-play providers, offering telephony, Internet access and television, which has significantly increased the complexity in the market in which they operate. Network and service operators no longer serving separate markets, but are now tightly interwoven (convergence). These changes in the market and the position of the various organisations within the value chain consequently have an impact on their internal business processes. Not only departments are changed or added when new services are offered, but logistics, billing systems, marketing and supply processes must also be adapted to this new and changed position within the value chain.

23

Multi-disciplinary environment In addition to having to deal with competitors, telecom operators face a range of significant strategic decisions in various domains. In the technological domain, they have to find their way with regard to the many technological possibilities available and the various migration paths towards the interactive broadband future. As far as upgrading their infrastructure network is concerned, operators in the telecom market are faced with strategic choices. To anticipate future increases in demand and keep up with their competitors, operators have to use existing as well as alternative transmission media. Technological developments take place in rapid succession. The life cycles of most broadband access technologies are very short, which makes for a highly dynamic market. There is a broad range of technological options, including upgrading cable connections, implementing new technologies for copper, installing optical fiber networks, roll-out wireless networks or opt in favour of hybrid solutions. The attractiveness of these technologies is determined by how they score with regard to relevant attributes, not only in terms of transmission capacity, but also with regard to reach, compatibility, standardisation, maturity, and technological and geographical scalability. Greater insight is needed into the relevant characteristics of the various available broadband local loop technologies, for instance their future-proofness, transmission capacity, upgrading costs, etc. Moreover, insight is needed into the possible evolutionary paths from the current state of broadband networks towards connections providing 10 Mbps symmetrical and higher bandwidth speeds. In the economic domain, operators have to choose between several investment methods and to deal with many, often large, investments, ROI timeframes and uncertain revenues. In the policy and regulation domain, they have to deal with regulatory regimes and public broadband initiatives, which adds competition to the private market. To summarise, we can conclude that operators are still struggling with problems that have to do with a lack of service development and adoption (the chicken egg problem) and not being able to reach the critical mass. These problems are related to the cost sensitivity of the access network segment and the high risks and uncertainties in the residential broadband market (Ims et al., 1998). Furthermore, market response, pricing strategies, uncertainty handling, strategic interaction between competitors and external effects influence the economics of telecommunication operator investment projects (Eurescom, 1999). On top of that, telecom operators find themselves in fast changing business environments. Convergence between services and infrastructures has resulted in a much broader and complex playing field with more competitors. New technologies and end-user services are developed at a high pace, which leads to more and more complex technological migration paths. Moreover, consumers are becoming more demanding when it comes to bandwidth, new services and service quality. Telecom operators must take all these factors into account when making strategic decisions

24

regarding where, how and when to invest if they are to make money in what is a highly competitive market. They must choose their market strategy (shaper or quick follower) and make sure they provide the latest technologies to prevent customers from switching to the competition (churning). As a result, although the social and economic benefits of broadband have been recognised by many research institutes as well as by the European Commission, and despite the fact that various policy initiatives have been introduced to stimulate the roll-out of fiber network, these networks have still not been rolled out on a large scale by private operators. It may be clear that the existing national and European policies have not yielded the expected and desired effects. For that reason, it is necessary to take a closer look at the content and processes involved in the roll-out of broadband access (local loop) networks. To gain insight into the content-related side involves an examination of the most important drivers and outcomes of broadband network roll-out from the point of view of private operators. Insight into the process-related side is gained by taking a closer look at the process involved in deciding whether or not to roll-out certain technologies in the local loop. Knowledge in these two areas will provide policy-makers with more options to deploy their policy instruments successfully and stimulate fiber network roll-out successfully.

Research unit This study focuses on the roll-out of broadband networks in the local loop from the point of view of telecom and cable operators that have their own access networks. The unit of research is the behaviour of telecom providers (or telecom network operators) with regard to the roll-out and upgrades of their access networks. This research examines the way these operators deal with factors from different domains, including market environment, technology and regulation. Other terms used to indicate these telecom network operators are (access) network operators or (access) network providers.

Content-related demarcations Although we are aware that users help create new broadband services, we focus exclusively on the supply side. We deal with customers only in terms of market demand, as an abstract notion of specific user needs and requirements. We focus from a social shaping point of view to the interrelatedness of technological, economic, regulatory and market-related factors. There is a second demarcation in the area of technology. Although there are all kinds of emerging advanced mobile technologies that are seen as alternative broadband infrastructures, mobile technologies are not included in this research. Only wireless systems that are possible alternatives for (parts of) a fixed broadband access network, the so-called fixed wireless access (FWA) networks, are included. Other applications like Wi-Fi hotspots and in-house networks, like Bluetooth and Powerline communication (PLC), fall beyond the scope of this research. Geographical demarcation The developments in decision-making concerning broadband are too complex to capture the behaviour of all telecom operators in the

25

world, or even in Europe, in a single research project. In addition, decision-making tends to be a typically national phenomenon, which means that each country adopts a different approach. For these reasons, the scope of this study is to a large extent limited to the geographical area of the Netherlands. Decision-making processes in other countries will mainly be used for reasons of comparison. Further cross-boundary insight will be gained through an international survey.

Relevance for stakeholders

Thus far, we have focused mainly in factors that complicate the decision-making process in the highly dynamic broadband environment. However, to know how these factors in the end result in the decisions involved, we also need to take a closer look at the decision-making process. Although decision-makers in telecom companies know roughly how the process works, they are usually unaware of the way the individual technological, regulatory and market-related factors function within the decision-making process and what their correlative influence is. As far as governments are concerned, understanding the complicated nature of the decision-making process with regard to broadband access network roll-out by telecom operators may help making policy more effective in terms of realising investments in broadband access infrastructures. Also, with regard to scientists, greater insight into the evolution of this decision-making process would provide more structure and allows for more focused research. There is a need for an integrated multidisciplinary scientific framework or a unifying theory which describes the way technological, economic, market-related and regulatory factors affect the migration of broadband networks in the local loop.

Such a framework or theory would be relevant to scientists, strategic decision-makers in broadband providing telecom companies and regulators. Although there are many relevant technological, regulatory and market-related factors, it is unclear how these factors are interrelated and which factors are dominant within the different phases of the roll-out process.

1.2 Research objective, research questions

If we look at the complexity in the decision-making process surrounding access networks, as discussed above, a number of knowledge gaps become visible. Firstly, as described earlier, there is no integrated multidisciplinary scientific framework or unifying theory combining technological, financial, market-related and regulatory factors affecting migration of broadband networks in the local loop. Secondly, there is a need for a theory combining complex decision-making theory with technology development and innovation. The third knowledge gap has to do with insight into the elements of various existing decision-making theories that can be applied to the decision-making process surrounding the upgrading of existing

26

and roll-out of new broadband access networks. More insight is desirable in which aspects of these theories fit this specific decision-making process in the Telecommunications industry. It is important to gain more insight into the complex relationships between technological, financial and market-related factors and the migration process from existing networks towards broadband networks (as defined in this research) and next generation broadband networks. This implies that we have to understand the ‘content’ side as well as the decision-making process involved in upgrading broadband access networks. This brings us to the objective of this research project:

To develop a scientific integrated multidisciplinary model that provides insight into the

drivers and outcomes with regard to local loop broadband roll-out by telecom operators

and to understand and explain the way decision-making with regard to broadband roll-out

evolves.

This insight into content-related and process-related aspects will make the specific decision-making process and drivers and outcomes of broadband roll-out in access networks more transparent and help explain the relationships between decision-makers and the dynamic multi-actor environment in which they operate. To reach the objective of this research, this thesis answers the following main research question:

What content and process-related aspects determine broadband local loop upgrades, the

outcomes of broadband roll-out and the decision-making process regarding the upgrading

of broadband access networks?

To answer the main research question, a number of steps will be taken. For each of these steps intermediate research questions are formulated and answered, after which it will possible to formulate an answer the main research question. Due to the complexity of the area under investigation, it is useful to apply several theories and discuss the various disciplines and the decision-making process. Because this research focuses on the decision-making processes of telecom network operators and on the way decisions are shaped from a content perspective, it is important to apply empirical research methods. Detailed insight into the evolvement of decision-making processes can be gained through in-depth interviews, whereas more generic insights can be gained better through surveys among a broader and more international group of experts. In addition to empirical data, theoretical methods, for instance Meta-analysis are very useful to explore the

27

existing theories and concepts within broadband literature and to place the empirical results in a broader context. For these reasons, several research methods will be applied to answer this main research question, forming together a multi-theory, multi-method approach. There is, moreover, a classic distinction between content-related and process-related approaches, both in decision-making theory and in strategy thinking. In this research we assume that the two are interrelated. The process-related approach has to do with the way the decision-making process evolves, while the content-related approach focuses on the technological, market-related and regulatory factors influencing this decision-making process. Chapters 2 and 3 of this thesis focus on the content-related side of this research. on the domain and literature of broadband roll-out respectively. In Chapters 4 and 5, we examine decision-making theory and risk and uncertainty theory. Chapters 6, 7 and 8 focus on content-related as well as process-related aspects. Chapter 6 contains the results of our qualitative research into content-related factors and the way decision-making with regard to broadband roll-out evolves in practice. In Chapter 7, we present the results of our quantitative research, while Chapter 8 contains the final conclusions of this thesis, bringing content and process together. The first step is to gain more insight into the domain in which telecom network operators are operating. More insight is needed to understand the broadband domain and the content-related factors network operators have to deal in their decision-making process. Consequently, the first intermediate research question, comprising three sub questions, is the following: RQ 1

What are the technological, market-related and regulatory factors that play a role with

regard to the upgrading of access networks by telecom network operators towards next

generation broadband?

With regard to the technology domain two more detailed sub questions need to be answered if we are to understand the core subject of decision-making with regard to broadband network roll-out: a) What are important variables determining the choice between several migration paths?

b) Which possible migration paths can be and are followed to reach next generation

broadband networks?

To gain insight into the existing theoretical concepts and hypotheses related to factors influencing broadband roll-out, we carried out a systematic literature analysis via a Meta-analysis. The objective of the Meta-analysis is to develop a conceptual model that can be used in the research project. The conceptual model

28

provides insight into the relationships between the factors involved in broadband evaluation paths. The development of a conceptual model based on Meta-analysis has proved very effective and efficient (Van den Hooff, 1997). This leads to the second intermediate research question:

RQ 2

What multidisciplinary frameworks or conceptual models regarding broadband roll-out are

currently available?

One of the problems with regard to infrastructural investments is the necessary high upfront investment in combination with long ROI periods and an uncertain market environment. These circumstances affect the extent to which network operators are willing to invest. Because, ultimately, investments and revenues determine the ROI time of network investments – and with it the operators’ annual revenues - makes the financial perspective highly important. Additional insight is needed into the financial aspects of broadband investments. Also, it is important to understand the high-risk character of these investment-related decisions, which brings us to the following intermediate research question: RQ 3

How do network operators assess several investment possibilities concerning upgrading

and roll-out broadband networks in the local loop?

An important factor operators must take in consideration with regard to investing in a specific technology to upgrade their network is the level of risk and uncertainty of this technological alternative in relation to aspects like price development, expected demand, future-proofness, etc. These risks and uncertainties affect the likelihood of investments being earned back within the expected timeframe. As a result, assessing risk and uncertainty is an important aspect of network investment-related decisions. To be able to minimise the undesired effects of risks and uncertainties on broadband roll-out investment-related decisions, there are three aspects of these factors that are relevant: the characteristics of the risks and uncertainties themselves, their effects on broadband roll-out and the methods that operators can apply to assess them. This results in the following research question:

RQ 4

Which risks and uncertainties are affecting these investment-related decisions, what are

their effects on broadband roll-out and how can risks and uncertainties be assessed and

minimised

29

In addition to greater insight into the multidisciplinary factors influencing broadband investment-related decisions, as well as the methods that can be applied to choose between several investment-related options, greater insight is needed into the decision-making process that ultimately results in these investment-related decisions. This brings us to the next intermediate research question:

RQ 5

How does the decision-making process by telecom network operators regarding the

migration of broadband access networks evolve?

Validation of the conceptual model resulting from the literature analysis and the development of a parsimonious content model regarding the drivers and outcomes of broadband roll-out will take place via a qualitative and quantitative data analysis. The qualitative data analysis is based on expert interviews with strategic decision-makers of major telecommunication operators and network providers in the Netherlands. The data will be analysed via concept mapping. The quantitative validation consists of two online surveys: a pre-test and a main study. Firstly, a small-scale pre-test will be conducted to develop scales and gain insight into the construction of our questionnaire and items. The results of the pre-test provide initial insight into the validity of the conceptual model. The pre-test is carried out to develop scales via exploratory factor analysis, which are used as the basis of our main study, another online survey, which was directed more at stakeholders. The main study first comprises a confirmatory factor analysis to create a measurement model, which in turn is the basis for a structural equation model. The hypotheses originating from the theoretical and content-related background are tested via structural equation modelling with AMOS 7.0. These validation steps are performed based on our final intermediate research question:

RQ 6

Can we develop a parsimonious model that provides insight into the determining factors

with regard to broadband-related decision-making by telecom network operators and the

outcomes of broadband roll-out?

The approach combining content and process, presenting our research outline as described above, can be visualised as follows:

30

TheoryFinancial assessment

methods and

risk & uncertainty

TheoryConcepts

broadband roll-out

TheoryDecision-making

Conceptual content model on broadband

roll-out

Process models Risk & uncertaintyGarbage can model

Logical incrementalism

Empirical resultsQuantitative validation

Content model on broadband roll-out

P

r

o

c

e

s

s

C

o

n

t

e

n

t

Empirical results

Qualitative validation

Domain description

Process of broadband

Decision-making

Figure 1.1: Research outline

1.3 Thesis outline

This research contains a content-related side and a process-related side. Chapters 2 and 3 shape the content-related part of our theoretical framework and focus on the main drivers and outcomes of broadband roll-out in the technological, market-related and policy-related dimensions. Chapter 2 provides greater insight into the technological, market-related and regulatory domains of the environment in which telecom operators operate. In Chapter 3, we explore the available literature and the existing concepts with regard to broadband roll-out in the local loop, as well as the possibility of developing an initial conceptual content model. Chapters 4 and 5 shape the process part of our theoretical framework and focus on the decision-making process of broadband roll-out in access networks and on the impact of risk and uncertainty on the decision-making process. Chapter 4 provides greater insight into theory on financial assessment methods for broadband infrastructural investments and provides a framework for measuring risks and uncertainties related to these infrastructural investments. In Chapter 5, we focus on theory related to decisions and decision-making, discussing two theories that could be applied to decision-making with regard to broadband roll-out. Chapters 6 and 7 focus on the validation of the content-related side and process-related side of our theoretical framework, via qualitative and quantitative data analysis. In Chapter 6, we present the results of our qualitative data analysis, which was based on expert interviews with strategic decision-makers from Dutch telecom companies. In Chapter 7, we focus on our final content-related model regarding broadband roll-out, examining validation via a quantitative data analysis based on survey analysis. In Chapter 8, finally, we present the conclusions of this research

31

and answer our main research question via answering the intermediate research questions formulated above, discuss the implications of our research for the various stakeholders involved and provide suggestions for further research.

32

1.4 References

ACIL Tasman (2004). Economic Impacts of Broadband Adoption in Victoria. Retrieved January 17, 2008, from http://www.mmv.vic.gov.au/uploads/downloads/BAO/EconomicImpactsofBroadbandREPORTFINAL2004.pdf.

Bouwman, H., Hooff, B. van den, Wijngaert, L. van de & Dijk, J. van (2005). ICT in

Organisations: Adoption, Implementation, Use and Effects. London: Sage Publications.

Centre for Economics and Business research (CEBR) (2003). The Economic Impact of a Competitive market for Broadband. Retrieved January 17, 2008, from http://www.bigfuture.org/cebr%20Final%20Report.pdf.

CPB (Netherlands Bureau for Economic Policy Analysis) (2005). Do market failures hamper the perspectives of broadband? Retrieved January 17, 2008, from http://www.cpb.nl/nl/pub/cpbreeksen/document/102/doc102.pdf.

Criterion Economics (2003). The Effects of Ubiquitous Broadband Adoption on Investment, Jobs and the US Economy.

Doraszelski, Ulrich (1998). The net present value method versus the option value of waiting: A note on Farzin, Huisman and Kort. Journal of Economic Dynamics & Control, 25 (8), 1109-1115.

Dialogic Innovatie & Interactie (2005). Breedband en de Gebruiker 2004-2005. Utrecht: Vermaas, K., S. Maltha, F. Bongers, J. Segers, & L. van de Wijngaert.

Dialogic Innovatie & interactie (2007). Breedband en de gebruiker, 2007. Utrecht: Karianne Vermaas, Jeroen Segers, Hugo Gillebaard, Bram Kaashoek, and Guido Ongena.

Eurescom (1999), Extended investment analysis of telecommunication operator strategies- Investment analysis framework definition and requirements specification. (Project P901-PF Deliverable 1).

European Commission (2000). Proposal for a Regulation of the European Parliament and of the Council on unbundled access to the local loop (2000/C 365 E/15). Official Journal of the European Communities, 9-12-2000. Retrieved January 17, 2008, from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:C:2000:365E:0212:0214:EN:pdf

European Commission (2000). eEurope 2002: An information society for all. Retrieved January 17, 2008, from

33

http://ec.europa.eu/information_society/eeurope/2002/documents/archiv_eEurope2002/actionplan_en.pdf.

European Commission (2002). eEurope 2005, an information society for all, An Action Plan to be presented in view of the Sevilla European Council. Retrieved January 15, 2008, from http://ec.europa.eu/information_society/eeurope/2002/news_library/documents/eeurope2005/eeurope2005_en.pdf.

European Commission (2005). i2010 – A European Information Society for growth and employment. Retrieved January 17, 2008, from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=COM:2005:0229:FIN:EN:PDF.

European Commission (2008). Preparing Europe’s digital future- i2010 Mid-Term Review. Retrieved June 5, 2008 from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=COM:2008:0199:FIN:EN:PDF.

Expertgroep Breedband. (2002). Nederland Breedbandland, recommendations to the cabinet from the national broadband expert group. The Hague.

Firth Lucy and David Mellor (2005). Broadband: benefits and problems. Telecommunications Policy 29, (2-3), 223–236.

Gartner (2002). The Payoff of Ubiquitous Broadband Deployment. Gartner: Kathie Hackler, Ron Cowles, Jin Shen, and Behram Dalal.

Herder, P.M. and Z. Verwater-Lukszo (2006). Towards next generation infrastructures: an introduction to the contributions in this issue. International Journal of Critical Infrastructures, 2 (2-3), 113-120.

Ims, L. A., D.Myhre and B.T. Olsen (1998). Costs of upgrading the residential telecommunications infrastructure to broadband. Conference proceedings of GLOBECOM'98: Global telecommunications conference, The Bridge to Global Integration, 6, 3153 – 3158.

Macklin, B.(2002). The value of widespread broadband, Entepreneur.com, August 13, 2002. Retrieved May 20, 2004, from http://www.entrepreneur.com/article/0,4621,291780,00.html.

NZier (2004). Sustainable infrastructure: A policy framework, Report to the Ministry of Economic Development. Retrieved January 17, 2008, from http://www.med.govt.nz/upload/18061/nzier.pdf.

Price Waterhouse Coopers (2004). The Broadband Future, Interactive, Networked, and Personalised.

Rand Europe/IDC Benelux (2003). TEN Telecom Guidelines Status Review. Brussels: Jonathan Cave, Maarten Botterman, Renske Ellens, Paivi Luoma, Gert Jan de Vries, Roel Westerhof.

34

Renkema, Theo J.W. (2000). The IT value quest, How to capture the business value of IT-based infrastructure., Chichester: Wiley.

The Allen Consulting group (2003). True Broadband Exploring the economic impacts. Retrieved January 17, 2008, from http://www.citynet.nl/upload/ERN01_Final_Report_2_Broadbandproductivity_1.pdf.

Telecommunication Industry Association (TIA) (2003). The economic and social benefits of broadband deployment. Retrieved January 17, 2008, from http://www.researchictafrica.net/images/upload/Broadbandpaperoct03.pdf.

TNO (2007). Marktrapportage Elektronische Communicatie. Retrieved January 15, 2008 from http://www.ez.nl/dsresource?objectid=156081&type=PDF.

Van den Hooff, B., Groot, J. and De Jonge, S. (2005). Situational influences on the use of communication technologies. Journal of Business Communication, 42 (1), 4-27.

Van de Wijngaert (2001). Fysieke en affectieve toegang, geschiktheid; vraag, aanbod en context. In Harry Bouwman (Ed.), Communicatie in de informatiesamenleving (51-71).Utrecht: Lemma.

Vermaas, Karianne (2007). Fast diffusion and broadening use (Doctoral dissertation, University of Utrecht, 2007).

Vlaam de, Helene I.M. and W. Martin de Jong (2002). Infrastructure competition and the creation of value: an Anglo-Dutch comparison of regulatory regimes for the provision of rail and telephony services. International journal of Technology, Policy and Management, 2 (2), 125-143.

Weening, Heleen (2006). Smart Cities- omgaan met onzekerheid (Doctoral dissertation, Delft University of Technology, 2006).

Weijnen, Margot P.C. and Ivo Bouwmans. Innovation in networked infrastructures: coping with complexity. International journal of Critical Infrastructures, 2 (2/3), 121-132.

35

Chapter 2 Domain description

In this chapter, we describe the broadband domain and the main actors involved. In the environment of broadband decision-makers, three main dimensions can be distinguished: technology, market and policy & regulation. In this chapter, we address these three domains. In paragraph 2.2, we focus on the market perspective, consisting of the supply side (competitors and equipment suppliers) and the demand side (end-users). In paragraph 2.3, we focus on existing technologies for the upgrading of broadband, comparing them on the basis of relevant characteristics, and discuss possible migration paths towards Fiber-to-the-Home (FttH) networks. In Paragraph 2.4 we provide greater insight into questions relating to policy and regulation, discussing regulation and legislation, broadband stimulation projects and public fiber initiatives.2

2.1 Three perspectives on broadband in the local loop

In view of the technological, market-related and regulatory developments, network providers have to decide how to invest in network upgrades. Their decisions are based on several considerations related to these three perspectives. There are many relevant technological, regulatory and market-related factors. In this chapter, we look at the developments in the three above-mentioned dimensions and at the way they affect the decision-making process with regard to broadband roll-out. The three dimensions are presented in Figure 2.1. We begin with a brief discussion of these three perspectives on broadband.

2 To write this chapter, neither new data nor new literature has been gathered after the end of 2007. In some cases, where considered relevant, some recent developments have, however, been used.

36

Telecom operator

- Internal business

processes

- Investment

decisions

- Decision-making

processesCompetitors

Suppliers

End users

Market

Technology

Policy &

regulation

Figure 2.1: The environment of telecom operators

Market - supply side From a supply side perspective, the central question is whether it is possible to develop viable and feasible business models for the infrastructure and service providers (Bouwman et al., 2006). Operators who own their own network provide network access as well as access to services like Internet, telephony and television. For that reason, broadband operators face competition from network providers and from access providers, which in most cases are market parties providing access to Internet services without a network of their own (ISPs). Thanks to the development of Voice over IP (VOIP), these ISPs are now also able to offer relatively cheap telephony services over the Internet. Another group of actors are equipment suppliers, who provide equipment for telephony networks, including central offices, switches, cabinets, etc.

Market - demand side Although we focus mainly on the supply side, we are aware of the importance of the demand side. Market developments are driven by push factors as well as pull factors. Market demand is an important driver for a continuous investment in higher bandwidth, the availability of which in turn leads to innovative use. Ultimately, end-user subscriptions determine the annual revenues of telecom operators. New real-time applications, such as multi-actor games, video-conferencing or streaming media, are the cause of increasing demand for bandwidth (Bouwman et al., 2006). When it comes to accepting broadband, users view subscription costs, lack of availability and installation costs as thresholds.

37

Higher speed, always-on, cost reduction and control are considered benefits of broadband access to the Internet (Dialogic, 2003; Vermaas, 2007). Furthermore, research indicates that, even though only a small portion of society has embraced the Internet and broadband technology as a way of life, there is a large group of light users that will only adopt broadband if it is the standard (for instance cable is) and prices do not differentiate (Dialogic, 2003). In paragraph 2.2 we take a closer look at the market perspective.

Technology The low throughput of networks linking the core to the ‘in-house’ network, the access network, continues to present a problem (Cuchran et al., 2001), although the capacity of the core telecommunication and data-communication networks as well as ‘in-house’ networks has grown significantly in the last years, increasing from Mbps to Gbps. In their present form, access networks are approaching their technological boundaries. Existing as well as alternative transmission media have to be deployed to anticipate further increases in demand and to allow companies to keep up with their competitors. High-speed broadband networks with at least 10 Mbps symmetrical data speed can be realised via several evolutionary paths, consisting of small incremental network upgrades (see paragraph 2.3). Network operators can choose from a number of options, such as upgrading cable connections, implementing new technologies for copper, installing optical fiber networks or rolling out wireless networks, as well as a host of hybrid transmission solutions. We take a closer look at the technology domain in paragraph 2.3 of this chapter.

Policy & regulation Governmental and regulatory issues are related to the different roles governments play: policy-makers, investors or regulators (Alleman, 2002). Local and national governments can play different roles. Firstly, governments can play an important role in realising the benefits of broadband (Firth & Mellor, 2005). They can play the role of stimulator, which is an important role particularly in the early stages, when industry or technology requires a new technological standard or infrastructure. One of the questions that need to be addressed is whether broadband should be defined as a public infrastructure in the sense that the infrastructure is owned by the government or that governments invest in infrastructure, or as a domain where market parties have to take the lead. When governments start investing in networks themselves, they fulfil a role of financer. By doing this, they are subjected to strict regulation and run the risk of violating European rules governing market distortion, unfair competition and state aid (see paragraph 2.4). Furthermore, government fulfils the role of regulator. There are several regulatory issues that play a role for network operators with regard to with competition between network providers, interconnection between networks, obligations related to Significant Market Power, competition law, etc. These issues are discussed in paragraph 2.4 of this chapter. Other issues that play a role are

38

regulation with regard to tax, general legal liability, piracy, etc. We will not discuss these issues.

Before describing the three perspectives outlined above in greater detail, we introduce the four layer model based on the seven layer Open System Interconnect (OSI) model, which is commonly used in describing the broadband market. The model provides a clear framework that can be used to provide a more structured description of the market.

The four layer model as a general reference model for broadband

An often used technological breakdown to describe the broadband market is the seven layer OSI model. For the sake of practical applicability of this model, a model with four functional layers is now commonly used. These layers are clearly recognisable in the financial power field (Expertgroep Breedband, 2002). Figure 2.2 shows the four layer model of the broadband market. We briefly discuss the four layers of this model.

Content & applicationsLayer 4

Service provisioning (xSP)Layer 3

Passive InfrastructureLayer 1

Active Infrastructure and switchingLayer 2En

d u

sers

Figure 2.2: The four layer model

Layer 1 The first layer, the passive infrastructure, contains ducts, cables (optical fiber, coaxial cable, copper twisted pair) and equipment set-up points, including main distribution frames (MDF) and central offices (CO). Layer 2 The second layer contains active infrastructure and switching equipment. This layer uses the transmission capacity provided by layer 1, offers transmission services to layer 3 and provides wholesale services. Layer 2 contains, among other things, active infrastructure in central offices, for instance switching, containing optical ports, transmission equipment and network maintenance. Layer 2 also provides so-called service aggregation, the connection from end-users with their service providers. Network access points, located at end-user homes as well as

39

additional added value services, like digital rights management (DRM), are also part of the second layer. This layer is also known as the ‘transmission layer’. Layer 3 The service-provisioning layer (Layer 3) contains access services to the content layer (layer 4). Examples are access services to the Internet, television, Video on Demand (VOD) and telephony. Layer 4 The upper-most layer, layer 4, contains the actual broadband applications. In this layer, the actual content and end-user services are realised. (Breedband Expertgroep, 2002) The focus of this dissertation is mainly on Layer 1 of the four layer model, the physical infrastructure. Developments at the other layers of the model will only be used and described to illustrate and explain developments with regard to Layer 1. In the next paragraph, which contains a description of the broadband market, we will frequently refer to this model to describe the market and actors.

2.2 The Dutch broadband market

This research focuses on broadband providers offering residential broadband services via their own backbone and/or local loop infrastructure. In the Netherlands, the companies involved are concerns KPN (incumbent), UPC, Tele2-Versatel and Ziggo. Ziggo was created in 2006 after the merger of cable companies @Home, Casema and Multikabel. The infrastructures include copper TP, coax and optical fiber. A small percentage of the Dutch residential broadband local loop connections consist of optical fiber (FttH), connections that are provided either by private companies or by local governments3. In this paragraph, we discuss the background and trends in the Dutch residential broadband market, paying attention to the main market players, developments with regard to broadband, the existing market structure and the role of end-users.

2.2.1 Supply side

Traditionally, there were two types of networks in the Netherlands, copper twisted pair (TP) networks for the transmission of telephony and several regional coaxial cable networks for television broadcasting.

KPN, the Dutch incumbent

Originally, telephony was provided at the local level, but by the end of the 19th century, telephony became more and more nationalised. In 1940, the last local telephone networks became state property. In 1915, the state company called Post and Telecommunications was founded, with a national monopoly for providing

3 Based on Article 5.14 of the new Telecommunications Act, broadband offering by local governments is now forbidden.

40

telephony services. In 1928, its name was changed to “Staatsbedrijf der Posterijen, Telegrafie en Telefonie” (PTT). From 1980 onwards, the network became more intelligent, ISDN networks were developed and the network’s backbone was replaced by optical fiber. On January 1, 1989, the PTT became liberalised and its name was changed to “Koninklijke PTT Nederland NV” or shortly “KPN”. KPN was listed at the stock exchange in 1994. In 1998, the postal activities were separated and an autonomous company called TPG (now TNT) was founded. From that moment on, KPN was only a telecommunication provider. Its name was changed to Royal KPN NV. In the Netherlands, KPN is the owner of the fixed telephony network, market leader in mobile telephony and possesses several Internet Service Providers (ISPs). KPN is now also focusing on full ICT service provisioning and is, moreover, offering more layer 4 services.

Cable providers

In addition to the telecom operator, cable providers also play an important role. Their roots are different from those of the telecom operator. During the 1950’s and 1960’s, most Dutch homes had antennae on their roofs. For aesthetic reasons and to reduce signal noise, the Netherlands switched to cable television en masse4. During the 1990’s and after, a great deal of money was invested in modernising the cable networks, making them suitable for two-way digital traffic, enabling end-users to receive as well as send data.5 Moreover, an enormous push from the hardware industry took place to roll out cable networks. The network capacity was enhanced by replacing the core of the network with optical fiber networks, resulting in the name “hybrid-fiber-coaxial” or HFC-networks. There were many regional and local public cable providers, each with their own network structures. During the 1970’s and 1980’s, the cable networks expanded. At the end of the 1990’s, the cable market was liberalised, after which the various municipal cable companies were quickly taken over by private companies. This process is illustrated in Figure 2.3.

4 Retrieved May 01, 2008 from www.NLKabel.nl 5 Retrieved May 01, 2008 from www.NLKabel.nl

41

’88 ’93 ’98

IJsseltry

EGD

Westergo

PNEM

MEGA

’99

Pa-

letEssent Kabelcom

@Home

Casema

NKM

Kabelnet Alphen aan den Rijn

Kabelnet Amstelveen

Casema

Casema

’00

MultikabelMultikabel

(Primacom)

’05

Multikabel (Warburg

Pincus)

’06

EWR

Gelrevision (Gamog)

ENWTele

kabel

UPC

Kabelnet gemeente Haarlem

UPC

Eneco

’07

Casema

(Warburg pincus)

@Home

Ziggo(War-

burgpincus

&Cinven)

UPC

Castel (EDON NV)

Figure 2.3: Historical consolidation of Dutch cable networks6,7

Figure 2.3 also shows the strong consolidation process that took place in the cable market between 1993 and 2007. The final major consolidation took place in 2007. In February of that year, the construction of Zesko Holding was completed. Zesko was the temporary name for the merger of three Dutch cable companies @Home, Casema and Multikabel. In May 2008, the name was changed to ‘Ziggo’. This new holding will serve 3.4 million subscribers. The holding is owned by the private investment companies Cinven and Warburg Pincus. As a result of this strong consolidation the Dutch telecom market is now dominated by three major players, each with their own local loop infrastructures: KPN, Ziggo and UPC. Although there are more cable operators in the Netherlands, they all have regional monopolies, due to the fact that they operate in different geographical areas. In that sense, these networks are complementary rather than redundant. As a result, end-users cannot choose between cable operators, which means that there is no competition between the cable operators. They are, however, competing with ADSL providers. Next, we discuss this infrastructural competition in greater detail.

6 Beacuse of the complexity of the historical mergers and acquisitions, this figure may not be complete, 7Sources:Wikipedia,www.mediamagazine.nl/medianieuws/archief/2000/2000_02_2.html, http://web.planet.nl/computer/multim/1-5-98/mm1-5-98a.html, http://tweakers.net/nieuws/9509/Amstelveen-verkoopt-kabelnet-aan-Casema.html,http://www.nieuwsbank.nl/inp/1999/01/0121U049.htm, http://web.planet.nl/computer/multim/13-11-97/mm13-11-97c.html, www.nieuwsbank.nl/inp/1999/04/0412F082.htm, http://www.sat-net.com/listserver/dutchmedia/msg00150.html, www.dutchmedia.nl/dm599.htm#6, http://archief.trouw.nl/artikel?REC=851caeb41888a6be79c5a49299f70c2d, www.nieuwsbank.nl/inp/2000/02/0203M066.htm, jaarverslag Essent,2000, all consulted 11-01-2008

42

Strong competition brings high ‘broad’band penetration in the Netherlands

In 1995, the cable companies started providing Internet access. CAIW was the first cable Internet provider in the Netherlands. In 2000, KPN started its ADSL roll-out, which meant the beginning of the development of broadband Internet. Also, cable operators continued improving their networks, which made higher bandwidths possible. This infrastructural competition continues to this day. KPN has started VDSL roll-out and is performing trials with FttH. As response to these developments, cable companies have been running pilots with new high-speed cable technologies. In 2003, cable provider @Home installed Ethernet to the Home (EttH) technology in a pilot network in the city of Boxmeer, providing a symmetrical bandwidth of 10 Mbps.8 In 2006, the company carried out new field tests with an upgraded version, providing 40 Mbps download bandwidth and 8 or 32 Mbps upload bandwidth.9 At the end of 2007, UPC carried out its first field test in Amsterdam with Docsis 3, a new version of the Docsis protocol with a potential of 160 Mbps download and 120 Mbps upload speed10 (see paragraph 2.3 for further details on these technologies). The serious competition between ADSL and cable operators was made possible by three important developments: regulation regarding access and Open network provisioning (ONP), the coexistence of two fixed networks with national coverage and the convergence of data, voice and images. We discuss the relevant regulation in greater detail in paragraph 2.4. First, we take a closer look at the other two developments.

Two fixed infrastructures Serious competition in the Netherlands is first of all made possible by the fact that there are two infrastructures with national coverage (PSTN and cable) that have been developed side by side. Although on both the copper network and on the cable network there are several operators who provide access services to these networks, the Dutch market, seen from an infrastructural

perspective, remains a duopoly. An important question, also for the national regulator, is whether or not two infrastructures are enough. The structure involved is shown clearly in Figure 2.4. In June 2007, the Netherlands had the most serious infrastructural competition in all of the OECD countries (OECD, 2008). Between 2007 and 2010, the market share of DSL is expected to fall slightly, in favour of cable and fiber. This development is also visible in Figure 2.4. Actual developments have to be followed very closely to see if these statistics really turn out to be true.

8 Retrieved may 20, 2008 from: http://corp.ziggo.nl/Over+ons/Innovatie/Ethernet+to+the+Home.htm 9 @Home lanceert 40Mbps internet abonnement. 31-08-2006. Retrieved November 26, 2007 from. http://tweakers.net/nieuws/44217/home-lanceert-40mbps-internetabonnement.html. 10 UPC in Dutch DOCSIS 3 test, Broadband TV news, September 7, 2007. Retrieved on May 20th 2008 from: http://www.broadbandtvnews.com/?p=2030.

43

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2004 2005 2006 2007 2008 2009 2010

DSL Cable Fiber

Figure 2.4: Market shares access technologies in the Netherlands, 2004-2010

(Source: OPTA, VKA, 2007)

Convergence Traditionally, communication media were separated and their services were provided via distinct infrastructures. Broadcasting, voice telephony and on-line computer services were physically separated and operated on different platforms: TV and radio sets, telephones and computers. Convergence is the combination of these different media into a single operating platform, in a merger of telecom, data processing and imaging technologies, which has introduced a new era of multimedia in which voice, data and images are combined into end-user services (NetTel, 2004). The 1997 EU Green Book on Convergence (European Commission, 1997b) mentions two types of convergence: technological convergence and service-related convergence. In the case of technological convergence, different types of infrastructures can essentially transmit the same kinds of services, which results in the disappearance of the fixed relationship between service and network (Asscher, 1999). The second type is convergence of services, which means that the functions of end-user equipment, like telephones, TVs and computers, start to overlap (Asscher, 1999; Dommering et al., 1999). Networks are no longer designed to transport a single service, but become a conduit via which any type of information can be transmitted, in a phenomenon known as ‘service neutral infrastructure’ (Dommering et al., 1999, p.9). Early examples of this kind of convergence are the fax (convergence between post and telephony) (Dommering et al., 1999, p.37) and Videotex (convergence between broadcast and telephone network) (Dommering et al.,1999, p.41). Recent examples include Internet services provided over TV sets,

44

via systems like Web TV; e-mail and Internet access via mobile phones; Web casting of radio and TV programmes on the Internet; and voice telephony (VoIP) and TV access over the Internet via ADSL networks. As a result of convergence, most Dutch cable operators, ISPs and KPN are now able to offer service packages like ‘dual play’ (a service package combining Internet access and (VoIP) telephony or ‘triple play’ (a service package combining television, (VOIP) telephony and Internet access). An important consequence of this development is that actors who traditionally operated in separate markets have now become competitors in pursuit of the same end-users.

Other infrastructure providers and cooperation with telecom operators

Due to a number of specific difficulties related to investments in infrastructure, in particular in the local loop (see also Chapter 1), the development of fiber networks is progressing very slowly in the Netherlands (see Figure 2.4). The investment horizon of telecom operators lies around three to five years. Rolling out an entire, nationally covering, new fixed network is, however, impossible within such short time frame. The payback time of such enormous operations is much longer than five years. Companies that by their nature have a much longer investment horizon are real estate companies. While telecom operators are hesitating with the roll-out of fiber for reasons of viability, as far as real estate companies are concerned, the business case for fiber to the home networks is at this moment a viable one.11 In 2005, the Dutch firm Volker Wessels acquired a 100% share in KPN’s network building division (Financiële Telegraaf, 2005). In 2006 and 2007, Reggefiber, the investment company of Dick Wessels, acquired several Dutch fiber infrastructure providers, including Telecom Utrecht, Eurofiber, Y3 Net and XMS services, as well as acquiring majority shares in the fiber infrastructure providers Fastfiber and Lijbrandt Telecom and shares in ONS net Nuenen, ONS Brabant net, TrenT and Citynet Amsterdam. In 2007, Reggefiber became the biggest private investor in fiber networks in the Netherlands.12 It provides FttH to 155,000 Dutch households and the plan is to increase this number to 1.5 million by 2015.13 In October 2007, KPN announced the cooperation between Reggefiber and KPN in the roll-out of a fiber network to 70,000 homes in the city of Almere (KPN press release, 2007).Reggefiber will build the network. May 2008, KPN acquired a minority interest of 41% in Reggefiber.

11 Glas klaar voor machtsovername, Kabelnetwerken ouderwets en te duur. Computable 08-06-2006. Retrieved May 20, 2008 from: http://www.computable.nl/artikel/ict_topics/ictbranche/1814695/2379258/glas-klaar-voor-machtsovername.html 12 See note 6. 13 Reggefiber mikt op 1.5 miljoen huishoudens in 2015, 05-09-2007. Retrieved May 20, 2008 from: http://www.shopadsl.nl/adsl-nieuws/121_reggefiber-mikt-op-15-miljoen-huishoudens-in-2015.html.

45

Equipment suppliers

Equipment suppliers are another group of actors in the broadband market. These actors operate on layer 2 of the four layer model (Figure 2.2), providing equipment for telephony networks, for instance central offices, switches, cabinets, etc. Examples of large international equipment suppliers are Alcatel-Lucent, Cisco Systems, Nokia-Siemens, Samsung, ZTE, Huawei, and Ericsson. Although these companies are not competing directly with network operators, network operators depend on them for their equipment. Moreover, these suppliers have expert technological knowledge, which most decision-makers do not have, due to the fact that technological developments have become increasingly complex and advanced over the years. Many technological developments take place in the R&D departments of these equipment suppliers, which means they have the specific technological expertise that is needed for network roll-out. The dependency on equipment suppliers with regard to the availability of network equipment and knowledge means that network providers face a certain level of risk and uncertainty. We address this issue in greater detail in Chapter 4.

Internet service provisioning

Access to the Internet and Internet services is provided by access providers, also-called ‘Internet Service Providers (ISPs). As a result of regulation on interconnection and special access, operators owning their own fixed networks are obliged to provide other operators with access to their networks. This has allowed players that do not have their own local loop infrastructure to provide access services to end-users by using the final part of the network of cable operators or KPN. In return, they pay an interconnection fee to the network owner involved. In addition to providing Internet access, ISPs currently also offer access to telephony provided via IP (VoIP), which makes them more into generic service access providers. Examples of this kind of providers are Tele2-Versatel, Alice and Orange. They operate at layer 3 of the four layer model (see Figure 2.2), i.e. the service provisioning layer and provide the link between the infrastructure (layers 1 and 2) and content (layer 4). They provide end-users with access to the content, which is delivered over the infrastructure. In other words, these actors provide access to the added value of the networks, which makes layer 3 of the network an interesting position from a financial point of view. This is the reason that all network operators, in addition to being active on layers 1 and 2, also operate on layer 3 of the network. In 2007, @Home was the largest ISP in the Netherlands with 639,900 subscribers (12.7% market share), providing Internet access via the coaxial cable network. With 601,000 users and an 11.1% market share, KPN’s Planet Internet was the second largest ISP. Planet Internet provides Internet access via the copper TP network. In 2006, the top six broadband ISPs served almost 70% of the overall market (Telecompaper, 2007).

46

Between 2005 and 2007, the Dutch ISP market was characterised by a strong consolidation trend. Especially KPN took over several ISPs. KPN services the broadband market with four major brands (Planet Internet, Het Net, XS4ALL and Direct ADSL). In 2005 and 2006, KPN acquired several ISPs (see Table 2.1). At the end of 2006, KPN’s ISPs serviced more than 40% of the Dutch Internet market (Telecompaper, 2007).

Table 2.1: Acquisitions of ISPs by KPN since 2005

Acquisitions by KPN Subscribers

Tiscali Netherlands (2006) 276,000

Speedlinq (2006) 40.000

Demon (2006) 70.000

HCCNet (2005) 55.000

Freeler (2005) 16.000

Cistron (2005) 5.000

Tiscali, via KPN-network (2005) 60.000

Total 522.000

Source: TNO

After discussing the main market players, we now continue with describing the development of the broadband market in the Netherlands and put this development into an international perspective.

Broadband market is still growing, but mainly ‘midband’

As discussed in Chapter 1, the term ‘broadband’ has changed over time. In most market studies, broadband is referred to as Internet access via ADSL and cable (Docsis 2) technology. However, these technologies do not provide the transmission speed that is defined as ‘broadband’ in this research (i.e. 10 Mbps symmetrical), which results in a somewhat distorted picture. In some of the latest market studies, FttH subscriptions are also included, which offer genuine broadband access as defined in this research. In this study, we refer to bandwidth provided by ADSL and Docsis2 technologies as ‘midband’. In Q1 of 2007, the total number of ADSL and cable Internet subscriptions in the Netherlands was 5.3 million (TNO, 2007a). At that time, 80% of Dutch households had an Internet subscription, of which 71% a subscription via ADSL or cable infrastructure, representing an annual growth of 12.3% compared to 2005 (Telecompaper, 2007). Penetration per 100 inhabitants increased to 31.9 % at the end of 2006 (TNO, 2007a). On an annual basis, ADSL and cable penetration per 100 inhabitants grew with 5.4% (Telecompaper, 2007). This growth is depicted in Figure 2.5.

47

0

1000

2000

3000

4000

5000

6000

Q404 Q105 Q205 Q305 Q405 Q106 Q206 Q306 Q406

Bro

ad

ba

nd

co

nn

ec

tio

ns

(x

10

00

)

Figure 2.5: Number of Dutch ADSL and cable connections 2001-2006 (x 1000), (source:

Telecompaper, 2007. TNO, 2007a)

Compared to other countries, the number of broadband connections in the Netherlands, presented in OECD and other market studies is among the highest in the OECD countries, together with Denmark. A comparison between the number of broadband connections per 100 inhabitants between 2001 and 2006 for ten countries is presented in Figure 2.6.

0,0

5,0

10,0

15,0

20,0

25,0

30,0

35,0

2001 2002 2003 2004 2005 2006

Nu

mb

er

of

'bro

ad

ba

nd

'co

nn

ec

tio

ns

pe

r 1

00

0 i

nh

ab

ita

nts

Netherlands

Denmark

South Korea

Finland

Canada

Japan

UK

France

US

Germany

Figure 2.6: International comparison: number of broadband Internet connections per 100

inhabitants, 2001-2006 (source: OECD)

48

It is not clear, however, whether the same technologies are included for all countries in this comparison. The problem with many market studies is that they compare different technologies among different countries, while using the term ‘broadband’ in all cases. In South Korea and Japan, the roll-out of fiber to the home (FttH) technology is much further advanced than in the Netherlands and Denmark, for example. In the latter two countries, ADSL and cable technology penetration is very high, which results in a somewhat distorted picture, providing them with a higher broadband penetration ranking than Japan and Korea. To avoid presenting a distorted picture, it would be more accurate to make a comparison per access, rather than comparing ‘broadband’ connections in general. A comparison per access technology is presented in Figure 2.7.

0%

20%

40%

60%

80%

100%

Denmar

k

The N

ether

lands

South

Kore

a

Finlan

d

Canada UK

Franc

e

Japa

nUS

Germ

any

Co

nn

ecti

on

s p

er

100 i

nh

ab

itan

ts

DSL Cable Fiber/LAN Other

Figure 2.7: International comparison: number of broadband Internet connections per 100

inhabitants, 2001-2006 per access technology (source: OECD)

From Figure 2.7, it becomes clear that the roll-out of fiber access networks in the Netherlands and Denmark lags behind compared to South Korea and Japan. It also becomes clear that, although the roll-out of ADSL is almost the same for the Netherlands and Denmark, Denmark is further advanced when it comes to fiber roll-out. In Finland, the UK, France, Japan and Germany, cable-based Internet access represents a much smaller portion of the market than in the other countries. In these countries, DSL is clearly the dominant access technology.

49

Worldwide, South Korea and Japan are the only countries where a substantial part of broadband Internet connections is realised via fiber/LAN (FttX) (TNO, 2007a).14 In June 2007, fiber connections accounted for 36% of all Japanese broadband subscriptions and for 31% of all Korean subscriptions (Lee, 2008). In terms of the actual number of subscribers, in the middle of 2007, Korea had 4.5 million FttH subscriptions. In January 2008, Japan had 9 million FttH subscribers. Figure 2.7 shows that, in the Netherlands, Fiber/LAN connections only cover 0.4 per 100 inhabitants. In Q1 of 2007, FttX coverage in the Netherlands has increased slightly to 0.74 per 100 inhabitants (TNO, 2007a). In small parts of the Netherlands, optical fiber has been rolled out in the local loop. These are local pilots and networks, with are either owned by private companies or by public institutions, municipalities, housing companies, provinces or public-private partnerships. These significant differences in the number of broadband subscribers can be attributed to three main differences between the various countries. The first difference has to do with the role of governments. The Japanese and Korean governments have created an interesting financial climate for the roll-out of fiber to the home networks. Through its ‘u-Japan’ broadband strategy, the Japanese government provides money to cities to wire schools and community centres. It, moreover, provides zero-interest or low-interest loans to cities and businesses to deploy broadband and provides tax breaks for the purchase of networking equipment.15 Japanese authorities also set out to devise significant incentives to persuade Japanese companies to invest in new ultra-high-speed cable, especially in rural areas.16 In Korea, besides financial stimuli, the government also uses other incentives for broadband roll-out. The ITU report on Korea highlights the fact that the government offered low-interest loans to bring high-speed access to apartment complexes (p.12). In addition, the Korean government acted as a launching customer by providing networked computer systems in schools and by spending heavily on its own computer systems, in what is now one of its top ten expenditures (p. 32). On top of this, in 1999 and 2000, the Korean government provided facilities-based Service Providers (FSP) with US$ 77 million at a very low interest rate, to invest in broadband networks. The effect was to create services worth over seven billion dollars and five to eight thousand jobs (p. 32)17. In Europe, public investments in private markets aren’t possible without any problems. They are subject to strict rules and often create problems with regard to state aid and false competition (see paragraph 2.4 on regulation for more details). The second difference between Korea and Japan on the one hand and Western European

14 http://www.cedmagazine.com/Article-International-010108-Japan.aspx 15 Japanese government investment in broadband pays dividends, April 4, 2007. Retrieved May 20, 2008 from: http://government.zdnet.com/?p=3048. 16Network neutrality and the false idol of innovation, May 10, 2006. Retrieved May 20, 2008 from: http://www.oreillynet.com/etel/blog/2006/05/network_neutrality_and_the_fal.html. 17 See previous note.

50

countries on the other has to do with architecture. The types of buildings in Japan and South Korea strongly are different than the ones that are common in European countries. The Asian cities are characterised by large numbers of high-rise buildings and apartment blocks. This makes it relatively easy and cheap to roll-out fiber to the building/curb, because many households can be served with one cable. In a country like the Netherlands, however, the predominance of low-rise buildings make it relatively expensive to roll-out fiber, because the costs involved cannot be shared among multiple residents in one building. The third difference is the difference in the culture and characteristics of Europeans on the one hand, and Japanese and Korean people on the other. Society within these Asian countries is characterised by many high-tech developments, gadgets and active service development. The popularity of available content, such as online gaming, has spurred broadband growth.18 Important reasons why the roll-out of ‘real’ broadband networks is lagging behind in a country like the Netherlands are the lack of broadband service development and lagging end-user demand. We focus on these market-related aspects in the remaining part of this paragraph.

2.2.2 Demand side

The importance of service development

When no services that demand genuine broadband speeds are broadly available, there is no need for end-users to take out subscriptions to higher data speeds, which means there is no trigger for infrastructure providers to upgrade their networks. This problem has to do with the interdependency between infrastructure roll-out and service development, which is a result of the fact that, in most cases, the benefits of network roll-out do not end up in the same place where the costs are incurred (Expertgroep Breedband, 2002). This problem is also known as the ‘chicken- egg problem’: development of useful services does not start until the infrastructure (high-speed broadband networks) is available, while suppliers do not invest in infrastructure when there is no clear view on useful content.

Diffusion theory and critical mass

Innovation Diffusion Theory (IDT) argues that the adoption of (new) media by later users depends on the presence of other users. Most people will only come on board if a sufficient number of users has preceded them (Van de Wijngaert, 2001). The costs of adopting an innovation at the beginning of the adoption cycle outweigh the benefits. Early adopters play an important role in the development of a technology. The moment enough initial investors have adopted a technology to

18 http://www.cedmagazine.com/Article-International-Spotlight-South-Korea.aspx

51

make it worthwhile for the other members of a social system to invest, the so called ‘critical mass’ has been reached. We would argue that this type of interaction is likely to occur with regard to broadband services and broadband infrastructures.

The European Commission applied diffusion theory in 1993 in its ’White Paper on growth, competitiveness and employment: The challenges and ways forward into the 21st century‘ (COM (93)700, 05.12.93), which was written to stimulate broadband roll-out and reap the social and cultural benefits (see also Chapter 1) It calls for “the development of services and applications so as to attain a critical mass sufficient for further infrastructural investments and for attracting new users adding to the value of a given network” (RAND, 2003). The moment critical mass of fiber to the home users has been reached, the costs will drop and the advantages will increase, making a commercially viable offering possible. When a commercially viable offering is possible, more and more people will adopt a new technology, which in turn becomes increasingly attractive. The increase in the number of users causes a large acceleration in the rate of adoption, in a self-perpetuating process. At that moment the so called ‘flywheel’ has started to work (see also paragraph 2.4.4). Because it is ultimately the end-users who make sure revenues are generated, decision-makers consider end-users as an utmost important variable in decision-making. However, an important problem with regard to end-users is that their behaviour is hard to predict, because they are capricious in nature, which means that there is no such thing as ‘the broadband adopter/user’ (Vermaas, 2007, p.208).

2.2.3 Conclusion

As has become clear in this paragraph, the broadband market is changing quickly and continuously. Traditional network operators have turned into triple-play providers, offering telephony, Internet access and television, which has significantly increased the complexity in the market in which they operate. Network and service operators no longer service separate markets, but are now tightly interwoven (convergence). Offering attractive end-user services is, however, highly important to operators. When no –legitimately acquired- services that demand genuine broadband speeds are widely available, there is no need for end-users to take out subscriptions to higher data speeds, which means that there is also no trigger for infrastructure providers to upgrade their networks. These changes in the market and the position of a company within the value chain in turn affect the internal business processes within the organisations involved. Not only are departments changed or added when new services are offered. Logistics, billing systems,

52

marketing and supply processes must also be adapted to the organisations’ new position within the value chain. In addition to the complexity caused by the dynamic competitive field, the unpredictable behaviour of end-users and the necessary changes in internal business processes, decision-makers also face an increasing complexity and fast changes in the technological environment. In the next paragraph, we focus on the technological domain.

2.3 Broadband technology19

This paragraph is an extended version of the paper published as:

Fijnvandraat, Marieke and Harry Bouwman (2006). Flexibility and Broadband evolution, Telecommunication Policy, 30 (8,9),424-444.

In recent years, the capacity of the core telecommunication and data communication networks, as well as ‘in-house’ networks, has grown significantly, increasing from Mbps to Gbps. However, the low throughput of networks linking the core networks to the ‘in-house’ networks (the access networks) continues to present a problem (Cuchran et al., 2001). In their present form, access networks are approaching their technological boundaries. It is virtually impossible to increase capacity any further without technological adjustments. Moreover, demand for bandwidth is expected to increase further. The existing access networks are not designed to accommodate today’s exploding need for services and applications (Cuchran et al., 2001).

In this paragraph, we provide greater insight into the broadband technologies that are currently available (and/or under development) and enable evolutionary paths for broadband deployment (symmetrical bandwidth of 10 Mbps20) in the local loop, avoiding path-dependency. We focus on the question what the paths that allow for a more phased, flexible and evolutionary approach to broadband roll-out may look like, taking combinations of existing and emerging technologies into account. We discuss and compare several wired and wireless transmission media on the basis of criteria like network capacity (bandwidth), reach, compatibility, standardisation and technological and geographical scalability. In this paragraph, we focus primarily on the capabilities of the technologies, in order to understand the opportunities that are available to network providers, and the technological options that play a role in their decision-making process. We assess broadband technologies like ADSL2 (+), Etherloop, EFMC, Narad networks, EttH, EPON, 19 The last update on the information on the technology dates from March, 2006. Information about the technologies that became available later has not been incorporated into this thesis. 20 Not all the technologies discussed in this paper can deliver this broadband capacity (symmetrical bandwidth of 10 Mbps). If not, they can, however, function as a step in an evolutionary path towards this transmission speed.

53

APON, Active optical Ethernet, LMDS, MMDS, FSO and HFR on the basis of these criteria. Starting from this assessment, we present and discuss migration paths for copper, cable and wireless technologies.

Although there are various emerging advanced mobile technologies that are seen as alternative broadband infrastructures, we do not discuss mobile technologies in this study. Only wireless systems that are possible alternatives for (parts of) a broadband access network, the so calling fixed wireless access (FWA) networks are included. Other applications, including Wi-Fi hotspots and in-house networks, like Bluetooth and Powerline communication (PLC), fall outside the scope of this study.

Transmission media and technologies Transmission media form the first layer of the four layer network (see Figure 2.2). They transport signals via electromagnetic waves. They can be divided into fixed and wireless media. Fixed media include copper, coax and optical fiber. Wireless media involve transmission through the air, for instance via radio and optical links. The public telephone network is predominantly based on twisted pair, while the cable network primarily uses coaxial cables in the access networks. Over the last 20 years, network providers have installed optical fibers in the transport network linking switches (Gagnaire, 1997). Point-to-point (PtP) microwave connections are examples of wireless communication.

Fixedlocal loop

Coax Optical fiberCopper Radio

Wirelesslocal loop

Optical

FSO

WIMAX

WLAN/WiFi

MMDS

LMDS

HFRADSL 2(+)

VDSL

Etherloop

EttH

Narad networks

Optical Ethernet

EPON

APON

EFMC

BPON

GPON

Figure 2.8: Broadband transmission technologies for the fixed and wireless local loop

Transport networks can use various transmission technologies to enable broadband. In order to realise a symmetrical bandwidth of at least 10 Mbps, new technologies have to be applied. Figure 2.8 provides an overview of these

54

technologies. Although not all the technologies we discuss here are currently able to deliver this symmetrical speed, we have decided to include them in our analysis when they play a role in the incremental evolution towards a 10 Mbps broadband capacity. In addition to transmission capacity and reach (the maximum number of Mbps that can be transmitted over a certain distance at a reasonable quality), we include the following attributes in our analysis: � Compatibility: the modifications that have to be made to the existing telephone,

cable or wireless network, � Standardisation: the presence of a de facto standard and (with regard to

wireless media) frequency spectrum and the frequency range being used, and � Technological and geographical scalability: the ability to respond quickly to

demand and network growth and to provide services for high-bandwidth applications.

2.3.1 Copper-based broadband technologies

The PSTN network has a hierarchical composition and consists of 1st order traffic exchanges on national level. These exchanges are connected to 2nd order traffic exchanges at a regional level that split up to the central offices (also called Local Exchanges) in cities and villages. The local exchanges are each connected to several remote nodes at neighbourhood level and after that to manholes at street level. Finally, each home is connected to a separate cable. Up to the level of central offices, the networks nowadays consist of optical fiber connections. The part from the central offices to the end-user residents consists of copper TP connections. This part is called the access network, also called ‘local loop’ or ‘first or last mile’. The composition of PSTN networks is presented in Figure 2.9.

Optical fiber cable

Copper TP cable

MDF

L2/L3

switch

Central office/ Local Exchange

Telephone

exchange

ADM

DSLAM

manholeRemote node

Figure 2.9: Composition of the PSTN network

55

The traditional access network has become increasingly important in providing access to the Internet and other broadband services, in addition to telephony. There are a number of ways to increase the capacity of existing networks: ADSL 2 (+), Very High Speed Digital Subscriber Line (VDSL), Ethernet in the first mile over copper (EFMC) and Etherloop. In this paragraph, we take a closer look at these technologies. ADSL2 offers an increase in coverage, a reduction in power consumption and cross-talk noise and an increase in service density. It is, therefore, ideal for offering competitive symmetric high-speed access connections at rates between 1 and 3 Mbps (Androulidakis et al., 2004). ADSL 2+ allows for a significant increase in data rates on shorter telephone lines (AWARE, 2004). ADSL2+ is predicted to cover distances of up to 2 kilometres from the Digital Subscriber Line Access Multiplexer (DSLAM), with a capacity of 10 Mbps (Stordahl & Elnegaard, 2004). A major advantage of ADSL2+ is that the line cards that are required are less expensive than the regular ADSL cards. What this means is that all that needs to be done to provide access to high-speed Internet is to replace the line cards. To further improve capacity, it is necessary to bring the fiber network closer to the access network and use VDSL (Foster et al., 1998). VDSL, an extension of ADSL, supports speeds of up to 52 Mbps at a distance of 900 metres (Cuchran et al., 2001). However, in all cases involving DSL, performance is highly dependent on subscriber line quality (Androulidakis et al., 2004). EFMC is a point-to-point copper topology that delivers Ethernet over copper cable at 10 Mbps for distances up to 750 metres and at 2 Mbps for distances up to 2,700 metres (Hackenberg, 2005). One example of EFMC is Cisco’s Long Reach Ethernet (LRE). Etherloop (ETHERnet Local LOOP) is a technology from Nortel Networks that combines DSL and Ethernet to deliver up to 6 Mbps between customer and central office (CO).

Table 2.2 shows how the broadband copper technologies perform on six important criteria, as mentioned above. Of the four serious technological alternatives, Etherloop has the biggest potential in terms of transmission capacity and reach, but it is yet not standardised. At the end of 2007, ADSL2+ had been implemented in the majority of KPN’s network. Several Dutch pilots with VDSL took place in 2007.

56

Table 2.2: Score of new fixed technologies that use the PSTN network21

ADSL 2(+) VDSL Etherloop EFMC

Transmission capacity (bandwidth)

24 Mbps down and 3 Mbps up; 10 Mbps at 2 km from the DSLAM.

Asymmetrical: 12.96-55.2 Mbps down and 1.6–2.3 Mbps up.

Symmetrical: 19.2 Mbps

6 Mbps over distances of up to 6.4km.

15 Mbps over 1.5 km

Etherloop2 will reach 100 Mbps

Short reach option: > 10 Mbps

Long reach option: > 2 Mbps

Reach Bandwidth decreases with distance between end-user and local exchange. Maximum reach is 7 km.

900 m

Bandwidth decreases with distance between end-user and local exchange.

Circa 7 km. Short reach option :> 750 m.

Long reach option: > 2700 m.

Compatibility ADSL 2+ line cards in DSLAMs

ADSL2+ modem (CPE)

VDSL equipment in all remote nodes

Electronics in local exchanges

VDSL modem (CPE)

Extension of fiber in access network towards local exchanges

Installing bridges

Etherloop equipment in street boxes and telephone exchange

Partial optification

New data link sub-layers (closer to the PHY) are defined by existing DSL standards, and there are some new layers above those for rate matching and loop aggregation

CPE modems

Standardisation Standard is ratified by ITU in 2003

Standardised by ETSI in 2001

Not standardised IEEE 802.3ah

Scalability Up to 25 Mbps in best cases. By installing new line cards in DSLAMs the network can be scaled up per home.

Bandwidth reasonably high; more distortion with increased usage; Can be upgraded to FttB/FttH by extended fiber roll-out in the access network.

Providing VDSL demands adjustments to every remote node.

To 100 Mbps (Etherloop2)

Flexible up- and downstream, according to demand.

Etherloop can be scaled up per home.

Data rates and distances can increase through DSL modulation technologies and by “bonding” multiple copper pairs into one channel.

100 Mbps out to 100 m (single copper pair)

100 Mbps out to 750 m or 20 Mbps out to 2700 m (eight copper pairs)

21 Sources: Stordahl & Elnegaard, 2004; Planet Multimedia, Wat is dat, ADSL 2+?, 02-03-2005; Gagnaire, 1997;Cuchran et al., 2001; Foster et al., 1998.

57

2.3.2 Cable-based broadband technologies

In its present form, the cable network is a combination of coax and optical fiber as used in the backbone, which is why it is referred to as a Hybrid Fiber Coax (HFC) network. HFC has a tree-like topology that starts at the central network and branches out to connect individual households (Low, 2004). In other words, it is based on a point-to-multipoint configuration. The composition of an HFC network is depicted in Figure 2.10. The networks consist of fiber optic rings that connect local centres. The change from optical fiber to coax takes place at neighbourhood level. From the neighbourhood centres, individual coax cables are connected to amplifiers located at about 50 metres from private residences. The amplifiers are the initial node in the Mini-Star nets. From these nodes, 20 to 25 individual coax cables are connected to individual homes (TNO, 2002). The maximum capacity of each of the downstream channels is 40 Mb/s (Gagnaire, 1997).

CATV was originally designed as a one-way broadcast network. To provide Internet access and facilitate other interactive services, however, it needs to be bi-directional (Low, 2004). This means that several technological adjustments have to be made to HFC networks to make bi-directional transmission of voice and upstream data service possible, which presents several problems. Upstream transmission is problematic, because it is more difficult to maintain the appropriate carrier/noise (C/N) ratio in the upstream direction due to noise accumulation (also known as ‘noise funnelling’). Another problem has to do with allowing multiple users access to the network, which has privacy and security implications. Careful traffic control to guarantee the privacy of users and to avoid the possibility of illegal access or identity theft is required (Low, 2004). Even when these problems are solved, there is still the matter of upgrading the capacity of existing CATV networks.

58

Radio & TV

main station

Local

office

District

centre

Op

ticalfib

er n

etw

ork

Multi

tap

Local

office

Local

office

CMTSCMTS CMTS

CMTS

CMTS

CMTS

WAN

subscribers

District

centre

District

centre

Figure 2.10: Composition of an HFC network (source: TNO, 2002)

One way to upgrade capacity is to transport the capacity benefits of the optical fiber network to the coax part of the HFC network. There are two broadband technologies that can be applied: Narad Networks and Ethernet to the Home (or Cable Ethernet). Narad Networks create a fast two-way transmission by using the previously untapped frequencies above 860 MHz, which implies that they bypass the final active amplifier. Ethernet to the Home (EttH), which was developed by Teleste, Tratec and Ekc, makes an interactive HFC network suitable for Ethernet transmission (Cnossen, 2004). ETTH offers speeds of over 10 Mbps, later 50 Mbps and eventually 100 Mbps, both downstream and upstream, to residential and business customers, without the need for active equipment at the consumer’s

59

premises22. Successful trials with this technology are taking in Boxmeer, in the Netherlands. Another way to realise higher bandwidth capacity in cable networks is to change to a newer version of the DOCSIS protocol. DOCSIS (Data Over Cable Service Interface Specification) is an international standard developed by CableLabs and contributing companies. The protocol defines the communications and operation support interface requirements for a Data Over Cable system. It makes it possible to add high-speed data transfer to existing Cable TV (CATV). The current version is DOCSIS 2.0 and DOCIS 3.0 is being tested.

Table 2.3: Score of new fixed technologies that use cable23

Docsis 3.0 Narad Cable Ethernet (ETTH)

Transmission capacity (bandwidth)

152/304 Mbps down; 108 Mbps up

100 Mbps/1 Gbps dedicated PtP

100 Mbps per household

Reach Depending on the number of amplifiers en route

400-500 metres (to neighbourhood centres)

Depending on the number of amplifiers en route

Compatibility ▪ Roll-out of new CPE (modems) necessary

▪ Head end re-cabling and recombining

▪ Spectrum re-allocation

▪ Replacement of passive components (splitters, combiners, taps, etc.) may be required to ensure signal quality

▪ Implementation of Ethernet switches in the fiber optics networks

▪ Optification neighbourhood centres

▪ At low fiber networks: WDM or electronics between fibers or high rather than low fiber networks

▪ Narad equipment in neighbourhood centres

▪ Narad equipment in amplifiers

▪ Fiber needed up to neighbourhood centres, 400-500 metres from residences

▪ Segmentation and/or fiber roll-out towards the end-user

▪ No modems necessary, but new wall socket Fiber needed up to neighbourhood centres, 400-500 metres from residences

▪ Minimal configuration changes

▪ No in-home active equipment needed

Standardisation ▪ ITU-T Not standardised ITU-T DOCSIS 2.0

Scalability Geographic Scalability per local centre

Can be scaled up from 4 channels (152 Mbps down) to 8 channels (304 Mbps down)

Geographic scalability per connection if network reaches final amplifier, if not per final amplifier

Scalability in bandwidth with (D)WDM to Gbit Ethernet

Geographic scalability to 100% penetration

Scalability in bandwidth up to 100 Mbps

22 http://www.nederlandbreed.nl/pilots.asp 23 Sources: Cnossen, 2004; Cablevision Plans Deployment of 100 Mbps Data Service, CED magazine, June 28 2005; Narad networks at a glance (http://www.naradnetworks.com/news.html); Junkus, Justin.J. Bringing networks to life- Docsis 3.0-drivers and speed bumps, April 1, 2008. Retrieved from www.cable360.net/ct/sections/features/28858.html

60

The reach of broadband speed in cable networks usually depends on the number of amplifiers within the network. Table 2.3 compares two alternative technologies for realising broadband on existing CATV networks. Although Narad and EttH offer a capacity of up to 100 Mbps, they are still being developed. The Docsis 3 protocol enables download transmission speeds of up to 300 Mbps. However, implementing Docsis 3 requires adjustments in the network in every head end, which makes this type of upgrade relatively expensive and time-consuming.

2.3.3 Fiber-based technologies

Optical fiber access networks are considered the most promising alternative when it comes to broadband services provision in the local loop (Gagnaire, 1997). There are two basic versions of FttH, an active optical network (active optical Ethernet) and a passive one (PON). ‘Active’ implies that a distribution/concentration step takes place using active electronics in street boxes/distribution cabinets. In the case of passive designs, the distribution/concentration takes place within the telephone exchange, using two protocols: Ethernet (Ethernet-PON or EPON) and ATM (ATM-PON or APON). PONs are a cost-effective way to benefit from the large capacity of optical fibers (Gagnaire, 1997). They are passive in nature and use no active components in the outside plant. Instead, they use only passive components to distribute traffic. APON was the first commercial product and it was used primarily in business applications. BPON (broadband PON) is an expanded version of APON and it is presently the most widely applied standard. GPON (GigaPON) is an ITU-T-approved G.984 standard that takes PON to gigabit rates (Low, 2004).

In Table 2.4, the six possible fiber technologies are compared. Active Optical Ethernet, APON and BPON already are mature technologies. All fiber technologies have been standardised, can bridge long distances and offer high-speed connections. The main advantages of passive technologies are that they do not require active equipment and that their scalability is better.

61

Table 2.4: Score of new fixed technologies that use fiber optics24

Active

Optical

Ethernet APON EPON BPON GPON

Transmission capacity (bandwidth)

1 Gbps (802.3z)

100 Mbps (dedicated)

622 Mbps (shared)

38 Mbps (dedicated)

1 Gbps (shared)

62 Mbps (dedicated)

622 Mbps up, 1.25 Gbps down

Symmetric: 622 Mbps or 1.25 Gbps

Asymmetric: 2.5 Gbps down and 1.25 Gbps up.

Reach 1-50 km (depending on the standard)

20 km (single mode)

10-20 km (depending on the standard)

Up to 20 km 20-60 km

One switch needed

Switch per provider and per service

Compatibility Active equipment

▪ Optical line termination (OLT) device is installed at the central office (CO), while an optical network termination (ONT) device is installed on the other end, in or near each home or business site.

▪ Glass/coax or glass/copper converters at transition points (in case of FttC and FttB)

▪ Telephony: special modem in home and neighbourhood centre

▪ CPE: media converter/modem, private buildings for equipment housing

▪ Equipment in central office + remote node

Standardisation Yes, IEEE FSAN and ITU- T G.983.x

IEEE 802.3ah ITU-T G.983.x series

ITU-T G.984 standard

Scalability Every additional customer requires a separate fiber cable.

Scalable up to Gigabit speeds

Up scaling from PtP to PtMP is possible, not the other way round.

Sometime a new user can be added to splitter.

Single mode versions supports new broadband technologies to achieve gigabit speed

Through combinations of TDM with WDM and, eventually, CDMA.

Through increase of data rate, with faster lasers and more sensitive burst-mode receivers

Up to 6 Mbps downstream by adding the downstream bandwidth in the 1,490 nm and 1,550 nm wavelengths.

Through combinations of TDM with WDM and, eventually, CDMA

Through increase of data rate, with faster lasers and more sensitive burst-mode receivers

24 Sources: Da Silva, H. J. A, Optical Access Networks, 03-09- 2005 (available at: http://www.co.it.pt/seminarios/webcasting/itcbr_09_03_05.pdf); Low, 2004; Fiber-to-the-User: the Ultimate Endgame, Alcatel technology white paper (available at: http://www.alcatel.com/bnd/fttu/18062_FTTU_twp.pdf);

62

2.3.4 Fixed –wireless broadband technologies

The cost and complexity associated with traditional wired cable and telephone infrastructure have resulted in the development of new wireless technologies. In suitable situations, wireless connections, compared to fixed networks, make it easier and more cost-efficient to connect new customers; they involve lower maintenance costs and reduce thresholds involved in implementation. The deployment costs of rolling out the network are an important issue, which has a significant impact on the choice between fixed and wireless networks. Whether or not a particular wireless technology presents a suitable solution depends on a number of factors, including geographic circumstances, network availability, the weather, density, etc. Although all wireless systems use the same two elements, i.e. one or more broadcasting antennae and receivers, there are various types of wireless systems in use today. The main categories are point-to-point (PtP) and point-to-multipoint (PtMP), and Line-of-sight (LOS) and non-line-of-sight (NLOS) systems. In a LOS link, a signal travels over a direct and unobstructed path from transmitter to receiver; in a NLOS link, a signal reaches the receiver through reflections, scattering and diffractions (Wimax forum, 2004b). The PtP system is the most straightforward type of system, whereby one transmitter broadcasts to one user at another location. PtP systems can use various frequencies, from radio waves to optical laser and require direct LOS between sender and receiver. Since there is only one sender/receiver combination, these systems are not scalable in geographical terms.

In addition to PtP, there are the PtMP systems, in which several terminals communicate with a single base station. Like PtP, PtMP requires a direct LOS between sender and receivers. Unlike PtP systems, however, they are geographically scalable, since additional terminals can be connected to the system. PtMP systems use radio frequencies, both in the licensed and in the unlicensed ranges. Although using a licensed spectrum has the advantage of protection against interference from other operators, the licensing process is cumbersome (Wimax forum, 2004a). Finally, there are so-called omni-directional systems, which transmit signals in all directions.

In the case of wireless connections transmission, speed is the bottleneck. The higher the transmission speed, the easier it becomes to share bandwidth with multiple users. The maximum transmission speed and distance listed under the various standards are, however, purely theoretical. In practice, transmission speed depends on the number of users and on atmospheric influences, which often make it impossible to realise the maximum reach and bandwidth. Other possible signal-degrading variables are multi-path, fading and shadowing. The technologies being discussed are wireless in both directions and provide a (theoretical) bandwidth of

63

10 Mbps or more. Frequently used expressions with regard to wireless radio links are ‘Fixed Wireless Access’ (FWA) and ‘Wireless Local Loop’ (WLL). Possible wireless broadband technologies are:

� WLAN (Wireless Local Area Network) or Wi-Fi (Wireless Fidelity): a generic term for radio-based LANs, which encompass the 802.11 x standards family.

� WIMAX (Worldwide Interoperability for Microwave Access) is a wireless metropolitan area network (MAN) technology that connects IEEE 802.11x hotspots to the Internet and provides a wireless extension to cable and DSL for last mile broadband access.

� LMDS (Local Multipoint Distribution System) is a microwave technology that facilitates high-speed wireless communication. LMDS is a PtMP system that requires line of sight.

� MMDS (Multi-channel Multipoint Distribution System) is a one-way PtMP system that requires line of sight. Important characteristics are long reach, low speed and low costs.

Table 2.5 shows four wireless broadband technologies that are considered to offer great potential, either today or in the near future. Bluetooth and Ultra Wideband are technologies that have a reach that is too limited. WLAN/Wi-Fi can already be marked as an ‘off the shelf’ technology. Although an advantage of this technology is that it uses unlicensed frequency bands, which make it a cheap and scalable technology, at the same time this may cause problems associated with interference from other technologies that use the same frequency bands. Wimax provides high bandwidth, although the bandwidth it offers has to be divided among all the users in a cell. Wireless technologies invariably involve a compromise between the reach, speed and density of the network.

64

Table 2.5: Score of new wireless technologies that use radio waves25

WLAN/Wi-Fi Wimax MMDS LMDS

Transmission capacity (bandwidth)

• a: 54 Mbps

• MIMO: 216 Mbps shared

• Proxim: 108

Mbps

• Dual band: 10 Mbps

• b: 11 Mbps

• g: 54 Mbps

• n: 100 Mbps

75 Mbps shared in cell with radius of 6-10 km.

Up to 35-50 Mbps (net) with large channels (14-20 MHz)

Data rates typically from 0.5 to 3.0 Mbps within a

cell area

1 Mbps up and 25 Mbps down (shared)

Reach • Depends on configuration and frequency

Wi-Fi: 4-6 km with accurate antenna

Ethernet - Wi-Fi: 32 km

Up to 48 km with a typical cell radius of 6-10 km, depending on

frequency, transmit power and receiver sensitivity.

Maximum reliable coverage range: 40 km.

• short range: 4.8 km

• at 2.4 GHz: 16 km

Frequency spectrum

Unlicensed frequency

bands:

2,4 GHz:ISM band

5 GHz: UNII-band

Licensed and unlicensed

frequency bands

10-66 GHz (LOS) and < 11

GHz (NLOS)

In Europe licensed

bands, In US also unlicensed, NLOS and LOS

US 2.5–2.7 GHz

Europe: 40.5–

42.5 GHz band.

Licensed bands, all LOS

Europe: 24.5–26.5 GHz and 27.5–29.5 GHz, some

space in the 10 GHz band

USA: 28-31

GHz

40 GHz

25 Sources: Intel white paper on IEEE 802.16* and WIMAX, 2003 (available at: www.techonline.com/pdf/pavillions/intel/intel_wimax.pdf); www.wimaxforum.org/technology; Senior, P., 2004; www.wikipedia.com; Wimax forum, Business Case Models for Fixed Broadband Wireless Access based on WiMAX Technology and the 802.16 Standard, October 10, 2004; McKeough, C.,WIMAX, presentation, (available at: http://csmaster.sxu.edu/alzoubi/WiMAX-McKeough.ppt); Gagnaire, 1997; Sari, H., 1999; Fong et al.,2004; Xilinx, LMDS, MMDS Fixed Wireless Broadband Technologies, (presentation available at: http://www.xilinx.com/esp/wireless/collateral/lmds_mmds_esp.pdf); Ralli, 2004; Bigan, 2005, Goin Wireless: broadband architectures and systems, available at: http://www.eurescom.de/~public-seminars/1998/TRIBAN/27Bigan.ppt#1; Official IEEE 802.11 working group project timelines (2007-11-15). Retrieved from http://grouper.ieee.org/groups/802/11/Reports/802.11_Timelines.htm

65

WLAN/Wi-Fi Wimax MMDS LMDS

Compatibility Fiber optics required to the point where

wireless connection begins (the access point)

Fiber optics required to the point where

wireless connection begins (the access point)

Very easy to install for end-users

Antenna on every building

• Antenna on every building

• Radio components for high frequency

Standardisation IEEE

Wi-Fi:IEEE 802.11a, b, g

802.11n: Draft of the standard is expected to be finalised

03/2009; expected publication in 12/2009 (IEEE 802.11 working group)

IEEE 802.16 Wireless MAN

IEEE 802.16 fixed wireless broadband

IEEE 802.16 fixed wireless broadband

Scalability New proposed standard: 802.11n up to 100 Mbps (Intel Labs)

Limited geographical scalability due

to interference problems caused by other devices in the same unlicensed frequency

bands.

Flexible channel bandwidths: spectra can be divided into narrow sectors.

Same spectrum can be re-used by proper isolation between base station antennas

Optimised to deliver 110 Mbps (net)

services (in 3.5 MHz).

Scalability is limited by the bandwidth and number of channels

available.

Capacity can be increased

by making the cells smaller.

High level of scalability can be assured by:

Cell planning and cell splitting

Appropriate modulation scheme

Maximise utilisation of channels

Optimal application of frequency reuse

Upstream capacity can be enhanced

by adding extra radio channels at the customer

unit and by advances in technology

66

Wireless optical links

An optical link or Free Space Optics (FSO) is a wireless broadband technology that uses lasers or infrared light pulses26 and that can be used to supplement radio and optical fiber networks. FSO short-range applications have a reach up to 1 kilometre (maximum 2 kilometres) and can be used to connect buildings or establish a connection to the backbone (Leitgeb et al., 2005). There are PtP and PtMP configurations. A PtMP system is only possible with infrared and has two possible network topologies: the wheel and the mesh. The availability of FSO depends on local atmospheric conditions and on distance. Although rain does not seriously affect optical transmissions, transmissions are dramatically affected by snow and heavy fog, which cause a great deal of scattering (Leitgeb et al., 2005). A temporary application of FSO can be set up where the implementation of a network is delayed to such an extent that it damages service provisioning. A small group of free space lasers can be deployed at various growth locations within a network, making it possible to meet an increase in demand. A permanent application of FSO is useful in areas where fiber optics are prohibitively expensive or in situations where fiber optics cannot be used for technological or administrative reasons (Van Bussel and Rood, 2002).

Hybrid FSO-Radio (HFR) is a wireless system that combines FSO technology with the 60 GHz Millimetre Wave (MMW) technology. HFR considerably increases the reach of FSO. Fog and dry snow, which are the main problems with regard to FSO systems, have virtually no impact on the propagation distance of 60 GHz radio waves. Although at 60 GHz there is path loss due to absorption, the impact with regard to the propagation range under discussion is usually not all that significant. The same goes for FSO with regard to rain, which is the main problem of MMW connections. The two technologies are extremely suitable as back-up for each other, resulting in a very high reliability of the total system. Combining FSO with microwave links within an HFR communication network has the advantage of providing added redundancy and higher link availability (Leitgeb et al., 2005).

The wireless optical broadband technologies are compared in Table 2.6. FSO can offer good quality transmissions at high-speed only at short ranges. HFR can offer this combination over longer distances. An important advantage of FSO is its very short installation time. Unfortunately, both technologies are relatively expensive.

26 http://wcai.com/fso.htm

67

Table 2.6: Score of new wireless technologies using optical links

Free Space Optics Hybrid FSO-Radio (HFR)

Technology

• P2P

• P2MP: mesh/wheel

PtP

Capacity

• P2P: 155 Mbps/622 Mbps

• 2.4-10 Gbps has been

announced

155 Mbps/622 Mbps/1.25 Gbps (IP1000 system)

Reach

• P2P: number of km around switch:

• P2MP wheel: 1–2 km:

• P2MP mesh: 200-450 m in 4 directions:

• With 99.999% reliability: 1 km

• 100 Mbps: up to 2 km

• In open area: up to 24 km with

powerful lasers

Frequency-

spectrum

• Around 194 THz

• Around 375 THz

• Radio wave: 60 GHz

• FSO beam: THz area

License required?

No No

Standardisation

IEC security standard for products

In 2006, ITU-T has announced it will publish the first ITU-T Recommendation in the area of free-space optics27

IEC security standard for products

Services

Large service spectrum Large service spectrum

Technological modifications to network

Optical fiber to be installed up to:

• Cable distributor (P2P)

• Central Office (P2MP)

Optical fiber to be installed up to cable distributor

Geographical

Scalability

++

• connection can be realised within 2-3 days

• Network architecture remains the same when additional buildings are connected to the network

++

Scalability in

bandwidth

++

• Follows progression pace of optical fiber: WDM and DWDM can be applied

• Can be tailored to individual use demands

• Increasing capacity is simple: by changing the number of nodes and configurations

++

• Cost-effective migration from Fast Ethernet to Gigabit Ethernet across distances up to 1000 metres is

possible, provided there is an exchange of application modules.

27 Free Space Optics Standards Start, February 20, 2006. Retrieved May 22, 2008 from: http://www.itu.int/ITU-T/newslog/Free+Space+Optics+Standards+Start.aspx

68

2.3.5 Analysis

The attractiveness of the technologies discussed above is determined by how they score on relevant attributes. For that reason, the broadband technologies mentioned in this paragraph are compared not only in terms of transmission capacity, but also with regard to reach, compatibility, standardisation, maturity and technological and geographical scalability. Although costs are also a very crucial element, they change too quickly over time and between vendors to allow for an up-to-date and detailed overview, which means that including them would render this research obsolete the minute it was published. Also, other context-specific criteria, such as habitation density, geographical circumstances, economic welfare, etc., although they are relevant factors as far as individual network operators are concerned, are too specific to include in this discussion.

Transmission capacity One option telephone companies have is upgrading their networks to ADSL2+ and later to VDSL. There are, however, various factors that limit the data rates that can be achieved over twisted pairs in the local loop, including thermal noise, impedance impairments, cross talk, attenuation and propagation speed distortion (Gagnaire, 1997). Telecom companies can also opt in favour of the Ethernet protocol by rolling out EFMC or Etherloop. However, because the latter is not standardised, it is the more risky option.

Although Docsis 3.0, Narad Networks and EttH offer dedicated speeds of 100 Mbps, as technologies they are not yet mature. Fiber connections, of course, offer the highest data speeds, varying from 38 Mbps (APON) to 1.25 Gbps (GPON). Wireless optical connections also reach high speeds. FSO can potentially offer hundreds of Mbps and, in combination with radio links (HFR), up to 1.25 Gbps. There are also various different WLL technologies that can serve as alternatives to fixed local loop connections. Wi-Fi can offer download speeds of up to 54 Mbps, Wimax up to 75 Mbps, MMDS only 3 Mbps and LMDS up to 25 Mbps. However, all these rates are theoretical and must in practice be shared by however many people there are in a cell. The real data rates depend on the distance between end-user and antenna, weather conditions, the number of users per cell and the overall quality of the system.

Reach First of all, it must be emphasised that all maximum reaches mentioned here are theoretical distances. Real possible distances depend, as indicated earlier, on all kinds of environmental circumstances, including weather conditions, fading, obstacles, etc. All the transmission technologies we have discussed so far are capable of covering the distance in the local loop network (between 1 and 2 kilometres). The transmission rate for xDSL services is inversely related to the distance between subscriber and central office (Cuchran, 2001). MMDS and Wimax are the most attractive alternatives, with a reach of 48 kilometres. The fiber

69

optics technologies come second, with a reach of 20 kilometres. Wireless technologies, for instance LMDS, come in third place with approximately 12-16 kilometres. Almost all of these technologies involve a trade-off between reach, speed and quality of service (QoS). In particular weather phenomena, like heavy rainfall and fog, have a negative impact on high frequency wireless technologies, which is a problem that can be solved through HFR. Compatibility Different technologies demand different adjustments to existing networks. The more adjustments are needed, the higher the costs involved in upgrading and the bigger the chance of path dependency. In cases where many adjustments to an existing network have to be made, careful decision-making and comparison between different technologies can prevent undesirable path dependency. Going from regular ADSL towards ADSL 2(+) only requires new line cards in the DLSAMs and new customer premises equipment (CPE). Moving towards VDSL requires, in addition to specific VDSL equipment, a further optification of the copper network towards end-users, which makes it a more expensive and complicated solution than upgrading from ADSL to ADSL2(+).

Implementing Etherloop or Narad requires installing proprietary equipment at various network levels. In addition to VDSL, Etherloop, Narad and EttH also require partial optification towards end-users. Rolling out wireless networks requires less far-reaching adjustments to networks. Wireless connections can be installed in a few days by installing antennas and fiber connections to the point where the wireless connection begins (transmission point). To implement active optical networks (active optical Ethernet), active (electrical) equipment must be installed at various points in the network. However, this is not the case for passive networks. In case of FttC and FttB, optic-electrical converters are required in the fiber-to-copper transition points.

Standardisation When a technology is standardised, network providers know it will be available worldwide and that it will be compatible with other technologies. Standards enable economies of scale, which can bring down the cost of equipment, ensure interoperability and reduce investment risks for operators. Besides, suppliers can amortise their research and development costs over much higher volumes (Intel, 2003). The need for standardisation depends on requirements on global roaming and economies of scale. Proprietary solutions may be sufficient if no global roaming is needed and sufficient economies of scale can be achieved. As far as ADSL2(+), VDSL, EFMC, Docsis 3.0, WLAN, Wimax, APON, EPON, GPON, BPON, Active optical Ethernet, EttH, HFC, LMDS and MMDS are concerned, mature standards have been defined. Etherloop, Narad Networks, FSO and HFR have not yet been standardised. With regard to FSO and HFR, norms have been specified concerning the power to be used within a certain frequency band.

70

Technological and geographical scalability All wireless technologies and Etherloop have a high geographical scalability. Due to the high construction costs involved, the fiber optical technologies are the least scalable. In the case of VDSL, there is a maximum number of connections for quality is to be maintained at an acceptable level. This problem does not occur with Etherloop, APON, EPON and Active optical Ethernet. With MMDS, LMDS, MMW, FSO and HFR, the required line of sight restricts penetration in populated areas with high buildings. In the case of wireless technologies, it is important to have a sufficiently large frequency range to allow for the (expected number of) users and the possibility of new high-speed technologies in the same bandwidth. For the use of LMDS and MMDS, licenses are required. In the case of FSO and WLAN, no licenses are required, as long as operators do not use too much frequency space.

In terms of bandwidth, fixed and wireless optical technologies are reasonably scalable, due to multiplex technologies such as (D)WDM. In the case of LMDS, MMDS, WLAN and Wimax, expected data speeds exceed the existing speeds. However, in many cases this will affect the reach, which also makes it necessary to expand the frequency spectrum. When VDSL is upgraded, its reach quickly becomes too limited. Partial implementation of optical fiber can help solve this problem. In the case of Docsis 3.0, channel bonding is the technology that delivers increased downstream and upstream throughput. By bonding eight instead of four channels the downstream speed can be doubled from 152 to 304 Mbps.

Costs As we mentioned earlier, costs are an important factor in the choice between the various available technologies. To implement new fixed technologies, existing networks have to be modified. Most of the costs involved have to do with construction. FttH requires the highest initial investments. Prices are also affected by the transmission technology being used. ATM is an expensive transmission technology. Ethernet, on the other hand, is less costly, which makes technologies like Etherloop, EttH and EFMC relatively cheap. Generally speaking, it is cheaper to install wireless technologies than their fixed counterparts. Over the years, the costs involved in wireless connections have dropped considerably. WLAN/Wi-Fi and Wimax are, therefore, relatively cheap wireless alternatives. After Wi-Fi was standardised, the prices of several of its components fell significantly. Wi-Fi access is competitively priced. As a result of the costly and sensitive radio and laser components, high-speed LMDS and MMW systems are still expensive. MMDS is relatively cheap to use. Laser equipment and exact directional antennae can increase the costs. The equipment used in hybrid technologies, however, is the most expensive: although high speeds are feasible, they require major investments.

71

2.3.6 Evolutionary paths

When a decision has to be made which technology to select, there are many things that have to be taken into account, for example the existing outside plant, the network location, the cost of deploying the network, subscriber density and the return on investment (ROI) (Rui, 2005). Moreover, every technology has its advantages and disadvantages, and any decision will depend on what a provider needs and on the context in which the network operator has to operate.

In this paragraph, we focus on what paths may look like that allow for a more phased, flexible and evolutionary approach to broadband roll–out, taking advantage of combinations of existing and emerging technologies. We discuss the likelihood of transitions between the various technologies taking place, based on the evaluation of the technologies with regard to the criteria mentioned earlier. We discuss possible migration paths for copper, cable, wireless and optical fiber. All evolutionary paths eventually lead to GPON, the technology that, for the time being, offers the highest bandwidth over fiber networks. Needless to say, this state of affairs may change in the future, adding new paths and perhaps new ends to the technology decision trees, decision-makers that adopt an evolutionary approach have the opportunity to change development paths and end goals as a result of new developments in technology and environment.

Evolutionary paths based on copper technologies

The existing telephone networks, which are based on twisted pair, will continue to be upgraded, and as developments in Japan and Korea show, they will provide high-speed Internet connections and video over a copper last mile, thanks to improvements in software compressions and Internet connection technology. As soon as it has been sufficiently developed, ADSL2+ will replace ADSL connections. When demand increases, the next step involves switching to VDSL or fiber optics. In the case of VDSL, fiber needs to be brought closer to the end-user. FSO and Wimax connections could be installed at places within the network, to deliver high-speed services on locations where digging is difficult, not allowed or very expensive. When there is sufficient demand for optical speeds the final stretch of the local loop up to the end-user can be replaced by optical fiber. In the meantime, FSO can serve as an interim network for very high speeds. This means that people have access to high-speed data-connections at reasonable costs, which gives them an opportunity to use real broadband. If the demand for high-speed connections remains relatively modest, or if a lower level of reliability is considered acceptable, maintaining a VDSL configuration with a complementary FSO or HRF network is an option. Installing HFR requires high investments, but also offers high reliability, which makes a transition from HFR to fixed fiber unlikely.

If a choice is made in favour of VDSL, there are three factors that play an important role: ownership of the copper lines being used, the required reliability of the

72

connections and standardisation. Possible new copper technologies can use the existing telephone network, which means that new providers depend on the incumbent telecom providers for interconnection.

Alternatively, the Ethernet technology can be implemented in the copper network. Etherloop and EFMC are alternative copper solutions that offer higher bandwidths than existing ADSL networks. The major benefits of Etherloop are its low costs and the fact that it is compatible with PSTN, ISDN and ADSL. In addition, Etherloop uses the capacity of the telephone lines more efficiently than DSL technologies do, thanks to the application of burst mode transmission. Also, it requires fewer modifications to the network. Etherloop is not yet standardised, however, which means that proprietary equipment is required. EFMC is a possible alternative that also uses the Ethernet technology. The advantage of this technology is that it is standardised; a drawback is that it allows for smaller bandwidths than Etherloop. When operators are reluctant to implement proprietary equipment, they can use the EFMC technology, awaiting the standardisation and development of Etherloop or as a step towards VDSL or fiber. However, in cases where a choice in favour of VDSL as a long-term solution has already been made, it makes more sense to move toward VDSL via ADSL 2(+), because it offers the same bandwidth, while demanding fewer adjustments to existing ADSL networks. A drawback of this transmission path is that some problems occur if ADSL2+ and VDSL transmit signals over the same transport lines, because their spectra overlap. This problem may be solved by lowering the VDSL signal in the overlapping part; another solution is to implement VDSL and ADSL2+ in different parts of the network. Figure 2.11 shows possible transition paths from ADSL towards GPON.

73

ADSL 2+ADSL

VDSL

APON

EPON

BPON

GPON

Active Optic Ethernet

Etherloop

EFMC

EPON

BPON

GPON

APON

EPONBPON

GPON

Active Optic Ethernet

BPON

GPON

GPON

BPONGPON

VDSLAPON

EPON

BPON

GPON

Active optic Ethernet

EPON

BPON

GPON

Active Optic Ethernet

EPON

BPON

GPON

Active Optic Ethernet

Figure 2.11: Decision tree evolutionary paths from ADSL

Evolutionary paths based on coaxial cable technologies

Because upgrades from CMTS are designed to allow for two-way traffic, there will be important infrastructural competition between PSTN and HFC networks on the Internet market. When the moment comes that the maximum (shared) download speed (about 34 Mbps) proves to be insufficient, it is expected that phased implementation of Narad equipment will take place, allowing for high-speed connections up to 100 Mbps/1 Gbps. Operators can also decide to implement the Ethernet protocol via EttH, which offers broadband speed up to 100 Mbps at lower costs than Narad. Another advantage of EttH is hat the equipment is standardised.

A problem for Narad Networks is that most of the existing fiber optic cables within the HFC network are relatively thin, which means that there are few free channels left to accommodate Internet services. Reliability is also an issue, in particular the question whether or not an Internet blackout is acceptable. End-user Narad cable modems are not standardised and as a result they still are expensive. Important considerations with regard to Narad Networks are the delayed standardisation of the equipment and the high costs involved, combined with the fact that those costs can be spread over time by incremental implementation of the technology. Figure 2.12 shows possible transition paths from HFC towards GPON.

74

HFC

EttH

BPON

GPON

BPON

GPON

Narad

Docsis3.0

GPON

Docsis 2.0

APON

EPON

Active Optic Ethernet

BPON

GPONBPON

GPONBPON

Figure 2.12: Decision tree evolutionary paths from HFC

Evolutionary paths based on fixed-wireless broadband technologies

In most cases, wireless broadband connections are installed to provide quick and relatively cheap physical access to broadband. The main advantage of wireless broadband access systems is their limited cost due to the flexibility and facility of their implementation (Gagnaire, 1997). For these reasons, wireless access systems are of particular interest to emerging network operators (Sari, 1999). They provide a low-cost way to roll out flexible service packages quickly (Merrett, 1998). Nevertheless, their capacity remains limited in the light of broadband service requirements (Gagnaire, 1997). Relevant issues are the degree to which transmissions are directed (PtP/PtMP versus omni-directional), the bandwidth being offered, the frequency band within which the system operates, the question whether or not a license is needed, the influence of weather conditions, demographics, reach and QoS.

Wireless Broadband Access above 20 GHz, like Wimax, is seen as a complimentary or alternative solution to high-speed cable and DSL versions, such as ADSL2+ and VDSL (Elnegaard, 2005) and as a cost-effective solution in areas beyond the reach of DSL and cable (Wimax forum, 2004a). LMDS is undoubtedly more appropriate than MMDS for designating short-range millimetre-wave radio systems that will offer data and telephony services as well as TV programmes. LMDS systems are also much more suitable for supplying high-speed data services in densely populated areas (Sari, 1999). However, MMDS is capable of

75

covering greater distances than LMDS, up to 50 kilometres. Moreover, FWA networks such as Wimax, LMDS and MMDS can serve as back-up lines in case high-speed fixed networks are viable to roll-out or at the moment the geographical roll-out has reached the areas in which FWA access lined served as temporary solutions.

Free space optics (FSO) links can be used to set up communication networks or to supplement radio and optical fiber networks and present a potential way of connecting end-users to the backbone (last mile access). A central FSO unit can be connected to satellites, directional radio links, (mobile) telephone networks or fiber networks (Leitgeb et al., 2005). When a reasonable capacity is required, FSO technologies can be used at distances below one kilometre. One of their drawbacks is their extreme sensitivity to atmospheric conditions. FSO has a strong negative correlation between performance, reach and reliability, although on the other hand it offers a far shorter time-to-market than fixed broadband networks: FSO can be operational in a matter of days. Reliability and availability can be further increased by combining FSO and microwave links (HFR) (Leitgeb et al., 2005). When there is sufficient end-user demand and investment capital, providers can also opt in favour of taking a further step towards an optical fiber configuration. The transition from wireless to wired optical links would, however, require a large investment, which will most probably not be economic, because there will be no real gain in data rate for the end-user. If operators wanted to go for the wired option, they would probably have done so much earlier. Figure 2.13 shows the evolutionary paths from wireless technologies.

76

WLAN Wifi

Wimax

MMDS

LMDS

APONEPON

BPON

GPON

Wimax

LMDS

FSO

Active OpticEthernet

HFR

APONEPON

BPON

GPON

FSO

Active Optic Ethernet

HFR

APON

EPON

BPONGPON

FSO

Active Optic Ethernet

HFR

APONEPON

BPON

GPON

FSO

Active Optic Ethernet

HFR

EPON BPON GPON

HFR

BPON

Figure 2.13: Decision tree evolutionary paths from wireless technologies

Evolutionary paths based on optical fiber technologies

An alternative to these development paths is a direct choice in favour of optical fiber. Optical fiber is lighter, covers greater distances and offers greater bandwidth than copper or coax. It is robust and has a long depreciation period. However, installing and maintaining optical fiber networks is more complex and costly. In addition, fiber optics is characterised by poor geographic scalability and very high initial investments. Although APON requires larger investments than EPON, it does not offer a guaranteed QoS. EPON requires fewer and/or smaller facilities at neighbourhood level in order to establish the connection. PONs lead to solutions that are up to 50% cheaper than active networks, while requiring less space and maintenance. However, high construction costs remain a problem. The choice between FTTC, FTTB and FTTH should essentially depend on the unit cost per end-user, which in turn depends on the customer density in a given area and on the services the operator plans to offer. As far as FTTC and FTTB are concerned, the last mile of the distribution network is still based either on coaxial cable or on copper twisted pair (Gagnaire, 1997).

77

2.3.7 Discussion and conclusion

A major problem that arises with regard to the development of evolutionary paths is the speed with which the technologies are developing and the fact that the stakes in the industry are enormous. New technologies are usually described in industry reports, white papers, etc. These publications usually reflect the interests of the technology providers, which makes it hard to judge the actual status of the technologies. In addition, while technologies are described and their pros and cons investigated, new, alternative and better versions emerge at a rapid speed. Despite the rapid development of these technologies, it is useful to analyse the concept of evolutionary development paths and to outline possible alternative evolutionary paths on the basis of relevant criteria. However, because there are many regional variations in phases and rates of network development, adopted services and implemented network technologies, an evolutionary step in one situation may be a revolutionary step in another. In this paragraph, we presented an analysis for western countries, where the roll-out of fixed networks, like copper TP and coax networks, is highly advanced. In countries where no such networks have been rolled out, the main choice is between different wireless technologies.

The most likely development paths are the separate developments of evolutionary paths in copper TP (via VDSL) and coaxial networks towards fiber to the curb (FttC). Fixed Wireless Broadband Access, like Wimax, is seen as a complimentary or alternative solution to high-speed cable and DSL versions, such as ADSL2+ and VDSL and as a cost-effective solution in areas beyond the reach of DSL and cable. In addition, FWA networks like Wimax, LMDS and MMDS can function as back-up lines when the roll-out of high-speed fixed networks is viable or when the geographical roll-out has reached the areas in which FWA access lines served as temporary solutions.

If they are to make a balanced decision, operators need to be aware of the flexible, cheaper and potentially better technological scenario’s that are available, allowing them to respond in a flexible way to changes in user demand. The technologies we have discussed in this chapter are still evolving. Each new generation brings improvements in functionality, performance, price and QoS. This is a continuing trend that will keep presenting operators with new opportunities as well as uncertainties.

78

2.4 The policy & regulation perspective: stimulating and

regulating broadband roll-out28

In this paragraph, we provide greater insight into the regulatory framework as it applies to telecom and cable network operators that are working in broadband roll-out and the stimulating action plans initiated by the European Commission and Dutch government to achieve broadband roll-out. Moreover, we provide some insight into the legislation that applies to local and regional fiber pilots in case of public intervention in the broadband market. We start with a brief introduction of the background and origin of regulation in the telecommunications market.

2.4.1 Introduction

As discussed in paragraph 2.2, provision of telephony and TV services originally was part of the public domain. Between 1870 and 1970, the telecommunication market was characterised by state owned monopolistic companies, in Europe as well as in the United States. These monopolies were obliged by the various governments to offer a ‘universal service’. A universal service is a basic service package of a prescribed quality level, which is accessible at a reasonable price and at the same conditions for all members of society within the territory where the service is offered (Van Damme et al., 1998, p.10; Dommering et al., 1999 p.22). The Universal Service definition moreover specifies requirements related to quality and reliability of the service. The reasoning behind the need for creating monopolies to provide the universal service was that only monopolistic companies could also offer this service against low prices in unprofitable areas, because they are able to compensate their losses in unprofitable areas with profits from lucrative areas (cross-subsidisation). The second rationale behind creating monopolies was the argument of the natural monopoly, a situation in which operators can produce against lower costs thanks to of large scale production (economies of scale) than larger numbers of operators on the same market could within a competitive market (Dommering et al., 1999). In this period of national monopolies, major operators that have succeeded in establishing a sufficiently strong market position benefit from a number of factors that can allow them to fight new market entries effectively (Helberger, 2002). These incumbent operators have several advantages compared to new emerging market players. First of all, they are able to integrate enterprises horizontally, to share costs and risks. On top of that, economies of scale and scope and optimised use of resources and distribution channels allow them provide services more cost-efficiently (i.e. at lower prices) than starting operators. If the 28 Within this paragraph, secundairy sources will be used with regard to legal issues. The correct interpretation of the underlying primary sources will not be verified, because this paragraph has the intention to provide a global overview of relevant legal issues and the expertise of the author lies not in the legal field.

79

incumbents are vertically integrated, they can furthermore compensate losses at one level with profits from another level. Finally, they can create a situation of artificial scarcity of resources (Helberger, 2002). Between 1980 and 1990, all kinds of new technological developments took place, including the development of intelligent networks, digitisation, the development towards digital communication and optical fiber cable as new transmission technology. On top of this, new transmission networks were built, e.g. the satellite network, local fixed cable networks and analogue mobile networks. Moreover, the transformation towards a service economy created new demand for value-added services on top of the basic telephony offer. However, in this new competitive market this led to undesired distortion of competition, because the monopolists could easily offer these value-added services on top of their basic offer against much lower costs than their new competitors (Dommering et al., 1999).

High entrance obstacles and network effects demand external regulation

New entrants face several specific obstacles when they want to enter a market where incumbents are the dominant parties. First of all, they have to make large, irreversible investments to install the necessary facilities and they face cost asymmetries when it comes to gaining access to necessary resources. Secondly, they lack the competitive advantage of the first mover (network externalities) and have to deal with last-entry disadvantage, e.g. higher production costs. In addition, new entrants have to deal with the fact that their potential customers already have subscription contracts with the established operator, which results in additional efforts for customers in case they want to change provider. These so called ‘switching costs’ make the changeover to an alternative provider less attractive for end users. Also, incumbent operators have already established a reputation, which means that new entrants need to spend money to build a reputation (Helberger, 2002). These developments demand intervention from external regulators. Without external regulatory intervention, network effects will even catalyse the strong position of these major players (Helberger, 2002) A network effect occurs when consumption benefits depend positively on the total number of consumers who purchase compatible products (Church et al., 2002). Katz and Shapiro (1985) draw a distinction between direct and indirect network effects. A direct effect occurs when there is “a direct physical effect of the number of purchasers on the quality of the product”. When the network effect is indirect, consumption benefits do not depend directly on the total number of consumers who purchase compatible products, but on the effect their purchase has on the incentive to provide complementary products (Church et al., 2002). The combination of high entrance obstacles for new entrants to the market and the advantages for major incumbents creates competitive conditions that allow the latter to keep new entrants from entering the market. As a result, the likelihood that

80

the entry of new market parties will increase competition is, considered low (Helberger, 2002). Due to these network effects, the presence of a dominant economic power can affect the structure of markets and services, even when non-competitive behaviour is not the case (Helberger, 2002). The telecom market is characterised by a few dominant market players that control several levels of the value chain by vertical integration. These major players control essential network access facilities (a.o. the local loop) and are in a position to abuse their power to hinder existing and potential competitors. The reach of general competition law is not broad enough to effectively prevent the abuse of bottleneck control (Helberger, 2002) and to create a competitive market. For that reason, sector-specific regulation is necessary to create and maintain a level playing field. Sector-specific regulation In the middle of the 1980’s, the desire to create open markets and an internal European market resulted in the first initiatives to liberalise the European telecommunication market. In June 1987, the European Commission issued a "Green Paper" calling for the restructuring of the region's fragmented telecommunications marketplace (Romanelli, 1992). In 1990, this paper resulted in a series of directives better known as the ‘Open network provisioning (ONP) directives’.29 These directives were designed to provide other operators on the telecommunication market with access to the networks of the (former) monopolistic telecommunication companies (v.d. Hoven van Genderen, 2003, p.8). Harmonised principles for open and efficient access to and use of public telecommunication networks and general available telecommunication services were established in the framework. Since the liberalisation and privatisation of the telecom market, investing in infrastructure is now the responsibility of private companies. However, physical infrastructures such as cables have remained difficult and expensive to build and maintain and must form integrated entities to become meaningful production factors. (de Vlaam & de Jong, 2002). For a long time, existing incumbent operators were protected by exclusive rights during the roll-out of their old copper local access networks and their ability to fund investment costs through monopoly rents (EC, 2000, p.1). New entrants, however, do not have wide-spread alternative network infrastructures and are unable, with traditional technologies, to match the economies of scale and scope of operators notified as having Significant Market Power in the fixed public telephone network market. (EC, 2000, p.1) For this reason, introducing competition in the area of access networks requires regulation of the access to the essential facilities such as the local loop and specific network equipment. This has been realised by defining obligations for operators with

29 Richtlijn 90/387/EEG van de Raad van 28 juni 1990 betreffende de totstandbrenging van de interne markt voor telecommunicatiediensten door middel van de tenuitvoerlegging van Open Network Provision (ONP), PbEG 1990 L 192/1

81

Significant Market Power (SMP)30, as described in the ONP directives. These operators, controlling essential network facilities, are subject to specific obligations to offer providers without their own local loops access to their (access) networks and services based on objective, transparent and non-discriminating conditions. These directions opened the way for alternative operators to offer Internet access services to end-users and were instrumental in introducing greater choice in favour of end-users. Between 1999 and 2002, a review on the ONP directives took place. In 2002, this review led to a new Telecommunications Act. In the next paragraph, we provide insight into the concepts and directives that are the most relevant for broadband network operators.

2.4.2 The 2002 Telecommunications Act

This paragraph will provide a broad impression of legal developments in the broadband field. The Dutch Telecommunications Act is a translation of European legislation and describes the way five European directives for electronic communication markets are implemented in the Dutch market. The five directives comprise the framework directive (2002/21/EC), the access directive (2002/19/EC), the universal services directive (2002/22/EC), the authorisation directive (2002/20/EC) and the privacy directive (2002/58/EC). Member states are not allowed to differ from these directives in a negative way. The new Act enables providers, national governments and National Regulatory Authorities (NRAs) to respond to new developments in the telecommunications sector more quickly and flexibly.31 At the time this chapter was written, the European Regulatory Framework was under evaluation by the European Commission.

Electronic communication

The core of the new framework is that it regulates more than just voice related traffic. Rather than focusing exclusively on telephony and leased lines, the 2002 regulatory framework uses the broader concept of ‘electronic communications’. The term is part of the definitions of ‘electronic communications network’ and ‘electronic communications services.’ We now explain these two terms in greater detail.

Electronic communications networks are defined in point 2a of The Framework Directive as “transmission systems and, where applicable, switching or routing equipment and other resources which permit the conveyance of signals by wire, by radio, by optical or by other electromagnetic means, including satellite networks, fixed (circuit- and packet-switched, including Internet) and mobile terrestrial

30 rt. 4 lid 3 Richtlijn 97/33/EG van het Europees Parlement en de Raad van 30 juni 1997 inzake interconnectie op telecommunicatiegebied, wat betreft de waarborging van de universele dienst en van de interoperabiliteit door toepassing van de beginselen van Open Network Provision (ONP), PbEG 1997 L 199/32 31 www.opta.nl/asp/en/newtelecommunicationsact

82

networks, electricity cable systems, to the extent that they are used for the purpose of transmitting signals, networks used for radio and television broadcasting and cable television networks, irrespective of the type of information conveyed.”

Electronic communications services The definition of an ‘electronic communications service’ has been laid down in point 2c of The Framework directive: “a service normally provided for remuneration which consists wholly or mainly in the conveyance of signals on electronic communications networks, including telecommunications services and transmission services in networks used for broadcasting, but exclude services providing, or exercising editorial control over, content transmitted using electronic communications networks and services.” With these definitions, convergence between electronic media has been realised at legal level (Dommering, 2001). See also paragraph 2.2.1. The regulations that most affect broadband network operators are the obligation to provide a universal service and regulation of access and interconnection. These are highly relevant topics for operators with their own access networks as well as for operators who need access to the networks of these operators to reach their subscriber base (i.e. carriers). The national state-owned monopolies, the incumbents, are faced with additional regulation on the markets in which they are considered to have Significant Market Power (SMP). In the following section, we will provide more insight into the obligations concerning universal services, access, interconnection and SMP.

Access & interconnection

As mentioned in Article 3 of the Access Directive (2002/19/EC), the term ‘access’ has a wide range of meanings. It can refer to market access by a new player, access of end-users to certain services and access to the network or facilities of an operator, for example access by an ISP to the infrastructure of a network owning operator (van Duijvenvoorde, 2003, p. 31). The term ‘access’ is defined as follows in article 2, sub a of the Access Directive:

The making available of facilities and/or services, to another undertaking, under defined

conditions, on either an exclusive or non-exclusive basis, for the purpose of providing

electronic communications services.

The article mentions several examples of access, including: � Access to network elements and associated facilities, which may involve the connection of

equipment, by fixed or non-fixed means (in particular this includes access to the local loop

and to facilities and services necessary to provide services over the local loop),

� Access to physical infrastructure including buildings, ducts and masts;

� Access to relevant software systems including operational support systems,

� Access to number translation or systems offering equivalent functionality,

83

� Access to fixed and mobile networks, in particular for roaming,

� Access to conditional access systems for digital television services;

� Access to virtual network services.

Access in favour of interconnection The primary objective of interconnection in to realise interoperability between operators to guarantee end-users the possibility to communicate with everybody. To realise this interoperability by means of interconnection, access is necessary. Not all access is access in favour of interconnection. As a result, interconnection is seen as a specific type of access implemented between public network operators. In Article 2 sub b, it is defined as “the physical and logical linking of public communications networks used by the same or a different undertaking in order to allow the users of one undertaking to communicate with users of the same or another undertaking, or to access services provided by another undertaking.” Article 4 of the Access Directive imposes a general obligation of ‘negotiated access’ via interconnection to all operators of public electronic communications networks, with the purpose of providing publicly available electronic communications services when so requested by other operators (2002/19/EC, Art 4.1). When these commercial negotiations fail and general legislation on access and interconnection proves not to be adequate enough to realise effective competition on the market, the relevant NRA must intervene in the market and is obliged to ensure the end-to-end connection via imposing suitable obligations to these operators controlling end-user access (2002/19/EC, rec.6). Besides access in favour of interconnection, another type of access is access to the network of another operator to provide services. Economic power allows providers of access-controlled services to exercise strategic control over access to essential technological resources. Particularly when powerful, vertically or horizontally integrated operators have exclusive control over bottleneck facilities, they have a wide range of possible ways to hinder potential and actual competitors, ranging from denial of access to disadvantageous conditions, lack of compatibility and software support (Helberger, 2002). Entry barriers are considered to cause most of the competition-related problems in electronic communications markets (Zwenne & Kroes, 2007). The Access Directive provides remedies that are aimed at removing these access barriers. The obligation to provide access in favour to provide services only counts for market players with so called Significant Market Power (SMP). We will now discuss SMP in some more detail.

84

Significant Market Power

The concept of Significant Market Power has its origin in the Open Network Provision (ONP) directives. The starting point that is used in these directives was that a market share of 25% constitutes Significant Market Power. This approach was abandoned with the revision of the regulatory framework in 2002. In these directions, the Commission has opted in favour of a close connection to the concept of economic power as formulated in general competition law, which made Significant Market Power a functional criterion rather than a numeric criterion (Tweede Kamer, 2007). In the Telecommunications Act, Significant Market Power is described in art 14-16 of the Framework Directive (2002/21/EC) and is defined as follows:

An undertaking shall be deemed to have Significant Market Power if, either individually or

jointly with others, it enjoys a position equivalent to dominance, that is to say a position of

economic strength affording it the power to behave to an appreciable extent independently

of competitors, customers and ultimately consumers. (2002/21/EC, Art.14.2) With regard to broadband services KPN is the only party with SMP within the Netherlands. KPN. Cable companies only have SMP for broadcasting services. Determining the status of SMP The 2002 framework introduces a new procedure for the enforcement of obligations on market parties with Significant Market Power (SMP). The basis for decisions as to whether obligations should be imposed on parties with Significant Market Power (SMP) lies in market definitions that are the results of in-depth market analyses to be conducted by NRAs. How these market analyses should be performed is laid down in Article 16 of the Framework Directive. The definition of relevant markets is an intensive and time-consuming process. Questionnaires containing over 3000 questions have been sent out to over 300 parties operating on the markets for mobile telephony, broadcasting, leased lines, broadband and fixed telephony (Zwenne & Kroes, 2007). In turn, the affected market parties have the right to raise objections. Because NRAs have to apply regulation ex ante (in stead of ex post, as is the case in general competition law), the demarcation of relevant markets must also comprise future markets. For this reason, during the process of defining markets, foreseeable economic and technological developments must be taken into account. Moreover, the Commission operates with a limited time horizon of two to three years and is intended to revise its recommendation and the market definitions that it comprises every few years. Each market analysis consists of three elements: � Demarcating the relevant product market and geographical market, based on

general Competition Legislation (Framework Directive, Art.15, par.3)

85

� Assessing the presence of parties with SMP in the requisite market (Framework Directive, Art.14),

� Determining appropriate and proportional obligations to be imposed on these SMP parties (Framework Directive, Art 16; Access Directive, Art. 9-13; Universal Service Directive, Art.16).

Article 15 of the Framework Directive describes the procedure followed by the Commission to define relevant markets. Before the assessment of SMP can take place, the geographical and product market on which an operator may have SMP must be defined. Products or services belong to the same product market when “...those products and/or services are regarded as interchangeable or substitutable by the consumer, by reason of the products' characteristics, their prices and their intended use." (European Commission, 1997a) Undertakings are considered to be part of the same geographical market when “...the area in which the undertakings concerned are involved in the supply and demand of products or services, in which the conditions of competition are sufficiently homogeneous and which can be distinguished from neighbouring areas because the conditions of competition are appreciably different in those areas." (European Commission, 1997a) In recommendation 2003/311/EC, the European Commission has identified and defined 18 relevant markets for products and services that it considers eligible for ex ante regulation. These markets can be clustered in five market areas: mobile telephony, broadcasting, leased lines, broadband and fixed telephony. At the end of 2007, the Commission adopted a new Recommendation on the markets to which telecom-specific regulation should apply, removing several of the 18 above-mentioned markets from the list of relevant markets. For several markets, the Commission “no longer sees an a priori case for sector-specific ex ante regulation by national telecoms regulators. Of the seven remaining relevant markets, two are most relevant to broadband network operators and operators interconnecting on the networks of the incumbent, being the markets of ‘Wholesale access to the local loop’ (previously Market 11) and ‘Wholesale broadband access’ (previously Market 12). Although these market definitions are not binding, they are directional. In case of specific national circumstances, an NRA is allowed to diverge from the recommended market demarcations of the European Commission. In these cases the EC must agree on the NRA’s proposal.32

32 www.opta.nl/asp/aanbieders/marktanalyses/europees-kader

86

Ex ante obligations The European Commission applies the following cumulative criteria to determine whether relevant markets merit consideration for ex ante regulation (Kuipers & Zwenne., 2003): 1. Is the market characterised by high and non-temporary access barriers (i.e.

structural as well as legal/regulatory)? 2. Do competitive conditions and other characteristics of the relevant market

indicate that in the long term there will be genuine competition? 3. Taking the specific characteristics of the electronic communications sector into

consideration, is applying general competition law sufficient or will additional ex ante regulation be necessary?

Article 8 of the Access Directive dictates that NRAs must impose the regulatory ex-ante obligations on undertakings with SMP when effective competition is lacking and when national and Community competition law are not sufficient to address the problem (2002/21/EC, Rec. 27).The ex-ante obligations are described in Articles 9-13 of the Access Directive. It concerns a broadly described and wide-ranging toolbox of possible obligations, covering transparency, non-discrimination, accounting separation, access to specific network facilities, price controls and cost accounting obligations. Which obligations an NRA should apply depends on the specific situation, but all obligations should be proportional and justified..

Unbundled access to the local loop The high cost of duplicating the local access infrastructure prevents new market entrants from entering a market, which affects the level of competition. From January 2, 2001, the Regulation on unbundled access to the local loop (Regulation (EC) No 2887/2000) is in force for all member states. The aim is to increase competition by enabling new competitors to offer high bit-rate data transmission services for continuous Internet access and for multimedia applications based on digital subscriber line technology as well as voice telephony services (unbundled access). The obligation to provide unbundled access to local copper-wire loops only applies to operators with Significant Market Power. Prices for unbundled access must be transparent, non-discriminatory and fair and proportionate to the costs involved. They must, however, deliver a return that ensures the development of the existing infrastructure. The regulation does not apply to new optical fiber loops, a market that is already much more competitive.33 Universal service obligation The obligation to provide universal services is determined in the Universal Access Directive (2002/22/EC), which contains the obligation to provide a “...defined minimum set of services to all end-users at an affordable price” (rec.4). The requirement of universal service provision refers to a connection to the public telephone network at a fixed location and is limited to a

33http://www.ictparliament.org/CDTunisi/ict_compendium/paesi/ue/UnbundledAccessToLocalLoop_Factsheet_en.pdf. Last retrieved January 22, 2008.

87

single narrowband network connection. The connection should be capable of supporting speech and data communications at rates sufficient to allow for access to online services such as those provided via the public Internet (rec.8).

Regulation for cable companies

The Recommendation on relevant markets of the European Commission defined the wholesale market for broadcasting transmission services, to deliver broadcast content to end-users as a relevant market (market 18). With regard to the broadcasting market, the Dutch NRA (OPTA) has determined that all Dutch cable companies have Significant Market Power within their own coverage areas on the retail market for the provision of freely available radio and television (rtv) packages via cable. With regard to the wholesale market, OPTA has determined that the five biggest cable companies (UPC, Essent, Casema, Multikabel and Delta) have SMP for transmission of paid as well as free rtv packages within their own coverage areas. These determinations of SMP for wholesale as well as some on retail level have been approved by the European Commission (Tweede Kamer, 2007). Parties designated as having SMP can be obliged to meet reasonable requests for transmission capacity for providers of TV programmes to offer programme packages. Three other areas of regulation that apply to broadcasting companies are conditional access, electronic programme guides (EPG) and application programme interfaces (APIs). Because the EC recommended in November 2007 that the market for broadcasting transmission (market 18) no longer be considered a relevant market for potential ex ante regulation, OPTA has to undertake new market analyses, in accordance with article 16 of Directive 2002/21/EC (Framework Directive), to determine whether to maintain, amend or withdraw the obligations currently imposed on the cable operators. With this new Recommendation, the Commission has implied that it is most likely that the European broadcasting markets are sufficiently competitive. In case of doubt no regulation may be preferable to regulation, taking the pros and cons of regulation into consideration. General competition law, however, applies to the broadcasting market in its entirety.

The effect of the Act of 2002: a changed environment for NRAs

The implementation of the new directives have brought about some changes for the Dutch regulator. First of all, the manner in which access to networks belonging to parties with Significant Market Power (SMP) can be enforced has been modified extensively, increasing the ability on the part of NRAs to customise regulation. Secondly, a new procedure is introduced to impose obligations on market parties with SMP, based on intensive in-depth market analyses by the NRAs. Furthermore, the broader focus on electronic communication networks and the technology-neutral formulation has increased the scope of the area covered by the NRAs.

88

Also, compared to earlier legislation, general competition regulations now play a more important part in the activities of the NRAs. Finally, consumer protection has been improved.34

Other regulation

In addition to the Telecommunications Act, network operators have to deal with several other pieces of legislation. During the roll-out of infrastructures, for example, operators are faced with permissions to dig and rules from local governments to coordinate digging, to prevent pavements being torn up too often. Moreover, operators have to pay ‘local tax for installation on the public ground’ when installing their equipment on and in the ground. With regard to the provisioning of TV programmes, cable operators are subject to the Media Law.

2.4.3 A second way of public intervention in the broadband market:

national and European stimulation programmes

In addition to market regulation, public involvement in the broadband market is also visible in the form of stimulating broadband roll-out. National and local governments, housing companies and the European Commission are all of the opinion that broadband infrastructures are essential for economic growth and for the knowledge economy (Ministry of VROM et al., 2005, p.5). Several market reports have demonstrated that increasing productivity, creating jobs, growing wages, adding time-related and financial efficiency, creating new or spin-off industries and increasing demand for computer broadband equipment are examples of significant economic benefits as a direct result of the use of broadband technologies (TIA, 2003; Price Waterhouse Coopers, 2004). In addition to these direct effects, there are also indirect effects, like increased e-commerce, less commuting, increased consumption of entertainment, greater efficiency in the distribution of goods, services and information and savings in healthcare (Macklin, 2002). Broadband also has significant social relevance, in that it has the potential to improve the quality of education and healthcare, improve the connectedness between government and society, provide jobs and prosperity, stimulate private investment, boost productivity and modernise public services (Firth & Mellor, 2005, eEurope, 2005).

2.4.4 Action plans to stimulate broadband roll-out

Although the social, cultural and economic benefits of broadband have been demonstrated in several market studies, the roll-out of broadband in the Netherlands and other European countries is considerably lagging behind compared to countries like Japan and Korea (see paragraph 2.2). One of the

34 http://www.opta.nl/asp/en/newtelecommunicationsact

89

problems associated with operating in a liberalised market is that private telecom operators cannot carry the investments for new broadband networks. From their point of view, it is not viable to roll out new fixed infrastructure on a large scale, because the main requested download speed lies between 1 and 2 Mbps (European Commission, 2008, p116) and because there is a lack of services that demand high transmission speeds (above 10 Mbps) (see also paragraph 2.2.2). However, governments want to realise the above-mentioned economic, social and cultural benefits of broadband roll-out. To realise these public benefits, the European Commission as well as Dutch national government have initiated several stimulation programmes for the roll-out of broadband (i.e. fiber) networks. Moreover, there have been several initiatives by regional governments to roll-out fiber networks with public investments. These financial impulses are, however, subject to strict European legislation and possible, in addition to this, to national legislation. In this paragraph, we provide greater insight into European and Dutch action plans aimed at stimulating broadband roll-out.

European Action Plans

In 1993, the European Commission published the White Paper on growth, competitiveness and employment: The challenges and ways forward into the 21st century (COM (93)700, 05.12.93). The White paper calls for the development of services and applications with the aim of attaining a critical mass that is sufficient to warrant further infrastructural investments and to attract new users, adding value to a given network (RAND, 2003). Chapter 5A of the 1993 White Paper focuses on the information society. It makes no mention of eGovernment, but calls for a Task Force on European Information Infrastructures to be set up35 and underlines the importance of the information society as a “boundary condition to maintain economic growth, competition, employment and quality of life for all European citizens in the future” (v.d. Hoven van Genderen, 2003, p.8). As a result of this White Paper, several Action Plans have been launched by the European Commission. The first European Action Plan was the ’eEurope 2002 Action Plan’, which was adopted in June, 2000. One of the main objectives of this action plan was to provide Internet access to every citizen, household, school, company and government agency. In 2002, the second eEurope action plan (eEurope 2005) was presented. In 2005, the third action plan, ’i2010: a European information society for growth and employment’ was launched, as a follow-up of the eEurope 2005 action plan. It underlines the positive contribution information and communication technology (ICT) can make to the economic, social and personal quality of life. We now briefly discuss the various action plans by the European Commission.

35 www.leeds.ac.uk/jmce/bib-egoTv.htm

90

eEurope Action Plans in December 1999, a notification entitled ’e-Europe - An information society for all’ (COM (1999)687) was published and presented to the Lisbon Special European Council in March 2000. The European council agreed to the strategic objective that by 2010, the EU should have become "the most competitive and dynamic knowledge-based economy in the world." The objectives of the so-called 'Lisbon Strategy' were to be achieved via the ’eEurope 2002 Action Plan’, which was adopted in June 2000.36 One of its main objectives was to provide every citizen, household, school, company and government agency with Internet access. In addition, it emphasized that nobody should be excluded, to enhance customer confidence and social coherence. However, by the end of 2002, there were few signs that the plan’s ambitious objectives had been translated into new jobs and services, and not were there any signs of increased productivity elsewhere. Although connectivity had been established, broadband use was clearly lagging behind. As a result, the central focus of the second Action Plan, eEurope 2005, was shifted away from infrastructure towards stimulating use and creating new services (European Commission, 2004c). The second Action Plan, ’eEurope 2005, an information society for all’ (COM (2002)263), was published in 2002 and presented at the Seville European Council in June, 2002. It was based on two mutually reinforcing groups of actions. To begin with, the aim was to stimulate services, applications and content, covering both online public services and e-business. Secondly, attention was paid to the underlying broadband infrastructure and security matters (EC, 2002b). The mid-term review of this action plan (COM (2004)108) took place in 2004. The conclusion was that, although there were successes in a number of areas (rising broadband connectivity, increasing fully on-line government services), in many areas progress continued to be supply-driven, concentrating on technology, applications and initiatives. The review stated that for the remaining part of the programme a move should be made towards a demand-driven approach that emphasises service delivery, end-user value for all and functionality. To realise these objectives, the focus should be on 1) interoperability, standards and multi-platform access, 2) reinforcement of the pan-European dimension, 3) increasing the availability of attractive content, 4) experimentation with new business and service delivery models that get more value out of the shift to e-services, 5) monitoring and quantifying e-inclusion and 6) measuring the impact of e-services in terms of efficiency or productivity gains and quality of work and life (European Commission, 2004c).

36 www.leeds.ac.uk/jmce/bib-egov.htm

91

i2010 Action Plan In 2005, a new action plan, ’i2010: a European information society for growth and employment’ was launched as a follow-up to eEurope 2005. The i2010 action plan has a broader scope than its predecessors and focuses on all electronic communication, services and media sectors (European Commission, 2005). i2010 is the EU policy framework for the information society and media. It emphasises the positive contribution information and communication technology (ICT) can make to the economic, social and personal quality of life.37 Its strategy is based on three main pillars. The first involves completing a single European information space, which promotes an open and competitive internal market for the information society and media. The second pillar has to do with strengthening innovation and investment in ICT research. The final pillar involves the ambition to realise an inclusive European information society that promotes growth and jobs, consistent with sustainable development and prioritising better public services and quality of life (Aandstad, 2005). The i2010 mid-term review, which was published in April 2008, concludes that “in general terms there is a need to bring forward more innovative policies, speeding up action on the interoperability of cross-border eGovernment services, stimulating business take-up and including e-skills strategies within lifelong learning and skills policies” (European Commission, 2008, p.18). In addition, it concludes that “the pattern of information society development has remained largely unchanged: more advanced in the Nordic countries plus the Netherlands and the UK and lagging in many of the countries of eastern and central Europe and the Mediterranean” (p.56). For that reason, the report states that there is “a crucial need to develop European policies that both encourage the competitiveness of the leading countries and address the gaps between the high and low performers, thus countering fragmentation among Member States” (p.18). To realise these objectives, the report mentions several actions for the remaining part of the i2010 programme in three fields: ‘single information space’, ‘ICT research and innovation’ and ‘inclusion, public services and quality of life’ (p.60). An overview of the development of the European broadband stimulation programmes and telecom regulation is shown in Figure 2.14.

37 ec.europa.eu/information_society/eeurope/i2010

92

ONP

Directives

1990 2000 2002 2005 2007

1st eEurope

action plan: Europe

Europe's Way to the Information

Society

19941993

EC White Book

on growth, competitiveness and

employment

eEurope

2002

Action Plan

eEurope

2005: An information society for all

Telecommunications

Act 2002: A common regulatory

f ramework for electronic communications

networks and services

i2010: A European information society for growth and

employment

Figure 2.14: Historical outline of European broadband stimulation programmes and

telecom regulation

In the framework of the ‘Lisbon strategy’, besides stimulation programmes at a European level, programmes were also initiated at a national level. In the Netherlands, two Committees were installed by the Dutch government to advise the government on broadband roll-out and stimulation. Furthermore, a large national public initiative “Kenniswijk” was started and many local public initiatives to roll-out fiber networks were initiated.

Dutch national action plans: Actuating the ‘flywheel’ for broadband roll out

Kenniswijk The aim of the ‘Kenniswijk’ project, which was started in 2000, was to stimulate broadband development. The project was initiated by the Assistant Secretary of State for Transport, Public Works and Water Management, in close cooperation with market parties, social organisations and other (local) governments. It involved providing a neighbourhood in the south of the Netherlands near the city of Eindhoven with a high-quality ICT-infrastructure and innovative products and services in the areas of computers, (mobile) communication and Internet (PKRE, 2002). Kenniswijk is a demarcated area in two municipalities that covers around 38,000 households (Weening, 2006). The aim is to create a residential market with an innovative character and international charisma within two years, in which, development and application of interactive electronic residential services take place, based on high-quality infrastructure. The project was finished in 2005. Although some concrete results have been realised with regard to infrastructure, the original intention of realising an FttH network to 40,000 households was not been realised. None of the infrastructure providers seemed to be able to develop a viable business case at their own expense for rolling out an FttH network. It appeared that the final mile (local loop) was an insuperable barrier. Of the 1400 FttH connections, which were eventually realised as a result of public stimulation, only 175 were actually used by subscribers. The remaining households were not willing to pay for the subscription. As a result, the

93

critical mass required to make the project a success (see paragraph 2.2.2) could not be reached. With regard to broadband services, 17 services were operational and online in 2004. However, more than half of these services were no broadband services. The conclusion may be that, although services and end-user provisions were realised, there was hardly any infrastructure. The chicken - egg problem (see paragraph 2.2.2) could not be solved. The main problems had to do with infrastructure roll-out and an inability to generate sufficient critical mass (Weening, 2006). However, based on the same programme, the roll out in Nuenen has evolved successfully. Broadband expert group In December 2001, the Dutch cabinet installed The Expert Group Broadband to formulate collective starting points for a realistic development of broadband in the Netherlands and to translate these starting points into practical development models. In 2002, the Expert Group published ’Nederland Breedbandland’. The Expert Group determines in this report that achieving a leading role within Europe requires intensive public-private cooperation if the possibilities of broadband towards applications for the benefit of education, healthcare, public administration, safety, culture, labour, trade and leisure are to be translated effectively. Many concrete policy measures concern the Dutch government, among other things initiatives with regard to aggregating demand for broadband connections and stimulating local applications with broadband technology by local governments.38 Although the Expert Group rejects the scenario whereby the government rolls out infrastructure on its own account, it argues that the government can play an important directive role. It should give an impulse to service innovation in the areas of education, healthcare and safety.39 The Expert Group defined four main roles the government could play to stimulate broadband roll-out: manager, demand aggregator, financial stimulator and launching customer (Press release Expertgroep Breedband, 2002). The report calls for public investments of over € 2 billion within four years to stimulate fiber roll-out in the Netherlands. Impulse Committee broadband On March 13, 2004 the Dutch Secretary of Commerce installed the ’Impulse Committee Broadband’. The committee was installed to provide an impulse to the innovative capacity of the Dutch knowledge economy within the framework of the ‘Lisbon strategy’. The Committee should advise the Secretary concerning choices which had to be made to benefit from the opportunities provided by broadband as much as possible. An important aspect was the organisation of the infrastructure market to enable an optimal development of (semi-public) broadband services (Impuls Commissie, 2004). Like the report of the Broadband Expert group, the recommendations by the Impulse Committee

38 http://www.expertgroepbreedband.nl 39 http://www.expertgroepbreedband.nl

94

focused strongly on the encouraging role government could play. The Committee recommends, among other things, that there should be no constraints concerning broadband initiatives by public-private partnerships; the government should stimulate the exchange of knowledge between sectors and encourage the coordination of the development of broadband applications between sectors; the government should act as a role model by intensifying e-government services and the possibilities for e-working; and the government should make broadband a national ‘reputation project’ supported by funds, aspirations and commitment.40 Public local fiber initiatives As a reaction to the reports of The Broadband Expert Group and the Impulse Committee, many fiber pilots were initiated by Dutch municipalities with government subsidies between 2003 and 2006, to stimulate broadband roll-out in the Netherlands. The pilots included fiber city rings and connections between municipal buildings (hospitals, schools, libraries, etc.). In some pilots, fiber to the home was rolled out in residential areas with new housing projects, for example in Rotterdam and Amsterdam. The projects vary in terms of the roles played by municipalities. Also, the local embedding and context vary. Although the municipalities all selected their own specific implementation, roughly speaking four main models were applied (Fijnvandraat et al., 2005): 1) Demand aggregation carrier-owned model (Deventer, Zwolle, Amersfoort), 2) Demand aggregation customer-owned model (Breda, Eindhoven, The Hague), 3) Public-private partnerships (Amsterdam, Almere), and 4) Publicly managed dark fiber model (Rotterdam, Leeuwarden). Most pilots were discontinued after the pilot period, because market parties started upgrading their networks to higher speeds and government involvement was no longer necessary. Although the so-called flywheel effect had been set in motion by these projects, which was the intention of the ‘Expertgroep Breedband’ (Expertgroep Breedband, 2002 p.12), the upgrades of networks did not include FttH technology, but technologies that provided lower transmission speeds, like ADSL. Another reason to terminate the pilot projects was, however, of a different order. The financial involvement of the local governments had many possible legal complications. Commercial telecom operators did – and do – not always agree with the public participation in their market, but consider this public participation as false competition and market distortion. However, participation of public institutions in a commercial market is, however, not always forbidden, although there are certain problems. We now address regulation regarding public involvement in the broadband market.

40http://www.ecp.nl/nieuws/id=566/Impulscommissie_Breedband_adviseert_Brinkhorst.html?PHPSESSID=d755c7bf378a

95

Regulation of public intervention in local Dutch fiber initiatives

Public participation in broadband initiatives must fit within the European and national legislative and regulatory frameworks. With regard to broadband projects, the most important regulatory question lie in the area of state aid (Van Lotringen et al., 2005, p.11). We now briefly discuss these two topics. Finally, we use the case of the FttH project ‘Citynet Amsterdam’ as an illustration. State aid The European Union sees a smoothly running internal market as an essential condition for economic and social progress. By providing state aid, a member state can favour its national undertakings at the cost of undertakings from other member states, which could lead to a distortion of competition. For this reason, state aid must, with the exception from some specific situations, always be reported to ex ante the European Commission. Based on Art 87, par. 1 of the EU treaty, affected parties can determine whether state aid is being provided, after which they can determine whether that support is allowed under a regulatory exemption, which would mean that state aid does not have to be reported to the Commission ex ante. Ex post reporting is, however, always obligatory. Examples of exemption for ex ante reporting are some ways of state aid to medium and small enterprises, research or development projects or education. In the case of broadband initiatives, the so-called ‘de minimis-dispensation’ exemption could apply. Formally speaking, a financial contribution of less than € 300,000 within three years does not constitute state aid. When there is no question of any regulatory exemption, the state aid must be reported to the European Commission for approval. In economic unviable areas providing state aid is possibly allowed by the Euorpean Commission. Investing public money in broadband projects in remote areas as the Pyrenees or Scottish Highlands has been allowed. Within the Netherlands, remote areas similar to these do not exist, nor do underdeveloped areas. For that reason, article 5.14 on State aid of the new Dutch Telecommunications Act states that municipalities are no longer allowed to invest in broadband projects. Running projects can, however, be continued. State aid was a major issue during the Citynet project in Amsterdam. Within this project, the so called ‘market investor principle’ has been applied, which should provide the municipality of Amsterdam the permission to invest in the broadband project. We conclude this paragraph with some more details on this special case. Citynet Amsterdam The roll-out of fiber to the home to residential areas in Amsterdam, known by its pilot name ‘Citynet’, will connect 37.000 households with fiber. The total investment amounts to € 18 million, divided equally between ING, Reggefiber and the municipality of Amsterdam. One of the market players saw the (financial) role of the municipality in the project as distortion of competition and

96

rather than permissible state aid and began summary proceedings involving the municipality of Amsterdam. Amsterdam applies the so-called ‘market investors principle’, in which the municipality of Amsterdam invests in the pilot under exactly the same conditions as would apply to any other commercial market party. The pilot was taken under consideration by the European Commission to investigate possible market distortion and unfair competition. In December 2007, the European Commission concluded that the investments of the municipality Amsterdam were in line with market conditions and for that reason the allegations were dismissed (Planet multimedia, December 12, 2007).

2.5 Conclusion

The Dutch market is characterised by a strong infrastructural competition between upgraded copper (ADSL) and cable (HFC) networks, which from an infrastructure

perspective, constitute a duopoly. However, on both networks, several operators provide access services to these networks, which was made possible through access regulation. Incumbents possessing local loop infrastructures and other essential network facilities have been identified designated as operators with so-called Significant Market Power (SMP) with regard to network access. These operators are subject to several forms of additional regulation. In addition to regulation and the coexistence of two fixed networks with national coverage, serious competition between ADSL and HFC operators was made possible thanks to the convergence of data, voice and visual content. As a result of this convergence, network and service operators no longer service separate markets, but are now tightly interwoven. Traditional network operators have turned into triple-play providers, offering telephony, Internet access and television, significantly increasing the complexity in the market in which they operate. An important consequence of this development is that actors who traditionally operated in separate markets are now competing for the same end-users. The operator that owns the customer owns the customer in total. Customers do not take telephony from operator A and television from operator B anymore, but more and more buy their whole service package from one operator. Due to convergence, issues related to customer lock-in now are much more important than they used to be. The changes in the market and the position within the value chain have consequently affected the internal business processes within the organisations of telecom operators. Not only are departments changed or added when new services are offered, but logistics, billing systems, marketing and supply processes must also be adapted to the organisations’ new position within the value chain. Offering attractive end-user services is, however, very important to operators. Without the broad availability of legitimately acquired services demanding real broadband speeds, there is no need for end-users to take out subscriptions to

97

higher data speeds, and for that reason, there is no trigger for infrastructure providers to upgrade their networks. Basically, this is a problem of the chicken and the egg (see paragraph 2.2.2). Innovation Diffusion Theory (IDT) argues that the adoption of (new) media by later users depends on the presence of other users. When enough users have adopted a technology, it also becomes worthwhile for other people to invest, and critical mass will have been reached. The moment the critical mass of fiber to the home users has been reached, the costs will drop and the benefits will grow, making the provisioning of services commercially viable. Moreover, the telecommunications market landscape is changing radically. The market faces a period of mergers, take-overs by venture capital companies and turbulent financial markets.

In addition to complexity caused by the market landscape, dynamic competition, unpredictable end-user behaviour and necessary changes in internal business processes, decision-makers also face a complex and fast-changing technological environment. Fiber to the home, which is strongly advocated in policy arenas, is a relatively revolutionary development involving major investments. Implementing a fiber optics network is an expensive way of introducing broadband in the local loop. The necessary large initial investments, coupled with a high level of uncertainty about future bandwidth demand and emerging new technologies and a lack of long-term investment capital in the telecommunication market, make fiber optics a relatively unattractive option. In the last 5 years, technologies like ADSL and cable have been rolled out in the Netherlands on a large scale, making the country one of the leaders with regard to these technologies. Although many pilots, projects and studies were carried out between 2000 and 2005, with the aim of stimulating the roll-out of FttH networks, the roll-out of a fiber infrastructure in the Netherlands clearly lags behind compared to countries like Japan, Korea and, within Europe, for example Denmark. The high coverage of fiber infrastructure within Japan and Korea can be attributed to reduced prices, the popularity of advanced services, and differences in culture and architecture, but mainly has to do with active government support. In Europe, public interference in the market is not without problems. Participation in broadband initiatives must fit into European and national legislative and regulatory frameworks. With regard to broadband projects, the most important regulatory questions are in the areas of tendering and state aid. However, it is still not possible for private operators to earn back the high initial investments in FttH within an acceptable time frame and many investors do not find the FttH business cases convincing. This has been demonstrated earlier by Weening (2006) in an evaluation research of the Dutch broadband pilot Kenniswijk.

Signals from the market indicate that at the moment alternative, less risky approaches to the development of broadband are more suitable than approaches involving high risks and huge initial investments. Development paths that are either

98

evolutionary or complementary in nature (see paragraph 2.3) are more likely to suit present market-related and economic conditions and to meet policy objectives. Solutions that allow the existing infrastructure to be reused, at least for the time being, offer a better match to ‘natural’ market developments and should be included in policy discussions. They present the parties involved with fewer hurdles and involve less government intervention. The most likely development paths are the separate developments of evolutionary paths in copper TP and coaxial networks towards FttC, via VDSL. FWA networks can function as temporary solutions, back up connections or in areas where rolling out fixed networks is not viable.

99

2.6 References

Aandstad, Stig Aga (2005). i2010-a European information society for growth and employment. Presentation at the conference on i2010: A European Information Society for Growth and Employment, Perspectives for Central and South-East Europe. Retrieved January 15, 2008, from 2008http://www.mindentudas.hu/attached/20051027aandstad.pdf.

Alleman, J. and Rappoport, P. (2002). Modelling regulatory distortions with real options.The Engineering Economics, 47(4), 390-417.

Androulidakis, S., D. Kagklis, T. Doukoglou & S.Skenter (2004). ADSL2 against SHDSL for symmetric broadband access networking. IEE Electronic letters,

40(16), 1027-1028.

Asscher, Lodewijk (1999). Constitutionele convergentie van pers, omroep en telecommunicatie, October, 1999. Retrieved January 20, 2008, from http://www.ivir.nl/publicaties/asscher/Gw-studie.pdf.

AWARE, (2002). White paper, ADSL2 and ADLS2+, the new ADSL standards. Retrieved October 14, 2005, from http://www.dslprime.com/a/adsl21.pdf.

Bouwman, Harry, Marieke Fijnvandraat and Lidwien van de Wijngaert (2006). Black spots and white holes: developing a conceptual model for broadband roll-out. INFO, 8(1), 72-90.

Church, Jeffrey, Neil Gandal and David Peter Krause (2002). Indirect Network Effects and Adoption Externalities. Foerder Institute for Economic Research Working Paper No. 02-30. Retrieved January 23, 2008, from http://ssrn.com/abstract=369120.

Cnossen, J., (2004). Advance Broadband access Via Cable TV Networks, Ethernet to the home (EttH). Paper presented at Broadband Europe December 10–12, 2004, Brugge. Retrieved October 14, 2005, from https://medicongress.be/UploadBroad/Session%2008/Paper%2008-05.pdf; last

Cuchran, J., & Roka, R. (2001). The evolution of the access network’s infrastructure for the expansion of new broadband services and applications. Proceedings from EUROCON ’01: The international conference on trends in communications, 2 (446–449).

Dialogic Innovatie & Interactie (2003). Breedband en de Gebruiker 2003. Utrecht: Maltha S., F. Bongers, K. Schuurman, R. Vandeberg, K. Vermaas & L. v/d Wijngaert.

100

Dialogic Innovatie & Interactie (2005). Breedband en de Gebruiker 2004-2005. Utrecht: Vermaas, K., S. Maltha, F. Bongers, J. Segers, & L. van de Wijngaert.

Dialogic Innovatie & interactie (2007). Breedband en de gebruiker, 2007. Utrecht: Karianne Vermaas, Jeroen Segers, Hugo Gillebaard, Bram Kaashoek, and Guido Ongena.

Dommering, E.J., N.A.N.M. van Eijk, J.A.M. Nijhof and M.L. Verberne (1999). Handboek Telecommunicatierecht. Den Haag: SDU uitgevers.

Dommering, E.J. (2001).De nieuwe Brusselse Telecommunicatierichtlijnen. Computerrecht, February 2001. Retrieved 20 January, 2008, from http://www.ivir.nl/publicaties/dommering/dnbt.html.

Elnegaard, N. K. (2005). Using broadband wireless access technologies as complimentary or alternative solutions to DSL and cable networks- Service offerings and investment levels. Proceedings of the 16th European ITS conference (CD ROM).

European Commission (1997a). Commission notice on the definition of the relevant market for the purposes of Community competition law. Official Journal of the

European Communities, December 9, 1997. Retrieved January 22, 2008, from http://ec.europa.eu/comm/competition/antitrust/relevma_en.html.

European Commission (1997b). EU Green Book on Convergence of the

telecommunications, communications media, and information technology and

about its consequences on regulation (COM(1997) 623), 3 December, 1997.

European Commission (2000). Regulation (EC) No 2887/2000 of the European parliament and of the council of 18 December 2000 on unbundled access to the local loop. Official Journal of the European Communities, December 30,

2000. Retrieved January 22, 2008, from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2000:336:0004:0008:en:pdf.

European Commission (2002a). eEurope 2005, an information society for all, An

Action Plan to be presented in view of the Sevilla European Council. Retrieved January 15, 2008, from http://ec.europa.eu/information_society/eeurope/2002/news_library/documents/eeurope2005/eeurope2005_en.pdf.

European Commission (2002b). eEurope 2005 executive Summary. Retrieved January 15, 2008, from http://ec.europa.eu/information_society/eeurope/2002/news_library/documents/eeurope2005/execsum_en.pdf.

European Commission (2002c). Directive 2002/19/EC of the European parliament and of the council of 7 March 2002 on access to, and interconnection of, electronic communications networks and associated facilities (Access

101

Directive). Official Journal of the European Communities, April 24, 2002. Retrieved January 15, 2008, from http://eur-lex.europa.eu/pri/en/oj/dat/2002/l_108/l_10820020424en00070020.pdf.

European Commission (2002d). Directive 2002/20/EC of the European parliament and of the council of 12 July 2002 on the authorisation of electronic communications networks and services (Authorisation Directive). Official

Journal of the European Communities, April 24, 2002. Retrieved January 15, 2008, from http://eur-lex.europa.eu/pri/en/oj/dat/2002/l_108/l_10820020424en00210032.pdf.

European Commission (2002e). Directive 2002/21/EC of the European parliament and of the council of 7 March 2002 on a common regulatory framework for electronic communications networks and services (Framework Directive). Official Journal of the European Communities, April 24, 2002. Retrieved January 15, 2008, from http://eur-lex.europa.eu/pri/en/oj/dat/2002/l_108/l_10820020424en00330050.pdf.

European Commission (2002f). Directive 2002/22/EC of the European parliament and of the council of 7 March 2002 on on universal service and users' rights relating to electronic communications networks and services (Universal Service Directive), Official Journal of the European Communities, April 24, 2002. Retrieved January 15, 2008, from http://eur-lex.europa.eu/pri/en/oj/dat/2003/l_186/l_18620030725en00430045.pdf.

European Commission (2002g). Directive 2002/58/EC of the European parliament and of the council of 7 March 2002 concerning the processing of personal data and the protection of privacy in the electronic communications sector (Directive on privacy and electronic communications), Official Journal of the European Communities, July 31, 2002. Retrieved January 15, 2008, from http://mineco.fgov.be/Internet_observatory/pdf/legislation/directive_2002_58_en.pdf.

European Commission (2003). commission recommendation of 11 February 2003 on relevant product and service markets within the electronic communications sector susceptible to ex ante regulation in accordance with Directive 2002/21/EC of the European Parliament and of the Council on a common regulatory framework for electronic communication networks and services, Official Journal of the European Union, 8 May, 2003. Retrieved January 15, 2008, from http://ec.europa.eu/information_society/policy/ecomm/doc/library/proposals/rec_markets_en.pdf.

European Commission (2004a). eEurope 2005 Mid-term Review. Retrieved January 15, 2008, from

102

http://ec.europa.eu/information_society/eeurope/2005/doc/all_about/acte_en_version_finale.pdf.

European Commission (2004b). Directive 2004/18/EC of the European parliament and of the council of 31 March 2004 on the coordination of procedures for the award of public works contracts, public supply contracts and public service contracts, Official Journal of the European Union, April 30, 2004. Retrieved January 15, 2008, from http://www.dkom.si/util/bin.php?id=2004121408020465.

European Commission (2004c). eEurope 2005 Mid-term Review, 18-02-2004. Retrieved May 26, 2008 from http://ec.europa.eu/information_society/eeurope/2005/doc/all_about/acte_en_version_finale.pdf.

European Commission (2007). Press release IP/07/1677: Commission proposes a single European Telecoms Market for 500 million consumers, November 13, 2007. Retrieved January 22, 2008, from http://www.europa.eu/rapid/pressReleasesAction.do?reference=IP/07/1677&format=HTML&aged=0&language=EN&guiLanguage=en.

European Commission (2008). Preparing Europe’s digital future, i2010 Mid-Term review, COM(2008) 199, SEC(2008) 470, Volumes 1, 2, 3, April 2008. Retrieved May 26, 2008 from: http://ec.europa.eu/information_society/eeurope/i2010/docs/annual_report/2008/i2010_ar_2008_en.pdf.

Expert Group Broadband (2002).Nederland Breedbandland, recommendations to

the cabinet from the national expert group. Retrieved May 20, 2008, from http://www.expertgroepbreedband.nl/documenten/NL_Breedbandland_english_version.pdf.

Fijnvandraat, Marieke, Menno Smidts, Pim den hertog, Rudi Bekkers, and Sven Maltha (2005). Breedbandvraagstukken: beschrijvingen vanuit de praktijk. In: Ministry of VROM, Ministry of BZK, Ministry of EZ, IPO, and VNG (2005), Handreiking Goed op weg met breedband (49-79). The Hague: Ministry of Economic Affairs.

Financiële Telegraaf (2002, March 15). Volker Wessels betaalt fooi voor KPN Netwerk Bouw. Retrieved January 1, 2008, from http://krant.telegraaf.nl/krant/archief/20020315/teksten/fin.eur.volker.stevin.kpn.html.

Firth, L. and D. Mellor (2005). Broadband: benefits and problems. Telecommunications Policy, 29 (2-3), 224-236.

103

Foster, K. T., G.Young, J. W.Cook, D. E. A.Clark, R. H.Kirkby and M. G. Booth (1998). Realising the potential of access networks using DSL. BT Technology

Journal, 16 (4), 34–47.

Gagnaire, M. (1997). An overview of broad-band access technologies. Proceedings

of the IEEE, 85(12), 1958-1972.

Leitgeb, E., Gebhart, M., & Birnbacher, U. (2005). Optical networks, last mile access and applications. Journal of Optical and Fiber Communications

Reports, 1(2), 56–85.

Hackenberg, C. (2005). A boost for Ethernet to the home-a new IEEE Standard, 802.3ah, Makes Last Mile Hookups Easier. Broadband properties, February

2005, 30-34. Retrieved October 14, 2005, from http://www.broadbandproperties.com/2005issues/feb05issues/A_Boost_for_Ethernet_Hackenberg.pdf.

Helberger, Natali (2002). Access to technical bottleneck facilities: the new European approach. Communications & Strategies, 46 (33). Retrieved May 1, 2008, from http://www.ivir.nl/publications/helberger/C&S2002.html.

Impulscommissie Breedband (2004). Naar een nationale strategie voor breedband,

Advies van de impulscommissie breedband. Retrieved May 22, 2008, from http://www.ivir.nl/dossier/breedband/naareennationalestrategievoorbreedband.pdf.

Intel (2003). Whitepaper on IEEE 802.16* and Wimax.

Katz, M.L. and C. Shapiro (1985). Network externalities, competition and compatibility. American Economic Review 75 (3), 424-440.

KPN (2007). KPN en Reggefiber gaan krachten bundelen voor nieuw glasvezelnetwerk in Almere (press release 04-10-2007). Retrieved January 11, 2008, from http://www.kpn.com/kpn/show/id=901900/contentid=26134.

Kuipers, D.P. & G-J Zwenne (2003). Marktafbakening ten behoeve van ex ante regulering in de e-communicatiesector., Mediaforum, 2003(1), 2-10. Retrieved January 22, 2008, from http://hdl.handle.net/1887/3721.

Lee, Byung Tak (2008). Broadband Access Prospects and City-scale FTTH Project in Korea. Retrieved May 27, 2008, from http://www.oecd.org/dataoecd/54/10/40461009.pdf.

Leitgeb, E., M. Gebhart and U.Birnbacher (2005). Optical Networks-last mile access and applications. Journal of Optical and Fiber Communication Reports

2 (1), 56–85.

Low, R. (2004). Last mile, light speed. IEE Communications Engineer, 2(4), 36-39.

104

Macklin, B. (2002). The value of widespread broadband. Entepreneur.com, August

13, 2002. Retrieved October 5, 2007 from: http://www.entrepreneur.com/article/0,4621,291780,00.html .

Merrett, R. P., P. V. E. Beastall and S. J. Buttery, (1998). Wireless local loop. BT

Technology Journal, 16(4), 101–111.

Ministry of VROM, Ministry of BZK, Ministry of EZ, IPO, and VNG (2005). Handreiking Goed op weg met breedband. The Hague: Ministry of Economic Affairs.

NetTel (2004). Macro Environment and Telecommunications- Convergence, Chapter 3: Module 2: Global Trends in ICT/Telecoms Sector. Retrieved May 21, 2008, from: http://cbdd.wsu.edu/kewlcontent/cdoutput/tr501/page29.htm.

OECD (2008). Broadband growth an policies in OECD countries- Ministerial background report.

Planet Multimedia (2007, January 30). Reggefiber koopt Telecom Utrecht. Retrieved January 11, 2008, from http://213.75.12.27/planet/show/id=118880/contentid=807502/sc=e634ca.

Price Waterhouse Coopers (2004). The Broadband Future, Interactive, Networked,

and Personalised.

RAND (2003). TEN Telecom Guidelines Status Review, 2003. Rtrieved January 15, 2008, from http://www.rand.org/pubs/monograph_reports/2005/MR1485.pdf.

Romanelli, Rosario P, E.C. (1992). Phone home - telecommunications in the European Community. The Chief executive, March. Retrieved January 16, 2008, from http://findarticles.com/p/articles/mi_m4070/is_n74/ai_12087976.

Rui, Luo, Ti-gang Ning, Tang-jun Li, Li-bo Cai, Feng Qiu, Jian Shui-sheng and Jing-jing Xu (2005). FttH- a promising broadband technology. Proceedings from the 2005 International Conference on Communications, Circuits and Systems, 1, 609-612.

Sari, H. (1999). Broadband radio access to homes and businesses: MMDS and LMDS. Computer Networks, 31(4), 379–393.

Stordahl, K., & Elnegaard, N. K. (2004). Roll-out strategies based on adoption forecasts for ADSL2+/VDSL. Telektronikk, 4, 149–156.

Telecompaper (2007). Dutch broadband 2006, full year update.

105

Telecommunication Industry Association (TIA)(2003). The Economic and Social Benefits of Broadband Deployment. Arlington: TIA. Retrieved October 5, 2005, from www.tiaonline.org/policy/broadband/Broadbandpaperoct03.pdf.

TNO (2002). Capaciteitsaspecten breedband toegang. Den Haag: Burgmeijer, J.W., A.H. van den Ende, W.J. Lubbers, H.Segers, and B.C.B. Vermeulen.

TNO(2007a). Marktrapportage Elektronische communicatie. Delft: Linda Kool and Silvain de Munck.

TNO (2007b), Verkenning van omroepmarkten in Nederland, marktontwikkelingen en beleidsinstrumenten. Delft: Silvain de Munck, Nico van Eijk, Linda Kool, Martijn Poel, and Paul Rutten.

Tweede Kamer (2007). Beantwoording vragen over toegang tot de kabel, letter to the Parliament, kamerstuk van 04-07-2007. Retrieved January 22, 2008, from http://www.minez.nl/content.jsp?objectid=152387.

Van Bussel, J., & H. Rood (2002). Verkenning van de Nederlandse ISP markt voor

netwerktoegang en Internettoegang- Rapport voor OPTA Interconnectie en

bijzondere toegang. Schiphol: Stratix.

Van de Wijngaert (2001). Fysieke en affectieve toegang, geschiktheid; vraag, aanbod en context. In Harry Bouwman (Ed.), Communicatie in de

informatiesamenleving (51-71).Utrecht: Lemma.

Van den Hoven van Genderen R. (Ed.) (2003). E-communicatie. Visie op een

nieuw kader of vissen achter het net? (pre publication). Retrieved Janaury 17, 2008, from http://www.nvvir.nl/doc/ecommunicatie.doc.

Van Damme, Eric, Jos Jansen, Jan Potters, Thijs ten Raa and Vincent Verouden (1998). Universele dienstverlening, Marktwerking ten bate van iedereen. CentER, KUB.

Van Duijvenvoorde, Gera (2003). Toegang. In: R. van den Hoven van Genderen (Ed.), E-communicatie. Visie op een nieuw kader of vissen achter het net(pre publication). Retrieved Janaury 17, 2008, http://www.nvvir.nl/doc/ecommunicatie.doc.

Van Lotringen, Robert-Jan, Ilze Jozepa and Jochem Berns (2005).Breedband en Europa: aanbesteden en staatssteun. In: Ministry of VROM, Ministry of BZK, Ministry of EZ, IPO, and VNG (2005). Handreiking Goed op weg met

breedband (9-47). The Hague: Ministry of Economic Affairs.

Vermaas, Karianne (2007), Fast diffusion and broadening use (Doctoral dissertation, University of Utrecht, 2007).

Weening, Heleen Weening (2006).Smart cities, Omgaan met onzekerheid (Doctoral dissertation, Delft University of Technology, 2006)

106

Wimax forum (2004a). Business case models for fixed broadband wireless access

based on Wimax technology and the 802.16 Standard. Retrieved May 20, 2005, from http://www.wimaxforum.org/technology/downloads/WiMAX-The_Business_Case-Rev3.pdf.

Wimax forum (2004b). WIMAX technology for LOS and NLOS environments. Retrieved May 20, 2005, from http://www.wimaxforum.org/technology/downloads/WiMAXNLOSgeneral-versionaug04.pdf.

Zwenne, Gerrit-Jan (2007). Keuzevak telecommunicatierecht, eLAW Leiden, 2007. Retrieved January 15, 2008, from http://weblog.leidenuniv.nl/users/zwennegj/Kv%20Telecom%20Inleiding%20(sep07)%20DEF.pdf.

107

Chapter 3 Meta-analysis –the

development of a conceptual model for

broadband roll-out

This chapter is an extended version of the paper published as:

Bouwman, H., M. Fijnvandraat en L. van de Wijngaert (2006). White spots and black holes: Developing a conceptual model for Broadband roll-out. INFO, 8 (1), 72-90.

As described in Chapter 1, an integrated multidisciplinary scientific framework or a unifying theory describing the way technological, economic, market-related and regulatory factors affect the migration of broadband networks in the local loop, would be relevant to scientists, strategic decision-makers in broadband and regulators. In existing literature on R&D, innovation and marketing, many models have been developed (for an overview, see De Reuver et al., 2008). Although many technological, regulatory and market-related factors are relevant, it is unclear how these factors are interrelated and which factors play a major role within the different phases of the roll-out process. In this chapter, we examine the available broadband-related literature to investigate whether such a framework or theory for broadband roll-out is at present available. We begin by describing the methodology we used, consisting of Meta-analysis, content analysis and network analysis. In paragraph 3.2, we discuss the results of these analyses and the development of our conceptual model and close by presenting our conclusions in paragraph 3.3.

3.1 Methodology

In this paragraph, we describe the methodology we used in our literature analysis to arrive at a conceptual model for decision-making regarding broadband roll-out. Our methodology includes Meta-analysis, content analysis and network analysis, which we discuss in greater detail in the next section.

3.1.1 Meta-Analysis

According to Light & Pillemer (1984), traditional literature reviews are subjective, scientifically unsound and not very efficient. Meta-analyses are more rigorous and systematic in nature. Meta-analysis is a literature research method that is based on an exact and accurately stated question, paying sufficient attention to the way the data were gathered, interpreted and analysed in the original studies (Bouwman & Neijens, 1991). A Meta-analysis of existing studies is a compound analysis of

108

relationships between variables in earlier studies. (Van den Hooff, Groot & De Jonge, 2005). In this research, the ‘exact, accurately stated question’ to be answered is: “What are the variables and relationships with regard to the decision-

making and roll-out process concerning migration paths towards broadband networks that

were found in earlier studies in this field?”

The aim of a Meta-analysis is to conduct literature research in an objective and scientifically accepted way by extracting data (variables and relationships) from relevant studies and analysing them in a way that can be imitated and replicated (van den Hooff et al., 2005). Meta-analyses are often used in domains where research is firmly developed, theories and hypothesis are explicit, concepts clearly defined, and research methods, measurement instruments and statistical analyses highly standardised, for example in areas like Psychology and Medical research. In other domains, Meta-analyses are less common, for example because there is a lack of rigor in research, a limited emphasis on empirical research, a preference for simulations and a lack of unifying frameworks. In particular in domains where the scientific tradition of hypothesis or model testing is not widely developed and accepted Meta-analyses are few and far between. Starting point of the Meta-analysis in this study is the collection of relevant research projects as published in academic peer-reviewed journals and conference proceedings in the broadband evolution domain. A list of relevant contributions was drawn up by consulting several databases. We used IEEE Xplore and relevant e-journals from commercial publishers like Elsevier Science Direct, Emerald Library, Kluwer, Springer and Wiley to trace and extract articles and papers. We consulted, for example, articles from the European Journal of Operational Research, IEEE Communications Magazine, IEEE Journal on Selected Areas in Communications, Telecommunications Policy and conferences like the International Symposium on Services and Local accesS (ISSLS) have been consulted. To select relevant contributions, we used the following concepts: Broadband, Access networks, Last Mile and Local Loop, in combination with concepts such as evolution, upgrading, migration and roll-out. More general concepts were combined with specific technologies such as Digital Subscriber Line (DSL), Gigabit Ethernet, wireless LAN, Powerline (PLC), MMDS, LMDS, Wimax, FSO and FttX. These concepts had to be mentioned in the title or in the keyword list for us to consider them relevant. Figure 3.1 provides an overview of the selection criteria.

109

Figure 3.1: Article selection criteria

During the initial selection, specific fundamental technological papers about broadband upgrade technology were excluded, because while these papers discuss technological developments and offer insight into possible new technological alternatives for broadband evolution, they do not relate directly or indirectly to the decision-making or roll-out process surrounding network upgrading. Also, policy articles that related to broadband access networks only marginally, for instance, discussing local loop unbundling, were omitted. As a next step we considered whether it was worthwhile to include the papers in the sample based on the summaries of the article or paper. Some papers, for instance, focused on network backbones, such as the high-speed Internet-backbone, and referred only marginally to the local loop and so were not included in the analysis. The final selection contained fifty articles for content analysis (see the separate literature list in Annex E). Based on extensive literature search, we can conclude that the literature we analysed was representative of the entire domain of studies into broadband roll-out and evolution in the local loop.

3.1.2 Content Analysis

As a next step, the articles were subjected to content analysis. From the various possible approaches to content analysis (Krippendorff 1980; Merten, 1983; Weber,

110

1990), we chose an approach that was developed by Van Cuilenburg, Kleijnnijenhuis en De Ridder (1988). This method has proved its value in earlier studies (Bouwman & Neijens, 1991; Van den Hooff, 1997, 2005; Nouwens, 2004). After reading the articles, statements were selected that contained propositions on relationships between relevant concepts. The search focused on clear statements in which a relationship was assumed between an independent and a dependent variable or concept. Although normally the focus would have been on hypotheses that could be tested, it was interesting to see that hardly any of the papers contained clearly formulated hypotheses. In some cases, models were indeed discussed and tested, but in most cases the models were not presented explicitly, which meant we had to select key sentences containing statements about relevant variables and concepts. Every statement was rephrased in a single statement containing an independent and a dependent variable. Based on this singular statement, the type of relationship was coded: causal, conditional (for instance if z =1 than x explains y, if z= 2 there is no relationship between x and y) or similarity (x1 is the same as x2, and are both part of x). The direction (positive or negative) of the relationship was also coded. It was also indicated whether (empirical) testing of the hypothesis actually took place or if it was only a postulation based on conceptual (theoretical) work. In cases where empirical testing did take place, the kind of statistical test that had been conducted was examined. Also, we looked at whether the data had been originally collected or was based on secondary data sources. A final coding of the data took place based on its source (journal, paper presented at a conference, year of publishing, etc.). An example is the following text-fragment from El-Sayed & Jaffe (2002, p. 74): “Market forces such as traffic and subscriber growth, equipment costs reductions and new technologies penetration, have a deep impact on network build outs”. This sentence includes three statements: (1)” traffic impacts network build outs”, (2) “subscriber growth impacts networks build outs”, and (3) “equipment costs impacts network build outs”. At an aggregated level, the concepts ‘traffic volume’, ‘broadband adoption’ and ‘costs network (technological)’ are used as independent variables and network upgrading as the dependent variable. Due to the fact that El-Sayed and Jaffe use the term ‘deep impact’, the relationship was coded as ‘strong’. Because the relationships were not tested empirically in this paper, we coded them as theoretical. No further information concerning empirical testing was added. In content analysis, the reliability of the results is a critical issue. A number of articles (10%) were coded independently by two researchers. The agreement between them with regard to the number of relationships between the core

111

concepts that could be retrieved from the articles and coded accordingly was 82%. With regard to the type of relationship (causal, conditional or similarity) they agreed in 88% of the cases. The direction of the relationship (positive-negative) caused more problems. There they agreed in 58% of all cases. This low level of agreement can be explained by differences in the way the coders framed the concepts: if the concept was formulated in negative terms and the relationship was also negative, this implied that if a positive formulation of the concept were to be used the relationship would be positive. Differences in the exact formulation of the concept between the two coders caused differences in the positive or negative interpretation of relationships. For example, ‘sparsely populated rural area’ and ‘density of population in rural area’ both indicate density, but in the first phrase the formulation is negative whereas the second one is positive. When this concept is related to a dependent variable, it will have an impact on the direction of the relationship. Based on this insight, every formulation of the concept and the direction of the relationship were carefully checked, and this was repeated after clustering the concepts. The agreement on the (lack of) empirical nature of the relationship was high: 94%. Although we did not use reliability measures to correct for change, it was assumed that the content analysis data were reliable.

Table 3.1: Terminology used in Meta-analysis

Field name Content Example

X_VAR Independent variable Traffic volume

Y_VAR Dependent variable Network build out

Relationship Nature of relationship (causal, similarity) Cau/sim

Direction Direction of the relationship, if indicated Strong

Level Measurement level -

T_E Theory or empirically tested relationship Theory

O_S Originally tested relationship, or

secondary analysis

References Study from which the relationship is

derived

El Sayed and Jaffe

Based on the coding, the descriptions found in the original texts were clustered into more general categories. To arrive at a more manageable list of meaningful categories, all 1324 relationships were reviewed in great detail. By clustering the variables, we were able to reduce their number from over 500 to 151. The clustering took place in three major rounds. After each round, the interpretations and clustering of the variables were discussed to improve consistency and coherence. By reviewing and carefully describing the interpretations, we attempted to eliminate subjective interpretations as much as possible. The clustering was based on the grouping of similar variables under a more general heading. Competition in the local loop, competition in networks and competition between network providers, for instance, were all grouped under ‘competition’. After

112

clustering the variables to a higher level of abstraction, we checked whether the direction of the relationship was affected by the way variables were renamed and adjusted the direction of the relationship when necessary. A negative effect of the clustering of variables was that in 101 cases the X and Y variables were given the same name, resulting in var x = var x. Basically, this type of relationship can be considered a similarity relationship. Although this type of relationship is interesting in itself, the analysis did not focus on these types of relationships, which meant that we ‘lost’ 101 relationships. However, for a more detailed analysis of the clusters, which calls for relationships at a lower level of aggregation, the original findings could still be used.

3.1.3 Network analysis

Networks are defined by nodes and ties. Nodes can, for example, be actors (in the case of actor networks). Ties are the relationships between these nodes. Network analysis focuses on the relationships between nodes and ties (Hanneman, 2001). Using concepts and procedures from network analysis makes it possible to analyse the collected data. The concepts (variables) can be considered the nodes in a network of concepts, while the hypotheses or assumptions about the relationship between the nodes represent the links in the network. To describe the data, we applied traditional network concepts. The data acquired on the basis of the content analysis and subsequent clustering were included in a database that served as input for the network analysis. To construct and analyse the networks and analyse the data, Netminer II (version 2.5.0) from Cyram (2004) was used. Netminer is a software programme that applies the spring embedding algorithm of Kamada and Kawai (1989) to construct the networks, the aim being to find a set of coordinates for each pair of nodes. In the algorithm, the Euclidean distance is approximately proportional to the geodesic distance between two nodes. Due to the complex structure of the data, so-called ‘ego networks’ were frequently used. An ego network consists of a focal node and a set of alter nodes from or adjacent to the focal node. A node that is directly related to the focal node has a distance of 1. When the distance is 2, the focal node can be reached through another node. There are various ways to analyse either the network as a whole or parts of the network, i.e. the ego networks. Some of the basic indicators in network analysis are size, density and degree. Size is the number of nodes in the (sub)network. Network density measures the level of connectedness among the nodes in a network and is computed as the number of present lines divided by the maximum possible number of the lines among all of the nodes. Nodes that point to themselves are, logically, excluded from this analysis.

113

Two central indicators in social network analysis are In-Degree (Din) and Out-Degree (Dout). Din relates to the number of lines that are incident to it. Dout refers to the number of lines that are incident from it. Based on the number of Din and Dout relationships for each node, four different types can be distinguished: � Isolates: nodes that are not connected to any of the other nodes (Din = 0, Dout =

0) � Transmitters: nodes that have lines going out, but no lines going in (Din = 0, Dout

> 0) � Receivers: nodes that have lines coming in, but do not have lines coming out

(Din > 0, Dout =0) � Carriers: nodes that have lines coming in as well as going out (Din > 0, Dout > 0) When we apply these types to the Meta-analysis, it becomes clear that, because the data structure is based on relationships between variables, there are no isolates. Transmitters compare to independent variables, receivers are independent variables and carriers are intermediating variables.

3.2 Results: Broadband roll-out- a black hole

In total, we managed to extract 151 variables and 1324 relationships from the papers we analysed. On average, variables were mentioned 8.1 times. Based on this average and a standard deviation of 12.2 for Dout and 9.4 for Din, we can conclude that the variables are distributed very unevenly: a small number of variables is mentioned often, whereas most variables are mentioned only a few times. Figure 3.2 presents the variables mentioned most frequently. The figure indicates the differences between Dout and Din, showing the most important independent, mediating and dependent variables in the dataset. At the top, ‘Network configuration/technological design’ is listed as the most important explaining variable, followed by ‘density’, ‘regulatory framework’ and ‘innovation technology’. At the bottom, various types of costs, revenues, risks and broadband roll-out are presented as important dependent variables. ‘Costs for the technological parts of the network’ is an important mediating variable. Clustering the variables into more general categories, it is clear that most of the 151 variables are related to financial issues (31%), technological issues (22%), demand (16%), service (7%), market-related issues (5%) and regulation (7%). The variable mentioned most frequently in the regulatory domain is ‘regulatory framework’ (36), while ‘government intervention’ is mentioned 15 times and ‘government investment’ is mentioned only 6 times.

114

0 20 40 60 80 100

Network configuration/ technical design

Density

Regulatory framework

Innovation technology

Costs operations/maintenance

Market share of operator

Broadband adoption/ penetration

Competition

Network upgrading

Costs network (technical)

Network quality

Broadband services supply

Broadband roll out

Broadband demand

Network capacity

Risks financial

Revenues

Costs

Transmitters (Dout >) or

Independent variables

Carriers (Dout ˜ Din) or

Mediating variables

Receivers (Din >) or

Dependent variables

Out Degree (Dout)

In Degree (Din)

0 20 40 60 80 100

Network configuration/ technical design

Density

Regulatory framework

Innovation technology

Costs operations/maintenance

Market share of operator

Broadband adoption/ penetration

Competition

Network upgrading

Costs network (technical)

Network quality

Broadband services supply

Broadband roll out

Broadband demand

Network capacity

Risks financial

Revenues

Costs

Transmitters (Dout >) or

Independent variables

Carriers (Dout ˜ Din) or

Mediating variables

Receivers (Din >) or

Dependent variables

Out Degree (Dout)

In Degree (Din)

Figure 3.2: Dependent and independent variables mentioned most frequently

With 101 of the 1324 we found, the relationships contained variables that pointed to themselves. The relationship mentioned most frequently was ‘Costs for operations and maintenance (OAM)’. Of the remaining 1223 relationships, 875 were (72%) different. The causal relationship mentioned most frequently was the one between ‘Network configuration/technological design’ and ‘Network capacity’. This relationship was found 17 times (1% of the total number of relationships). Most relationships (84%) are mentioned only once, which means that in 875 cases a unique relationship was assumed to be relevant in understanding the process of broadband roll-out. This would seem to indicate that there is as yet no coherent conceptual framework. Figure 3.3 gives an impression of what a conceptual network would look like on the basis of these primary findings. This black hole more or less represents the lack of coherent and focused conceptualisation. Most of the concepts are strongly and closely related, indicated by the fact the Euclidean space is densely populated and there are only a few outliers, like availability of wireless networks, choice in favour of an ISP, current service supply and taxes, which can clearly be recognised. The density of the network is .054. Of the

115

potential relationships between the 151 nodes, only 5.4% (N=875) are also found within the network.

Figure 3.3: Explaining broadband roll-out: a black hole

3.2.1 Towards a conceptual model for broadband roll-out

We now turn to the empirically tested networks found in this study, whereby the term ‘empirically tested’ should be taken in a very broad sense. This category includes not only studies where data were collected, but also studies based on simulations.

116

Figure 3.4: Conceptual network explaining broadband roll-out based on direct, empirically

tested relationships between core concepts.

The density of this network (see Figure 3.4) is .43, including 7 nodes and 9 links or, in terms of a conceptual model: 7 variables and 9 hypotheses. In the ego network – the network that only displays the direct relationship between the central concept and the directly related variables – it can be seen that the degree of density has an impact on initial costs, revenues and broadband roll-out. Another explanatory variable is the interconnection costs, which implies that sharing costs by interconnecting may have a positive effect on broadband roll-out. Broadband roll-out itself is directly related to revenues, financial uncertainty and NPV, which indicates that the business case for broadband roll-out is not as financially secure as may be desirable. Including also indirect relationships with broadband roll-out as the core variable, a more complex network emerges (see Figure 3.5). The density of this network is .061, including 45 nodes and 96 links or, in terms of a conceptual model, 45 variables and 96 hypotheses. The most important independent variables on the market side are uncertainty about market developments and demand forecast, while on the technological side network configuration/technological design is mentioned a number of times as an important predictor, and on the cost side there is concern for interconnection costs. If we look at the dependent variables within

117

the network model, it becomes clear that the variables that only have incoming relationships (Din) are all more or less related to the outcome of an economic analysis of the feasibility of broadband roll-out, i.e. cost-effective solution, investments in upgrading and financial uncertainty. Observing the variables that act as intermediaries (Dout and Din), it can be seen that density has a low number of incoming relationships (1) and a high number of outgoing ones (15). The opposite was found for Net Present Value (17 incoming relationships, 1 outgoing) and revenues (10 and 2). If we look at the network more specifically, a clustering around density, costs and network configuration/technological design at the upper right side can be observed. On the lower right side, there is a clustering around uncertainty in terms of market development, competition, NPV and revenues. In between these two clusters, there are concepts that are related to broadband service supply, broadband adoption and broadband roll-out.

Figure 3.5: Conceptual network explaining broadband roll-out based on direct and indirect

relationships between core concepts.

If we compare the theoretical and empirical models for the ego networks of broadband networks two levels deep, it becomes clear that many of the variables were not mentioned in the empirical models, whereas a limited number of variables found in empirical models were not mentioned in the theoretical models. These

118

variables are primarily related to uncertainty, i.e. of business case, costs and forecast. There were 39 variables that were found in both models, which more or less confirms that the model presented in Figure 3.5 is the most valid one from both the theoretical and the empirical perspective.

Figure 3.6: Proposed conceptual model

On the basis of the results from the Meta-analysis, we developed a conceptual model in which it becomes clear that the clusters found in this analysis (see Figure 3.6), such as uncertainty around market developments (risks), cost-related issues surrounding the technological configuration and the density of the network, and expected revenues are placed in the initial phase of the decision-making process. The development of attractive and innovative services can be seen as a condition for the actual adoption and use. Scalability and competition also play a role in generating actual use. The actual revenues depend on the actual use and operational costs, including the recuperation of network-related investments. Although an assessment of an economic trade-off in the early phases is important, it may prove helpful to adopt an approach in which a clear distinction is made between the decision to roll out broadband, the actual roll-out, (monitoring) the actual adoption and use by consumers, and an assessment of the effects. This does not only provide a clear picture of the dynamics involved, including feedback loops between phases (which, for the sake of simplicity, have not been included in Figure 3.6), it also creates the foundation for an evolutionary (real option)

119

approach. In addition, the multi-level aspects (the decision by network providers versus the decision by users to adopt broadband) and different stakeholder roles and benefits (social and economic welfare for the government, strategic advantage and revenues for network providers) become clear.

3.3 Conclusion

To arrive at an initial conceptual model on broadband roll-out, we carried out an extensive literature analysis. We carried out a three-level analysis (Meta-analysis, content analysis and network analysis), on the basis of which we can conclude that, in existing broadband literature there is no coherent high-level conceptual model that explains broadband roll-out. Most of the concepts are strongly and closely related, indicated by a densely populated Euclidean space. This ‘black hole’ more or less represents the lack of coherent and focused conceptualisation. A large number of micro-economic concepts are used that are usually closely connected and related to specific investment methodologies, like Net present Value or Internal Rate of Return. There is a noticeable emphasis on cost-benefit approaches focusing on the costs involved in the implementation and roll-out of broadband networks, while more general objectives, such as strategic value, innovation in technology and services, economic and social welfare are ignored. In general, we can conclude that a middle range theory is missing. In light of the fact that regulation acts as an important framework in which operators have to make decisions, it is interesting to see that regulation plays only a minor role in broadband-related decision-making. Government is mainly seen as an investor in broadband roll-out, and regulatory issues are not explicitly addressed. Technological issues are relevant mainly as cost-related drivers and as alternative investment opportunities. Although specific user needs and requirements have not been examined, it was noticeable that user demand is ignored in empirical studies. In cases where references are made to demand, they are mostly combined with uncertainty, i.e. uncertainty about demand, business uncertainty, market uncertainty and demand forecast. We can conclude that, although technological, regulatory and user aspects play a role, it would appear that the emphasis in assumptions and propositions with regard to broadband roll-out is of a highly economic nature. In the next chapter, we take a closer look at the financial aspects of broadband investments. The conceptual model presented in this chapter incorporates the major findings and adds a more dynamic perspective to the analysis of broadband roll-out. The decision to build the conceptual model around four core phases appears to be a valid on based on the analysis of central role played by the core variables we have chosen.

120

In the phases 1 (decision-making broadband roll-out), 2 (broadband roll-out) and 4 (actual revenues) of the conceptual model, the output of the network analyses provides a very strong indication that the four phases included in the conceptual model are related to aspects that are assumed and in some cases empirically tested in other studies, which gives our model a certain degree of plausibility. The core perspective, i.e. supply side, economically driven, decision-making must also be taken into account. We can assume that, if we were to adopt a more user-oriented perspective, adoption would be a far more crucial question. Generally speaking, it would appear that the concepts we have chosen are valid. It is obvious that the model needs to be tested further and validated. This will be performed on the one hand via qualitative research via interview analysis (see Chapter 6) and, on the other hand, via quantitative research via survey data (see Chapter 7).

121

3.4 References

Bouwman, H. & P. Neijens (1991). Een Meta-analyse van videotext literatuur: een aanzet tot een acceptatiemodel voor de consumentenmarkt. Massacommunicatie,19 (2), 134-148.

Cuilenburg, J. van, J. Kleijnnijenhuis & J. De Ridder (1988). Tekst en betoog naar

een computergestuurde inhoudsanalyse van betogende teksten. Muiderberg: Coutinho.

De Reuver, Mark, Harry Bouwman and Ian McInnes (2008). Business models dynamics for start-ups and innovating e-businesses. Accepted for publication in The International Journal for Electronic Business.

Hanneman, R.A. (2001). Introduction to Social Network Methods. California: University of California. Online book, retrieved from: http://faculty.ucr.edu/~hanneman/SOC157/NETTEXT.PDF

Hooff, B. van den (1997). Incorporating Electronic Mail: Adoption, Use and Effects of Electronic Mail in Organisations. (Doctoral dissertation University of Amsterdam, 1997).

Hooff, B. van den, J. Groot & S. De Jonge (2005). Situational Influences on the Use of Communication Technologies. Journal of Business Communication,

541 (2), 4-27.

Krebs, Valdis (2005). An Introduction to Social Network Analysis. Retrieved August 2nd, 2005 from: http://www.orgnet.com/sna.html.

Krippendorff, K., (1980). Content Analysis, An introduction to its Methodology. Newbury Park: Sage Publications.

Light, R.J. & Pillemer, D.B. (1984). Summing up. The science of Reviewing

Research. Cambridge: Harvard University Press.

Nouwens, J. ( 2004). The Outside In. Questioning the use of electronic information services in organizations. (Doctoral dissertation University of Amsterdam, 2004).

Weber, R. P. (1990). Basic Content Analysis. Newbury Park: Sage Publications.

122

123

Chapter 4 Infrastructural investments:

dealing with uncertainty and irreversibility

One question that keeps arising is that of the return on investment. As described in Chapter 1, a problem related to infrastructural investments is the necessary high upfront investment in combination with long payback periods and uncertain demand. As a result, telecom operators are unlikely to make major and long-term investments. In light of the fact that it is, ultimately, the investment-related costs and revenues that determine the payback time of network investments - and, by extension, the annual revenues of the operators - the financial perspective is a highly important one. In Chapter 3, we concluded that, although technological, regulatory and user-related aspects play a role, the emphasis in the assumptions and propositions with regard to broadband roll-out is largely economic in nature. This would appear to make sense, in light of the fact that it is, ultimately, the investments and revenues that determine the payback time of network investments and, by extension, the annual revenues of operators. In this chapter, we first describe the traditional and commonly applied methods that are available to telecom operators to assess the business value of corporate investments in the local loop infrastructure. Next, we address a financial assessment method called Real Options Analysis (ROA). Although the relevance of this method to decision-making in the telecommunications sector has been demonstrated within literature, thus far the method has not been applied in practice on a large scale. An important factor operators must take in consideration with regard to investing in a specific technology for their network upgrades is the level of risk and uncertainty associated with this technological alternative in relation to aspects like price development, expected demand, future-proofness, etc. These risks and uncertainties affect the likelihood of investments being earned back within the expected timeframe. As a result, risk and uncertainty assessment is an important aspect of network investment-related decisions. After discussing financial theory, we take a closer look at theory regarding risk and uncertainty, providing a framework to measure the source, level and nature of risks and uncertainties. Moreover, several assessment methods that can be applied by operators to manage these risks and uncertainties will be discussed.

124

4.1 Financial methods to assess business value of

investments

Remer et al. (1995) define five basic methods to assess investment projects: Net Present Value methods, Rate of Return methods, ratio methods, payback methods and accounting methods. In the next paragraphs we discuss the methods that are applied most frequently in practice: discounted cash flow and ratio methods. We investigate the actual use of these assessment methods by telecom operators during our qualitative and quantitative research (see Chapters 6 and 7).

4.1.1 Discounted cash flow methods

Net Present Value (NPV) and internal Rate of Return (RoR) are often referred to as ‘Discounted Cash Flow’ (DCF) methods. From a financial perspective, they are considered to be superior to other financial methods because they take the time value of money into account, which means that the value of revenues will be less when they are earned further into the future (Renkema, 2000, p.107). We now take a closer look at NPV and RoR. Net Present Value The Net Present Value rule has emerged as the dominant factor governing investment-related decisions (Keswani & Schackleton, 2004). On the basis of NPV, a potential investment project should be undertaken if the present value of all cash inflows minus the present value of all cash outflows (which equals the Net Present Value) is greater than zero. If the Net Present Value is equal to zero, the investor is indifferent to the project. In case two or more projects are considered, the project which has the greater present value is usually selected (Remer & Nieto, 1995a).

Rate of return methods The internal rate of return method (IRR) is the threshold at which, after discounting the incoming and outgoing cash flows, the present value of a project equals zero (Renkema, 2000, p.107). The term “internal” implies that the interest rate only represents “internal” limits to the cash flow. This method is also known as the rate of return method, discounted cash flow rate of return, interest rate of return, return on investment, investor’s method, break-even rate of return and profitability index (Remer & Nieto, 1995a). As an investment decision tool, the calculated IRR is used to rate alternative investments. The investment alternative with the highest IRR is preferred. The criteria for accepting or rejecting a project depend on the available minimum attractive rate of return (MARR). If the calculated IRR is bigger than the MARR, a project is acceptable. If the calculated IRR equals the MARR, the investor remains indifferent to the project. In all other cases, the project is rejected (Remer & Nieto, 1995a).

125

4.1.2 Ratio methods

A second frequently applied group of financial assessment methods is that of the ratio methods. Ratio methods assess investment alternatives on the basis of the relationship between criteria. The resulting ratio score for an alternative can be compared to target values or benchmarks (Demkes, 1999). Examples of meaningful ratios are project expenditures versus total turnover and all earnings that can be attributed to a project versus the total profits. Ratios do not by definition take only financial numbers into account. Product expenditures can also be related to, for example, the total number of employees or some output measure, like products or services (Renkema, 2000). Two ratio methods that are well-known and applied frequently are the Return On Investment (ROI) method and the benefit/cost (B/C) ratio method. Other ratio methods are the methods of premium worth percentage analysis, profit-to-investment, savings-to-investment, return on management and cost effectiveness. Below, we describe the ROI and B/C methods in greater detail.

Return on Investment (ROI) The ROI method has two versions, the return on original investment and the return on average investment. The return on original investment measures a percentage-based relationship between the average annual profits and the initial investment. The return on average investment measures a percentage relationship between the average annual profits and the average annual investment. The project with the highest ROI is considered the most favourable. Because these methods do not consider the changing value of money over time (i.e. the calculated averages are not discounted), one project could incorrectly be favoured over another. For this reason, the ROI method should always be applied in combination with another assessment method (Remer & Nieto, 1995b).

Benefit/Cost ratio The Benefit/Cost ratio method has its origin in US Congressional legislation, which stated that the criterion for accepting a project was that the ‘benefits’ should outweigh the ‘costs’. As long as the costs and benefits are clear, the benefit/cost ratio provides simple results. However, it is not always possible to express benefits in monetary values. The ratio is commonly written as ‘Benefits divided by Costs’ (B/C) or ‘Benefits minus Costs’ (B-C). If a proposed project results in a reduction of benefits, benefits are defined to be negative. The same goes for negative costs: they arise when a project results in cost reduction (Remer & Nieto, 1995b). In addition to these traditional assessment methods, a great deal of research focuses on the techno-economic field related to another financial assessment method: Real Option Analysis (ROA). Although real options theory is increasingly

126

used in other industries, it has not yet been applied on a large scale in the telecommunications industry (Alleman, 2002). This is striking, because ROA is relevant to telecommunications in three areas: strategic evaluation, estimation and telephony cost modelling (Alleman, 2002). As Alleman (2002, p.91) concludes:”managers cannot afford to ignore the implications and methods developed by Real Options Analysis”. In the next paragraphs, we take a closer look at the origins of the Real Options Analysis and the method itself.

4.2 From financial options to ‘real options’

The options theory has its origins in the financial world in the pricing model of Black and Scholes (1973). In the financial world, options represent the right to buy (call options) or sell (put options) a financial value, mostly a stock, for a predetermined price (the exercise price), without being obliged to do so (De Ruijter & Janssen, 1996). Options provide the possibility to buy insurance against increases or decreases of the stock value. With call options, one "insures" oneself against a price increase, while with a put option, one insures oneself against a price reduction (De Ruijter & Janssen, 1996). Modern financial theory applies this approach to real world projects: the real

options thinking. Thinking about how future options affect the value of projects has, therefore, come to be known as the area of real options. In this approach, real options are based on the same principals as financial options. To have a ’real option’ means to have the possibility for a certain period to either choose in favour or against something, without committing oneself in advance (De Ruijter & Janssen, 1996). The idea of real options was first mentioned and discussed by Myers (1977). At the heart of the real option approach is the notion that the values of real resources vary over time in ways that cannot be fully predicted (March & Shapira, 1987). The value of an option is largely determined by market conditions, not by technological performance (Neufville, 2001). In the last decade, finance researchers have developed Real Option Analysis as a way to value investments under conditions of uncertainty (Miller & Waller, 2003). Real options valuation can be used when the following three conditions apply (Weeds, 2002): the future is uncertain; the (investment) decision is (partly or entirely) irreversible; the firm holding the (investment) option has the ability to delay the investment.

4.3 Types of real options

Alleman (2002) describes seven forms in which managers possess flexibility: defer, abandon, shutdown and restart, time-to-build, contract, switch, expand and growth. They form the several types of options in Real Options Analysis. Four types of options are embedded in investments, the option to expand, delay and abandon an

127

investment and the option to add flexibility. Next, we discuss these four types of options in greater detail. Option to expand An option to expand provides management with the ability and right to expand into different markets and products and to diversify its operations. It makes it possible to take advantage of future, interrelated opportunities and to expand if developments are better than expected (Alleman, 2002). It provides the opportunity to make further investments in existing products and to increase output and market share (Hermantzis & Tanguturi, 2004). In some cases, investing in projects allows firms to take on other projects or to enter other markets in the future. Option to delay Not making an investment now provides the opportunity to do so later or to adjust the initial decision when market opportunities are more favourable or more information is available that could affect the desirability of the investment. This flexibility is clearly important in the evaluation of many investment opportunities under conditions of uncertainty, because it gives decision-makers the option to discontinue a project if market-related conditions turn out to be unfavourable and wait conditions improve (Smit, 2003). On the other hand, investing earlier may have strategic benefits, like first-mover advantages and early cash inflows. These advantages must be weighed against the loss of flexibility (Smit, 2003). This option also makes it possible to accelerate when new information becomes available (Moro & Carvalho). Option to abandon This option allows decision-makers to avoid or minimise losses with unsuccessful investments by abandoning or terminating a project before the original time limit for termination (Moro & Carvalho). This option normally leads to an improvement of the Net Present Value. Using this option is useful in case a company does not have the option to delay an investment. Abandoning the project means that no operating costs will be incurred, but also that no cash flows will be received. It allows decision-makers to hedge themselves against the downside risk of an investment and wait until conditions are right to re-enter the market. Option for flexibility This option provides the possibility to add flexibility to the business operations, especially in the area of production. Instead of construction of plants or infrastructures demanding long periods of investments being able to meet demand immediately, companies can work with alternative sources of input, using flexible technologies of production, a new configuration of installations, equipment and acquisitions (Moro & Carvalho). In the telecom market, this kind of flexibility can be seen in the use of alternative local loop technologies like ADSL and ADSL 2+ or wireless broadband networks, rather than rolling out national FttH networks that demand long investment and roll-out times.

128

The broadband infrastructure market is characterised by uncertainty, risk, major investments (that can be spread sometimes) and quick developments in the environment (Eurescom, 1999). The real options methodology focuses on overall project risk and recognises the value of deferring, contracting or expanding the initial investment opportunity, as part of the most likely options for an investment project in a competitive environment with a great deal of uncertainty with regard to technology evolution and market response (Eurescom, 1999). The ROA method explicitly recognises the value of flexibility and the added value associated with options in the context of uncertainty and adds values to different future possibilities. Traditional methods like he NPV method, on the other hand, discount the NPV because of uncertainty (Neufville, 2003). Because the level of risk and uncertainty clearly has a significant effect on investment assessments, and with it on the expected payback time of these investments, insight into risks and uncertainties are very important to decision-makers on broadband roll-out. In the last part of this chapter more insight will be given in risks and uncertainties. Paragraph 4.4 defines risk and uncertainty. In this paragraph, we introduce a framework to categorise risk and uncertainty. In paragraph 4.5, we examine several ways to assess risk and uncertainty.

4.4 Risk and uncertainty

Risks and uncertainties are inherent in large infrastructural projects, and include high initial costs, long periods with regard to recapturing the return of investment, irreversibility, indivisibility, high entry and exit thresholds, and inelastic demand. Economic, political, social and other related risk issues have been recognised as crucial criteria for investment-related decision-making (Piyatrapoomi et al., 2004). Factors causing uncertainty are the (unknown) behaviour of competitors, the (unknown) development of end-user demand, the rapid development of new broadband technologies and the different development paths available to operators (Fijnvandraat & Bouwman, 2006).

4.4.1 Uncertainty

The term “uncertainty” emphasises that decisions must be made on the basis of incomplete knowledge about projects that do not yet physically exist (Walker et al., 2003). Walker et al. (2003) define uncertainty as “any deviation from the unachievable ideal of completely deterministic knowledge of the relevant system.” This broad definition covers uncertainty that can be reduced by increasing knowledge as well as uncertainty that cannot be reduced by knowledge and that is therefore indeterminable (Meijer et al., 2006). Walker et al. (2003) divide uncertainty into three dimensions: nature, location and level. Meijer et al., (2006) add ‘source’ as a fourth dimension of uncertainty.

129

Nature The nature of uncertainty describes its origin. The question is whether uncertainty is caused by imperfect knowledge or to the inherent variability of the uncertain variables. The nature of uncertainty is divided into epistemic uncertainty

and variability uncertainty, whereby epistemic uncertainty is defined as “uncertainty

due to the imperfection of our knowledge, which may be reduced by more research

and empirical efforts”. Meijer et al. (2006) call this knowledge uncertainty. Variability uncertainty is defined as “uncertainty due to inherent variability of the

system (changes of the environment), which is especially applicable in human and

natural systems and concerning social, economic and technological developments”

(Walker et al., 2003). Knowledge (or epistemic) uncertainty can be overcome by gaining more knowledge by, for example, forecasting studies and market research. An example is the uncertain economic value of new technologies. Variability uncertainty, however, cannot be overcome this way. Uncertainty regarding the actions of a competitor is an example of variability uncertainty. For this reason, it is important for managers or decision-makers to determine the nature of uncertainty (Meijer et al., 2006). Location The location of uncertainty reflects where uncertainty manifests itself within a model. It describes uncertainties concerning model boundaries (context uncertainty), variables and relationships (model uncertainty) and the reference system (input uncertainty). Because we focus on the uncertainty that is experienced by telecom operators (the object) ‘location uncertainty’ will not be included in this research. Level The level of uncertainty indicates how uncertain a situation is and covers a spectrum from complete deterministic understanding to total ignorance (Walker et al., 2003). Meijer et al., 2006 use the continuum low-high in their uncertainty matrix. Courtney et al. (1997) distinguish between four levels of uncertainty. Level 1 (a clear-enough future) is the situation in which decision-makers can develop a single forecast of the future. At level 2 (alternate futures) there are several alternate outcomes for the future. At level 3 (a range of futures) the future is not one of a range discrete scenarios, but lies along a continuum of potential futures. Level 4 (true ambiguity), finally, is an interaction of multiple dimensions of uncertainty that creates a future that is virtually impossible to predict. At least half of all strategy-related problems fall into levels 2 or 3 (Courtney et al., 1997). Because these 4 categories are more concrete than a continuum, we include them in our typology. Source The source of uncertainty is the domain of the organisational environment about which the decision-maker is uncertain (Milliken, 1987). To cope with uncertainty, decision-makers need to choose appropriate strategies. For this reason, distinguishing between different sources of uncertainty is important (Wernerfelt & Karnani, 1987). The source of uncertainty is context-dependent

130

(Meijer et al., 2006). Two known theories about defining the environment of companies are the PEST analysis model (Johnson & Scholes, 2002, p.102) and the Five Forces of Competitive Position model by Porter (1980). The PEST model discusses political, economic, social and technological uncertainties; the Five Forces model of Porter discusses uncertainties such as the behaviour of new entrants, substitutes in the market, buyer power, supplier power and rivalry. Meijer et al. (2006) draw the following distinction with regard to sources of uncertainty: technological, resource, competitive (new entrants & rivalry), supplier, consumer (buyer power) and political/regulatory. These sources of uncertainty are sources that are mainly being discussed in organisational management and innovation literature (Meijer et al., 2006). In the following section they will be discussed briefly. Technological uncertainty Meijer et al. (2006) distinguish three types of technological uncertainty. The first type is uncertainty about the characteristics of the new technology. Several characteristics of new technologies determine the choice in favour of a technology (see Chapter 2). When a new technology emerges, information is limited and the characteristics of the technology will keep changing while further improvements are made (Meijer et al., 2006). In the broadband market, this can clearly be seen in the developments within ADSL technology, resulting in several successions of technologies, i.e. ADSL, ADSL2, ADSL2+, SDSL, VDSL, etc. These continuing changes complicate decisions and make characteristics like, for example, future-proofness very uncertain. Since incremental innovations constitute relatively minor changes, it is easier for decision-makers to forecast the performance of the innovation in terms of the perceived innovation characteristics than in the case of radical innovations (Meijer et al., 2006). The second type of technological uncertainty Meijer et al. mention is uncertainty about the relationship between the new technology and the infrastructure in which it is embedded, i.e. uncertainty to what extent adaptations to the infrastructure are needed. Some new technologies only demand small adaptations to the existing infrastructure (for example the change from DOCSIS 1 to DOCSIS 2). Other technological changes demand, however, large infrastructural adaptations (for example the transition from ADSL2+ to VDSL) or, for example in the case of FttH, an entirely new access network. Due to the necessary large investments and long life spans associated with infrastructural adaptations, they are difficult to implement. Uncertainty about the extent of necessary adaptations as well as uncertainty about the effect these adaptations have on the existing network (for example irreversibility) influence the decision-making process on broadband roll-out. The third technological uncertainty Meijer et al. (2006) mention is uncertainty about the possibility of choosing alternative (future) technological options. As we saw in Chapter 2, operators can choose between many (combinations of) competing technological options to upgrade their existing networks. This uncertainty is

131

increased due to the ongoing process of technological changes and the possibility for operators to wait for the arrival of a new generation of technologies. Other uncertainty-increasing factors are the speed with which new technologies arrive, the degree of improvement of these technologies as compared to the existing ones and uncertainty about the possibility that future technologies can cause existing technologies to become obsolete (Meijer et al., 2006). A final important aspect is the timing of roll-out and upgrades. Investing too soon involves the risk of having sunk cost but no adoption because the market is not ready yet. There is a risk that competitors copy your technology against lower costs the moment the market is ready. This risk must be balanced against the possible first-mover advantages. On the other hand, by investing too late operators run the risk of losing the market to their competitors. Resource uncertainty Meijer et al. (2006) define resource uncertainty as uncertainty about the amount and availability of raw material, human resources (knowledge and skills needed), financial resources (how much capital to invest) needed for innovations and the best way to organise the process. This type of uncertainty is caused by the difficulty of making accurate forecasts regarding the necessary resources and its level will be higher for radical than for incremental upgrade processes. Competitive uncertainty: Competitive uncertainty relates to uncertainty about the behaviour of (potential or actual) competitors and the effects of this behaviour on the competitive position of the own firm (Meijer et al., 2006). Wernerfelt & Karnani (1987) mention five factors being very important within competitive strategy under conditions of uncertainty: first-mover advantages, economies of scale, the number of competitors, the relative competitive position and past investments of the firm.

Supplier uncertainty Uncertainty about the actions of suppliers levels to uncertainty about the timing, quality and price of delivery (Meijer et al., 2006). Supply networks (including various types of inter-company relationships) have become very complex and vulnerable to various risks (Halikas et al., 2005). The sources of this complexity are, among other things, globalisation, increasing product/service complexity, outsourcing, e-business and demanding customer needs (Harland et al., 2003). These trends have made companies increasingly dependent on the risks of their external resources. Supplier uncertainty becomes increasingly important when the level of dependence on a supplier is high. When that is the case, suppliers have greater bargaining power and are able to create uncertainty with regard to the quality and price of their products (Meijer et al., 2006). Chopra and Sodhi (2004) have identified categories and drivers of supply chain risks. Their risk categories include disruptions, delays, system forecasts, intellectual property, procurement, receivables, inventories and capacity, each of which may have

132

several variations with regard to their source and type of impact (Halikas et al., 2005). Consumer uncertainty: According to Meijer et al. (2006), uncertainty about consumers can be attributed to three uncertain sources. The first is consumers-preferences with respect to the new technology: will they adopt a new technology or not? Operators perceive uncertainty about the market opportunities of a new technology due to uncertain preferences of potential customers with regard to product characteristics, quality, prices and distribution channels. Although operators can forecast several of these aspects, also forecasts themselves are an important source of uncertainty (Piyatrapoomi et al., 2004) (See also paragraph 4.5). The second source of consumer uncertainty is end-users’ characteristics. The compatibility of new technologies or services affects the chances of adoption. A problem is that operators strive to adoption by the so-called ‘critical mass’ (see also Chapter 2) whereas it is very difficult, if not impossible, to generalise end-user characteristics. As Vermaas (2007) states: “the end-user does not exist.” As a third source of consumer uncertainty, Meijer et al. (2006) mention the uncertainty of long-term development of demand over time. This uncertainty includes uncertainty regarding the size of demand and whether demand will stabilise (Meijer et al., 2006). This uncertainty is partly caused by general macro-economic developments. Political/regulatory uncertainty According to Meijer et al. (2006), political or regulatory uncertainty includes uncertainty regarding government behaviour, regimes and policies. They mention six underlying causes for regulatory uncertainty: existing policies, unclear or inconsistent regulation, a lack of regulation, future changes in regulation which occur rapidly or are hard to predict and uncertain government behaviour in general. The last one is caused by, for example, contradictory political statements or confusion about a mix up of the public and private role of governments in, for example, FttH pilots (see also Chapter 2).

4.4.2 Risk

In traditional decision-making theory, risk is most commonly conceived as reflecting variation in the distribution of possible outcomes, their likelihoods and subjective values. Risk is measured by the variance of the probability distribution of possible gains and losses associated with a particular alternative. A risky alternative is one for which the variance is large. Expected value is assumed to be positively associated with the attractiveness of an alternative, while risk is, in contrast with uncertainty, assumed to be negatively associated (March & Shapira, 1987). In most literature, the concept of ’risk’ is defined as consisting of two elements, i.e. the probability (or likelihood) of occurrence of a negative event during the lifetime of

133

operation of a facility and the consequence of this negative event (Berdica, 2002). The former relates to risk assessment, the latter to risk management (Piyatrapoomi et al., 2004). However, the evaluation of different influencing factors and the consequences of risk depend on the perception of the actors involved. Hence, measuring risk involves considerable uncertainty, caused by various degrees of subjectivity, which makes it more difficult to obtain an objective composition of the two (Berdica, 2002). There are several additional complications with theoretic concepts of risk that are used for the analysis of actual processes underlying choice behaviour (March & Shapira, 1987). The ways in which actual decision-makers define risk may differ significantly from the definitions in literature, as well as perceptions of risk will vary between individuals. Managers see risk in ways that are less precise, but that are also different the way risk is perceived in decision-making theory (March & Shapira, 1987). The key point is that dealing with risk is a balancing act that should address both positive and negative aspects, keeping in mind the likelihood and subsequent consequences of any defined event (Dalgleish et al., 2005). Most risk classifications apply to medical and environmental research and for that reason did not provide a useful ontology for decision-making in broadband upgrading, which is why we decided to use the same approach to classification of risk we used for uncertainty, classifying risks according to their nature, level and source. The risks of infrastructural investments are substantial, hard to assess beforehand and difficult to manage in financial, technological and organisational terms. (Renkema, 1999, p. 9).

4.4.3 A typology framework for risk & uncertainty

The source, dimension and scale of risk and uncertainty as discussed in the previous paragraphs can be summarised in a typology framework. Source includes the six sources described in paragraph 4.1.1. Dimension includes the nature and level of risks and uncertainties. These dimensions are measured on a different scale. The scale for nature is nominal: epistemic (knowledge) versus variability uncertainty. The level of uncertainty is measured on a nominal scale from 1 to 4 (1: a clear-enough future, 2: alternate futures, 3: a range of futures, 4: true ambiguity). The risk and uncertainty framework presented in

Table 4.1 is based on Meijer et al. (2006), Walker et al. (2003) and Courtney et al. (1997). Meijer et al. (2006) use this framework merely to categorise uncertainty, but also risks can be placed into this framework.

134

Table 4.1: Typology of perceived risks and uncertainties (to Meijer et al., 2006, Walker et

al., 2003 & Courtney et al., 1997)

Source Dimension Scale

Nature Epistemic vs.

variability

Technological:

- uncertainty about the technology itself

- uncertainty about the relationship between the

technology and the technological infrastructure

- uncertainty about the availability of alternative

technological solutions

Level 1,2,3,4*

Nature Epistemic vs.

variability

Resource:

- uncertainty about the amount and the availability

of financial resources

- uncertainty about the amount

and the availability of human resources

Level 1,2,3,4*

Nature Epistemic vs.

variability

Competitive:

- Uncertainty about the actions of (potential or

actual) competitors

- Uncertainty about the effects of competitor’s

actions

Level 1,2,3,4*

Nature Epistemic vs.

variability Supplier:

- uncertainty about the timing, quality and price of

the delivery Level 1,2,3,4*

Nature Epistemic vs.

variability

Consumer:

- uncertainty about consumers’ preferences

- uncertainty about consumers’ characteristics

- uncertainty about the development of demand Level 1,2,3,4*

Nature Epistemic vs.

variability

Political:

- unclear or inconsistent regulation

- lack of regulation

- future changes in regulation

- government behaviour

Level 1,2,3,4*

* 1: a clear-enough future, 2: alternate futures, 3: a range of futures, 4: true ambiguity

The applicability of this theoretical framework is validated in Chapters 6 and 7. In Chapter 6, we validate the categorisation in source with qualitative data analysis. In Chapter 6, categorisation into Level and Nature will be measured based on quantitative data analysis.

135

4.5 Reducing risk and uncertainty

Most uncertainties cannot be eliminated, but must be accepted, understood and managed. There are several theoretical approaches to dealing with risk and uncertainty. The first (and traditionally the most commonly used) approach is to ignore it, implicitly assuming that the future world will be structurally more or less the same as it is today (Walker & Marchau, 2003). Decision-makers can also try to deal with risk and uncertainty. They can try to reduce or minimise risk and uncertainty. Several methods are available to reduce risk and uncertainty. Examples include scenario investigation, sensitivity assessment and probability-based assessment (Piyatrapoomi et al., 2004). There are other methods available to assess and reduce risk and uncertainty as well. We describe scenario investigation, forecasting and several ways of obtaining information, i.e. gaining information from suppliers, technological analysis, cost-benefit analysis and benchmarking with competitors. Scenario investigation The future is uncertain and risky. Constructing possible scenarios and looking for options that perform reasonably well with minimum risk is one way to deal with uncertainty and risks (Piyatrapoomi et al., 2004, Walker 2000). Scenario assessment is a basic tool used to assess risk and uncertainty with regard to future forecasts. Scenario planning creates the possibility to identify, evaluate and act upon contingencies, uncertainties, trends and opportunities that are often unanticipated. Moreover, it encourages managers to think about how to take advantage of opportunities and avoid potential threats (Miller et al., 2003). The creation of flexible business cases, in which input variables can be altered and scenarios introduced is used by telecom operators to incorporate flexibility into their decision-making. This way, worst case scenarios can be chosen to check whether an option is financially feasible. Forecasting The planning of a telecommunication network is based on the estimation of future needs. For this reason, forecasting in telecommunication is essential for successful planning. It is, however, a complex procedure that involves great number of factors affecting the network development (Acimovic-Raspopovic, 2003). Telecom operators use forecasting to determine the impact of developments in competition, equipment costs, broadband development and demand for their own position. Companies also make technological forecasts to gain insight into technological developments in the market that may be applied by the own company. Dynamic forecasting models contain options to create insight into consequences of changing competitive behaviour or demand and financial margins. The possibility of these models to adapt input variables to changes is necessary, given the dynamics of the telecommunication market. Operators invest much time and capacity in forecasting via research into marketing, demand and

136

customer satisfaction and following competitors’ behaviour. Although forecasting is used to reduce uncertainty, a certain level of uncertainty is intrinsic to forecasting and forecasting errors are also a source of uncertainty (Piyatrapoomi et al., 2004). As Piyatrapoomi et al. state: “no matter how hard we try and how advanced our technologies, the future is unknowable”. Obtaining information Epistemic or knowledge-related uncertainty can be reduced by obtaining more information (Meijer et al., 2006). This can be done via different methods. First of all, it can be done by gaining information from suppliers on new upcoming equipment and technologies. This information, whether or not combined with knowledge from other sources, like a company’s own specialists or market studies, can be applied to perform detailed technological analyses to compare technological options on the basis of several relevant criteria (see also Chapter 2). Cost-benefit analysis (see 4.1.2) can be applied as a next step to gain more insight into the financial aspects and to reduce uncertainty related to unforeseen costs or to payback periods. Another commonly applied approach is benchmarking with

competitors to learn from their experiences, successes and mistakes.

4.6 Conclusion

When considering whether to pursue an investment project, it is typical to use a decision rule to determine whether or not the project should be undertaken. Strategic decision-makers have several evaluation methods at their disposal. The discounted cash flow methods have proven themselves to be adequate in the case of investments over a limited range and with a predictable future, but the procedures are not very suitable to major technological investments in situations that are characterised by considerable technological and market uncertainties. In an environment which is characterised by uncertainty these traditional procedures fail to take into account that effective risk management enhances the value of a system (Neufville, 2001). In traditional investment approaches, investment-related activities or projects are often seen as ‘now or never’, irreversible, not divisible, finite, static and stand-alone (De Ruijter & Janssen, 1996; Keswani & Schackleton, 2004). The NPV approach, for example, assumes that managers will remain passive when circumstances change, even if market conditions worsen dramatically (Keswani & Schackleton, 2004). In case of uncertain future cash flows, a higher discount rate is used, reducing the present value of these cash flows. This means that uncertainty affects the NPV in a negative way and hence affects the investment decision (Oslington, 2004). The procedures associated with discounted cash flows also fail to recognise the value of flexibility (Neufville, 2001). Keswani & Schackleton (2004) show that increasing future flexibility can dramatically increase the expected NPV of a project.

137

Thinking in terms of options alters the way actors deal with uncertainty, in the sense that conventional design is reactive to risk, whereas real options thinking recognises that uncertainty adds value to options, which makes uncertainty a positive element (de Neufville, 2003). Under conditions of uncertainty, real options give management the flexibility to acquire, divest and switch resources when such moves prove advantageous (Miller & Waller, 2003). The theory is explicitly based on the idea that most investment projects embed a series of alternative actions (Doraszelski, 1998). It provides an analytic framework to design flexible strategies, making it possible to respond in a flexible way to changing circumstances within the business environment or the organisation (De Ruijter & Janssen, 1996). In contrast to the conventional decision analysis, which works with a predetermined set of possible decisions, the options approach seeks to identify new possible paths, to change the decision tree by adding flexibility. A real options approach to a problem inserts additional decision nodes into the decision tree to reflect the options involved. The traditional valuation methods of Discounting Cash Flows (DCF), however, fail to account for these alternative choices (Suto et al., 2003). ROA is relevant to telecommunications in three areas: strategic evaluation, estimation and telephony cost modelling (Alleman, 2002). The broadband infrastructure markets are characterised by uncertainty, risk, major investments (that can sometimes be distributed) and quick developments in the environment (Eurescom, 1999). ROA can have a significant impact on decision-making processes in broadband upgrades and technology choice, due to its different way of assessing risk and uncertainty. For this reason, more insight into risk and uncertainty in the broadband market is very important. Telecom operators are faced with several types of risks and uncertainties during their decision-making process concerning the upgrading and roll-out of their broadband networks. These risks and uncertainties are in one respect intrinsic characteristics of large infrastructural projects and on the other hand are caused by (unknown) competitor behaviour, (unknown) end-user demand, rapid technological development and the different development paths available to operators. A combination of the uncertainty typologies from Meijer et al., (2006) Walker et al., (2003) and Courtney et al., (1997) provides a suitable framework to assess the risks and uncertainties involved in the decision-making process concerning the upgrading and roll-out of broadband networks. Telecom operators use several strategies to assess and reduce risks and uncertainties in their decision-making process. Examples are scenario investigation, sensitivity assessment, probability-based assessment and forecasting. In Chapters 6 and 7, the actual usage of traditional assessment methods as well as ROA by decision-makers in telecom companies will be investigated via qualitative and quantitative analysis. Also, the applicability of the theoretical framework for risk and uncertainty developed in this chapter will be validated.

138

4.7 References

Acimovic-Raspopovic, V., A. Kostic-Ljubisavljevic & V. Radojicic (2003). Dynamic routing design under forecast uncertainty. Proceedings from TELSIKS ’03: the 6th International Conference on Telecommunications in Modern Satellite, Cable and Broadcasting Service, 1, 30-33.

Alleman, J. (2002). A new view of telecommunications economics. Telecommunications Policy, 26 (1) 87-92(6).

Berdica, Katja (2002). An introduction to road vulnerability: what has been done, is done and should be done. Transport Policy, 9 (2),117-127.

Chopra, S., Sodhi, M.S. (2004). Managing Risk to Avoid Supply-Chain Breakdown. MIT Sloan Management Review, 46 (1), 53–62.

Courtney, H., J. Kirkland and P. Viguerie (1997). Strategy Under Uncertainty. Harvard Business Review 75(6),67-79.

Dalgleish, Fraser and Barry J. Cooper (2005). Risk management: developing a framework for a water authority. Management of Environmental Quality: An

International Journal,16 (3), 235-249.

Demkes, R. (1999). COMET: A comprehensive methodology for supporting telematics investment decisions (Doctoral dissertation Telematica Institutuut, 1999).

Doraszelski, U. (1998). The net present value method versus the option value of waiting: A note on Farzin, Huisman and Kort (1998). Journal of Economic

Dynamics and control 25 (8), 1109-1115.

Eurescom (1999), Extended investment analysis of telecommunication operator strategies- Investment analysis framework definition and requirements specification. (Project P901-PF Deliverable 1).

Fijnvandraat, Marieke and Harry Bouwman (2006). Flexibility and broadband evolution. Telecom Policy 30 (8-9), 424-444.

Hallikas, J., K. Puumalainen, T. Vesterinen, and V.-M. Virolainen (2005). Risk-based classification of supplier relationships. Journal of Purchasing & Supply

Management 11(2/3), 72–82. Harland, C., H. Brencheley and H. Walker (2003). Risk in supply network. Journal

of Purchasing and Supply Management 9 (2), 51–62.

Hermantzis, Fotios C. and Venkata Praveen Tanguturi (2004). Delay in the Expansion from 2.5G to 3G Wireless Networks: A Real Options Approach.

139

Proceedings from ITS 2004: 15th biennial international conference of International Telecom Society, Berlin, Germany.

Johnson, G. & K. Scholes (2002). Exploring Corporate Strategy. Text and Cases

(6th ed.). Harlow: Prentice Hall/Financial Times.

Keswani, A. and Schackleton, M.B. (2004). How real option disinvestment flexibility augments project NPV. European journal of operational research 168 (1), 240-252.

March, James G. and Zur Shapira (1987). Managerial perspectives on risk and risk taking. Management Science 33 (11), 1404-1418.

Meijer, Ineke S.M., Marko P. Hekkert, Jan Faber and Ruud E.H.M. Smits (2006). Perceived uncertainties regarding socio-technological transformations: towards a framework. International Journal of Foresight and Innovation Policy, 2 (2), 214-240.

Miller, Kent D. and H. Gregory Waller (2003). Scenarios, real options and integrated risk management. Long Range Planning 36 (1), pp.93-107.

Milliken, F. (1987). Three types of perceived uncertainty about the environment: state, effect, and response uncertainty. The Academy of Management

Review, 12 (1), 133–143.

Moro, C.A.C and. Carvalho, S.V. Strategic Investments Evaluation in a Flexibility Context with a Real Options Application. Retrieved January 30, 2008 from: http://hermes2.dpi.inpe.br:1905/col/dpi.inpe.br/hermes2@1905/2006/10.20.15.56/doc/strategic_worcapVI.pdf.

Myers, S.C. (1977). Determinants of corporate borrowing. Journal of Financial

Economics 5 (2), 147–176

Neufville, R. (2001). Real options: dealing with uncertainty in systems planning and design, paper prepared for presentation June 29, 2001 at 5th international conference on Technology, Policy and Innovation, Technical University Delft, Delft, The Netherlands. Retrieved: January, 29 2008 from: http://msl1.mit.edu/mib/dsp/curricula.MIT.edu/~dsplan/docs/papers/Delft.htm

Neufville, Richard de (2003). Real options: Dealing with uncertainty in systems planning and design. Integrated Assessment,4 (1), 26-34.

Oslington, Paul (2004). The Impact of Uncertainty and Irreversibility on Investments in Online Learning. Distance Education, 25 (2), 233-242. Retrieved May 13th, 2008 from: http://www.unsw.adfa.edu.au/sbus/OslingtonLinks/TimingInvestEdTech.pdf

140

Piyatrapoomi, N., A. Kumar & S. Setunge (2004). Framework for investment decision-making under risk and uncertainty for infrastructure asset management. Research in Transportation Economics, 8, 199-214, 2004.

Porter (1980). Competitive strategy: Techniques for analyzing industries and

competitors. New York: Free Press. Remer, S. & Nieto, A.P. (1995a). A compendium and comparison of 25 project

evaluation techniques. Part 1: Net present value and rate of return methods. International Journal of Production Economics, 42 (1), 79-96.

Remer, S. & Nieto, A.P.(1995b). A compendium and comparison of 25 project evaluation techniques. Part 2: Ratio, payback, and accounting methods. International Journal of Production Economics, 42 (2), 101-129.

Renkema, Theo J.W. (2000). The IT value quest, How to capture the business

value of IT-based infrastructure., Chichester: Wiley.

Ruijter, de Paul and Nico Janssen (1996). (Real) option thinking and scenarios. Retrieved May 12, 2005 from: www.deruijter.net/nl/artikelen/option_thinking/option_thinking.html.

Smit, Han T.J. and L. Trigeorgis, (2003). Real Options: Examples and Principles of Valuation and Strategy. In Joe McCahery and Luc Renneboog (eds.), Venture capital contracting and the valuation of high tech firms. Oxford: Oxford University Press. Retrieved May 2003 from: http://publishing.eur.nl/ir/repub/asset/12258/RealOptions_2003.pdf.

Suto, H., Alleman,J., and Rappaport, P., Simple decision-making criterion as real options. Proceedings from IEEE International Conference on Computational Intelligence for Financial Engineering, 2003, 17-24.

Vermaas, Karianne (2007), Fast diffusion and broadening use (Doctoral dissertation, University of Utrecht, 2007).

Walker, Warren (2000). Policy analysis: a systematic approach to supporting policymaking in the public sector. Journal of Multi-Criteria Decision Analysis,

9 (1-3), 11-27.

Walker, W.E. and V.A.W.J. Marchau (2003). Dealing With Uncertainty in Policy Analysis and Policymaking. Integrated Assessment,4 (1), 1-4.

Walker, W.E., P., Harremoes, J. Rotmans, J.P. van der Sluijs, M.B.A. van Asselt, P. Janssen and M.P. Krayer von Krauss (2003). Defining Uncertainty- A Conceptual Basis for Uncertainty Management in Model-Based Decision Support. Integrated Assessment, 4 (1), 5-17.

141

Weeds, H. (2002). Real Options and Game Theory: When Should Real Options Valuation Be applied? Paper presented at the 6th Annual International Conference on Real Options 2002.

Wernerfelt, B., and A. Karnani (1987). Competitive strategy under uncertainty. Strategic management Journal, 8 (2), 161-177.

142

143

Chapter 5 Strategic decision-making in a

complex, dynamic and uncertain

environment

To roll-out a broadband network, decisions have to be made on what the long-term objectives of the organisation are, whether an evolutionary or revolutionary technological path should be followed to support reaching these objectives in the course of time, when the roll-out of a technology should start and how much money should be invested. The enormous financial risks associated with high initial investments, together with the irreversible character of infrastructures, make decisions concerning infrastructure extremely critical. In Chapter 4, we provided more insight into methods to assess investments for broadband roll-out as well as to assess risk and uncertainty. Although it has been concluded in Chapter 3 that network upgrade decisions have a dominant financial character, decision-makers have to consider more aspects than only money. Moreover, these decisions aren’t taken in isolation, but in a highly dynamic environment with many actors with diverging interests. Decisions on broadband roll-out are strategic, as well as complex. Several decision-making theories are available to describe decision-making processes. This chapter will provide more insight into decisions and the decision-making process regarding broadband roll-out. Paragraphs 5.2 and 5.3 describe two decision-making theories (Garbage Can and Logical Incrementalism) in greater detail. Before discussing the decision-making process, we first focus on several characteristics of decisions. We begin by describing two characteristics of decisions regarding broadband roll-out: ‘strategic’ and ‘complex’.

Decisions on broadband roll-out are strategic

Decisions like the ones involved in the roll-out of new technologies, “determining the basic long-term goals and objectives of an enterprise and the adoption of courses of action and the allocation of resources necessary for carrying out these goals”, are defined as strategic decisions (Robertson, 2003, citing Chandler, 1962). Simon (1960) refers to strategic decisions as non-programmed decisions. According to Simon (1960), strategic decisions are novel, unstructured and consequential. Non-programmed decisions cannot be solved with standard solution methods because these decisions have not arisen before, their nature and structure are complex and elusive, or their importance demands a tailor-made solution (Simon, 1960). The consequences of strategic decisions are difficult to quantify and involve a great deal of uncertainty, emphasising the significance of managerial judgement (Lin,

144

2004). In general, strategic decisions answer three main questions: how the organisation should be equipped (by investment in productive technology or in data processing equipment), what the best structure for the organisation is and what the organisation should do (Hickson, 1987). Examples of strategic decisions are the introduction of a new technology, investments in buildings and equipment, internal reorganisation, electronic data processing and related planning. Common are decisions to introduce new services; less frequent decision topics are mergers and acquisitions and raising capital (Hickson, 1987). Strategic decisions are taken by senior management (Lin, 2004).

Decisions on broadband roll-out are complex

The issues that play a role in the roll-out of new broadband infrastructures are also complex. In literature, several typologies of complexity can be found. According to Liew (2005), complex problems consist of interrelated decisions that span multiple domains, paradigms and/or perspectives. De Baas (1998) provides a typology of complexity based on several characteristics. Next, we discuss the framework of De Baas (1998) and apply this framework to the broadband market based on theory, the domain description from Chapter 2 and five years of personal practical consultancy experience within the broadband market. According to De Baas (1998) complex situations can be characterised on the basis of six factors: 1. Unclear or controversial objectives, 2. Uncertainty about the required means and their effectiveness, 3. A changing pattern of participants in the decision-making process, 4. Various actors with diverging interests, 5. Mutual interdependency between actors in reaching their objectives, 6. Mutual independence (autonomy) of actors in determining their own behaviour. All six characteristics clearly apply to the broadband roll-out domain.

Unclear or controversial objectives - Several conflicting objectives are visible within companies. Long-term value drivers for business strategies (to enhance strategies for the continuity of the company) are quite different from short-term value drivers (reputation management) (Keijzers, 2003). A common characteristic of short-term interests is that they can be identified, qualified and quantified. Long-term interests, on the other hand, have in common that the objectives in terms of sustainability in general or in terms of the company’s success remain rather vague. Because company divisions increasingly operate as separate companies with their own financial balances (see, for example, KPN Annual report, 2007) different objectives, coupled to the various divisions of these companies, can be

145

distinguished. In telecom companies, conflicting interests like long-term network planning objectives versus short-term marketing objectives, objectives of business versus retail divisions and mobile versus fixed network divisions can be seen. Also, on the higher level of the telecom market as a whole several conflicting interests are visible. It is of interest to operators to tie customers to their company as long as possible because this generates certain revenue streams. Customers, however, want flexibility and they want to be able to change operators when they want to do so. The roll-out of most future-proof broadband networks is very interesting to end-users, but – at the moment – not a viable option for broadband operators from an economic perspective. Furthermore, as described in Chapter 2, the intervention of local governments in roll-out of broadband realised high-speed connections for end-users and is for that reason within end-user interest. This roll-out is, on the other hand, against the interests of current operators who consider these local governments being new competing market entrants. Moreover, the final objective of the entire long-term roll-out is not clear in the sense that the most future-proof network technology remains unknown. Although at this moment, optical fiber is still seen as most future-proof technology, operators never expected that, due to new compression technologies, the transmission capacity of the existing coax and copper networks could be stretched out so far. Also, new wireless broadband technologies seem to be promising (see also Chapter 2).

Uncertainty about the required means This type of uncertainty is closely connected to ‘unclear objectives’. At the moment, it is impossible to predict how much bandwidth the market will demand in five years time or which services will be developed that may demand higher bandwidth than can be provided with the existing networks. The adoption and use of new technologies are not constant, but rather evolve over time, and patterns of technology adoption and diffusion are never fully predictable (Vermaas, 2007, p.215). As a consequence, the necessary means to provide enough bandwidth are still not known and it is also not certain whether the chosen technology path will be effective in meeting future bandwidth demand. Many technological migration paths are possible to provide higher bandwidth in the future (see Chapter 2, section II). For that reason, it appears to make sense that, within telecom companies, different visions can be distinguished on which technology path should be followed to arrive at a next generation broadband network. Also, a very general objective, for instance increasing revenues, may be realised in many ways. To realise short-term revenues or for competitive reasons, a fast introduction of new services sometimes demand quick-and-dirty network adjustments. These adjustments could, however, conflict with long-term objectives and network strategies. Another example has to do with the convergence of mobile, data and telephony networks (see also Chapter 2). Because telephony can also be provided via data networks (VOIP) and mobile networks, several divisions within a single company (Internet division, fixed network division, mobile division) can start

146

cannibalising each other’s customers, which makes answering the question on how to reach long-term objectives even more complicated. A changing pattern of participants in the decision-making process The participants in the decision-making process are not fixed. At different times, different people are important in the decision-making process. The moment alternatives must be mapped, several expert groups and technology boards will be consulted. Also, discussions with equipment providers will take place to gauge their vision on future technological possibilities. In comparing possible alternatives financial experts and in some cases consultants will enter the process. Also, marketing forecasts are an important input for determining the expected future demand. Moreover, people enter and leave the company, altering the people involved in decision-making processes. Various actors with diverging interests In the telecom market, many different actors with diverging interests can be seen. Like all private companies, operators act in the interest of sustainability and growth of their company and want to make a profit. As far as end-users are concerned, on the other hand, ‘convenience at reasonable costs’ seems to be important in the decision whether or not to adopt broadband (Vermaas, 2007). Regulators are interested in stimulating sustainable competition in the electronic communications markets (OPTA, 2005). Also, within companies, several conflicting interests between divisions can be seen. Marketing divisions most of the time work with short timelines and ad hoc competitive pressure. Network divisions have, on the other hand, a much longer network planning horizon, which can cause conflicts between these divisions of telecom companies. A practical example of conflicting interests between the network and marketing division is the introduction of the ‘Internet + bellen’ service by KPN. This dual play offer (telephony in combination with Internet access) was, as a result of high competitive pressure, introduced in the market before the network was ready for a large-scale adoption of this service. This resulted in many problems related to roll-out and QoS and caused many complaints by customers. Mutual interdependency between actors in reaching their objectives Although different divisions within telecom companies have diverging interests, they are dependent on each other and cannot make decisions independently because they all have critical knowledge needed to arrive at well-considered decisions. These interdependencies also can be seen in a broader market perspective. Although carrier operators and incumbents are strong competitors in the market, a substantial part of the revenue stream from incumbents is created by wholesale earnings from these carrier operators (KPN annual report, 2007). On the other hand, these carriers are completely dependent from the incumbent’s network in reaching their subscribers, because they make use of the local loop from the

147

incumbent to connect end-users to their own backbone networks. Another example is the interdependency between end-users and broadband providers. End-users will not have Internet without operators who make sure they get access to these services, but on the other hand, end-users are the ones that generate the revenues for these operators. Mutual independency (autonomy) of actors in determining their own behaviour Although they depend on each other, several divisions within companies, equipment providers, carriers, end-users and so on, are fully independent in determining their own behaviour. Because of these interdependencies, there are several boundaries within which they have to operate. Due to regulation, incumbent operators cannot, for example, increase end-user or interconnection prices without the approval of the national regulator. To summarise, based on De Baas (1998), decisions regarding broadband roll-out are strategic decisions that are characterised by a high level of complexity, expressing itself in large numbers of interdependent actors with different, conflicting interests and in large numbers of technological, organisational, regulatory and market-related factors.

Other types of decisions in broadband roll-out

In addition to strategic or non-programmed decisions, more straightforward, repetitive and routine decisions, appearing in day-to-day decision-making, can be distinguished in telecom companies. Simon (1960) refers to these decisions as programmed decisions. Gilligan et al. (1983) divide programmed decisions into two categories: ‘short-term operational decisions’ and ‘periodic control decisions’ (Lin, 2004). Periodical control and short-term operating control decisions are tasks of middle and operational management respectively (Lin, 2004). Short-term operating control decisions concern mechanical decision models requiring simplified assumptions and generating information for regular feedback for control decisions. An example of short-term operating control decisions is stock control. Periodical control decisions occur regularly but infrequently. Examples are decisions regarding equipment replacement, investment-related decisions and performance monitoring (Lin, 2004). These decisions are mainly operational and occur when the strategic decision to roll-out a certain technology already has been made. The decisions associated with the actual roll-out and implementation of a network occur more frequently and involve less uncertainty.

148

5.1 The way towards a decision: the decision-making

process

Complex, strategic decisions are not the result of a few hours of discussion, but the outcome of a decision-making process. The process from developing a decision strategy via implementation to appraisal of results occurs in all organisations. Strategic decision-making processes are characterised by long time periods, novelty, complexity, open-endedness and ambiguity, revealing itself in a low level of understanding of the situation, the routine to follow or the solution (Mintzberg, 1976). Most of the major strategic choices are made in two years or less with a mean period of about a year (Hickson, 1987). The literature on (strategic) decision-making is multidisciplinary, voluminous and complex. It is at different times intuitive, mathematical, philosophical and empirical and rooted in economic, political and organisational theory as well as social psychology. In a word, it is daunting (Tarter, 1998).

Classifying decision-making theory

Many classifications have been made in literature of the various modes of strategic and complex decision-making. A distinction needs to be made between several types of decision-making theories. A first, fundamental distinction can be made between normative and descriptive decision-making theories. Descriptive decision-making theories can be subdivided into purely descriptive decision-making theories and theories that also give decision-makers some leads to handle the environment in which they have to make their decisions. Below, we discuss a number of theories that seem to be applicable to the decision-making processes concerning broadband roll-out and discuss these theories in greater detail. Normative decision-making theory Normative decision-making theory consists of a group of models that set normative rules on how decision-making should develop to arrive at ‘good’ decisions. The normative approach is an optimisation strategy that is based on traditional economic theory, assuming clear goals, complete information and the cognitive capacity to analyse the problem (Tarter, 1998). Decision-makers act as completely rational creatures in arriving at the single best solution to a problem. Literature describes these decision-making processes as transitions from one state to another, through a number of intermediate, sequential stages. Six common stages can be defined: identifying the problem, diagnosing the problem (collecting information), generating alternatives, examining consequences, making the decision and implementing the decision (Tarter, 1998). This rational approach to decision-making is, however, quite oversimplified and comes together with several practical problems. Firstly, the notion of clear and simple goals for organisations is not realistic. Normally speaking, decision-makers

149

are not faced with concrete and clearly defined problems. Most of the times, organisational goals and problems are most complex and often conflicting. Secondly, organisations are always faced with considerable uncertainty, due to incomplete knowledge on consequences of certain decisions and the fact that it is practically impossible to consider all possible alternatives (Gaffron, 2002). This makes maximisation an impossible task. Furthermore, the optimisation strategy places unrealistic demands on the human cognitive ability and the ability to gather and handle the quantity of information needed to select the best option. Moreover, there is the question whose values and goals should be used in the decision-making process, because most of the time actors have conflicting interests. A final problem with rational decision-making is the question how to separate facts and values and means and ends in a decision-making process (Gaffron, 2002). To summarise: the traditional model is an ideal rather than a description of how decision-makers function in reality (Tarter, 1998). Descriptive decision-making theory A second group of decision-making theories is a response to the rational view and addresses the problems associated with the normative models. This is the group of descriptive decision-making theories, which have been developed to give users a methodology for describing and characterising the decision-making process (Scherpereel, 2001). In descriptive models, two subgroups can be recognised. Firstly, a group of models that are purely descriptive and only function to create understanding of and insight into how -sometimes seemingly unexplainable - decisions occur. Examples of models that give insight into how decisions originate in dynamic, complex and uncertain environments are the Garbage Can model by Cohen, March & Olsen (1972), the streams model of Kingdon (1984) and the rounds model of Teisman (2000). The rounds model is, besides a descriptive model, also an analytical framework for researchers studying decision-making processes on a certain topic. In this model, ‘decision-making rounds’ demarcate the starting point and ending point of decision-making processes on a certain studied topic in retrospect, where the researcher determines the most crucial decisions or decision-making aspects. The second group of descriptive models is not purely descriptive, but also gives decision-makers some leads to cope with characteristics of the environment in which they operate and to follow a decision-making process that is more adapted to these characteristics. Two examples of these models are the model of Logical Incrementalism (Quinn, 1980) and the satisficing theory by Simon (1959). Applying decision-making theory to broadband roll-out Based on the characteristics of the broadband market and of the companies in which strategic decisions regarding broadband roll-out take place, there are three theories that qualify themselves as theories for describing decision-making with regard to broadband roll-out. The first is the Garbage Can model by Cohen, March & Olsen, dating from 1972. This

150

model will be used to gain some more understanding on how decisions are made within this complex environment. The second theory we discuss is Logical Incrementalism (Quinn, 1980). These two theories will now be discussed in greater detail. Combining the theoretical descriptions and the characteristics of the broadband market will give us greater insight into the applicability of these theories with regard to the decision-making process we investigate in this research In Chapter 6, the level of practical applicability of these models will be validated based on extensive interviews with decision-makers on strategic level within Dutch telecommunication companies.

5.2 Describing the broadband market and explaining the

realisation of decisions: the Garbage Can

The Garbage Can model of organisational choice (GCM) has been highly influential in organisational theory (Lipson, 2007). The garbage can is a descriptive metaphor for how organizational decisions are made. The model was developed by Cohen, March and Olsen in 1972 and serves to explain decision-making processes in so-called ‘organised anarchies’. These are organisations that exhibit three general characteristics: problematic goals, unclear technology and fluid participation. We will now explain these three characteristics.

5.2.1 Telecom companies as organised anarchies

In case telecom companies can be characterised as organised anarchies, it can be assumed that the Garbage Can theory can be applied to strategic decision-making within these companies and that we can use the theory as a starting point to investigate decision-making with regard to broadband-roll-out.

Problematic goals refer to ambiguity regarding problems and goals and to what the organisation is really about. Organisational actors may be uncertain about the nature of the problems they face and about what they hope to accomplish. They may discover their preferences through acting (Lipson, 2007). The operation functions based on a variety of inconsistent and ill-defined preferences and can better be described as a loose collection of ideas than as a coherent structure (Cohen et al., 1972).

Unclear technology refers to a situation in which members within the organisation do not fully understand how their organisation works and are uncertain of the rules, structures, and processes by which decisions are made. Technology refers in this case to organisational processes and methods to deliver the product, not to technological artefacts (Lipson, 2007). In Chapter 2, we saw that there are many technologies to realise a symmetrical bandwidth of 10 Mbps. The visions on which

151

technology path should be followed to realise the most future-proof technology as well as the visions on which technology is most future-proof vary between operators, as becomes clear in by the different paths they follow. As discussed earlier, strategic decisions are taken by the highest management levels of organisations, which could result in unclear processes for the remaining parts of the organisation. Besides, as discussed earlier, these decisions are very complex and, as we saw in Chapter 2, many factors in the areas of technology, finance, market and regulation must be taken into account. Also, this complex deliberation process can lead to unclear technology.

Fluid participation refers to the involvement of different actors in different situations. As we explained earlier, different actors are involved in different parts of decision-making processes. These actors vary from internal company divisions to external experts, equipment providers and, in some cases of significant network changes of incumbent operators, regulators and carriers connected to the incumbent network. All these ‘interest groups’ are in a position to influence the decision-making process. The theory of the Garbage Can and the criteria of an organised anarchy were originally applied to universities (Cohen et al., 1972; Hayes et al., 1998). The Garbage Can theory has, however, been applied to other types of companies as well. Lipson (2007) demonstrated that the criteria for organised anarchies and, by extension, the Garbage Can theory, can even be applied to globally operating organisations like the United Nations. Based on this broad applicability and on several market insights as described in this and earlier chapters, we assume that telecommunication companies can also be characterised as organised anarchies.

5.2.2 Decision-making within the garbage ban

The garbage can model is a descriptive metaphor for how organizational decisions are made. According to Ansell (2001) in the International Encyclopedia of the Social & Behavioral Sciences (2001) “the model conceives of organizations as conglomerates of semiautonomous decision arenas or `garbage cans' through which problems and solutions flow as independent streams.” At specific moments of decision, decisions are outcomes of a precise mix of problems and solutions represented in a garbage can. The precise composition of this mix depends on several factors: the number of decision arenas, the structure of access to them, the overall organizational load of problems and solutions, and the allocation of energy and attention across arenas (Ansell, 2001). An important feature of the Garbage Can model is that problems and choices are partially decoupled. It frequently occurs in organisations that solutions are proposed before problems have been identified, especially when organisations are faced with a great deal of ambiguity,

152

conflicts, uncertainty and a changing environment. Specific decisions do not follow orderly processes from problem to solution, but are outcomes of four relatively independent streams of events within the organisation: problems, solutions, participants and choice opportunities. • Problems Problems are points of dissatisfaction, which may or may not lead to a solution. Examples in the case of telecom companies comprise high churn, competitive threats, disappointing revenues, decreasing end-user demand, networks falling behind on competitors, regulatory threats, etc. • Solutions Solutions can exist independent of problems and the attractiveness of a solution can also stimulate a search for a problem. According to Cohen et al. (1972), organisations sometimes have the answer before they find the question and justify their answer with the right question (Tarter, 1997). Examples of solutions are different technological alternatives to upgrade networks or develop new services. These solutions may become more or less interesting according to external developments like the behaviour of end-users and competitors, regulators, the availability and price of equipment, the possibility to buy frequency licenses, etc. In case of an upcoming frequency auction for a technology, this technology suddenly attracts attention, as well as possible services that could be interesting to attract new customers or to beat competitors. • Participants are organisational members. Problems and solutions change when participants join and leave the process. Substantial variation in participation arises from the fact that other issues demand time from the participants (Cohen et al., 1972). • Choice opportunities are moments when organisations make decisions. Choices are the result of unique combinations of problems and solutions at a given time and place. Based on the broader applicability of the Garbage Can model, as proven by Lipson (2007), we assume that the Garbage Can theory can also be applied to telecommunication companies. We illustrated that, based on general characteristics and insight into the market, the streams defined by Cohen et al. (1972) seem to be applicable to decision-making processes concerning broadband roll-out. Based on insights provided by Chapters 2, 3 and 4, it can be assumed that the streams of problems and solutions lie in the areas of intra-organisational, technological, financial and demand factors.

5.3 Logical Incrementalism

In Chapter 4 we saw that Real Options Analysis can very well be applied to broadband-related decision-making processes, due to its ability to deal with uncertainty by keeping open options for the future to adjust decisions based on changes in the environment (see Chapter 4). The theory of Logical Incrementalism

153

also specifically takes uncertain environments as a starting point. Logical Incrementalism is based on the assumption that it is impossible to predict the future with any accuracy because so many, most very strong, forces are involved (Kippenberg, 1998). The most appropriate way to grow, therefore, is to move forward in incremental steps. Seen from the perspective of logical incrementalists, strategic decision-making is an incremental process. Top executives work consciously towards their objectives, but they are also prepared to be flexible on the way to these objectives. They experiment and hold off decisions to gather more information, going further if conditions are suitable and, if they are not, retreating and seeking another path. This incremental process is a steady process with an underlying logic, tying analysis, organisational politics and individual needs together into a cohesive new direction (Kippenberg, 1998). Quinn stresses that “Successful executives continually reassess the total organisation, its capacities and its needs related to the surrounding environment” (Quinn, 1980). Das & Teng (1999) mention three characteristics of the logical incrementalist decision-making process. Firstly, the process is incremental in nature and managers do not make dramatic decisions. Secondly, the decision-making process is a consistent movement aimed at reaching a clear objective. The strategic objectives are kept broad and relatively vague to keep the option for developments and adjustments open in case more information becomes available. This option to gather more information and feedback from the initial action is the final characteristic of the Logical Incrementalism (Das & Teng, 1999). The power of Logical Incrementalism lies in the fact that managers have access to broad options and subsequently narrow the range of relevant options down over time. Moreover, they constantly develop and evaluate options, based on feedback from earlier actions (Das & Teng, 1999). The incremental approach provides organisations the flexibility to consider emerging alternatives and keep the company open for future options instead of making drastic, irreversible decisions. Alternatives will not be missed, since managers keep getting information and keep conducting global analyses (Das & Teng, 1999). Quinn argues that Logical Incrementalism is an “effective, active management technique for top executives to practice in a purposeful, proactive and conscious way” (Kippenberg, 1998). Das & Teng, (1999) provide insights in several underlying explanations other researchers have given for Logical Incrementalism. A first explanation is that, due to an unstable environment and their limited cognitive capabilities, the best way for managers to achieve their strategic objectives is by choosing the smallest possible increments. Other researchers see the possibility to remain flexible enough to be able to assimilate new information as the argument in favour of incremental

154

decision-making. Logical Incrementalism also allows organisations to make adaptations based on feedback from initial actions (Das & Teng, 1999). As discussed in Chapter 4, the broadband market is characterised by many uncertainties in the areas of technology, economic circumstances and the behaviour of end-users, suppliers and regulators. In Chapter 4, it was proven that the Real Option Analysis, a financial assessment method which takes a high level of environmental uncertainty into account, would be well applicable to decision-making regarding investments in broadband network roll-out. This method enables a flexible strategy in which environmental changes can be incorporated and can lead to adapting earlier chosen paths. Flexibility and incremental decision-making as an answer to uncertainty reflect a clear similarity between Real Option Analysis and the decision-making theory of Logical Incrementalism. Based on this similarity and the fact that it was shown in literature that Real Option Analysis can be applied very well to uncertain markets as the telecom market (a.o. Neufville, 2001 & 2003; Oslington, 2004; Miller & Waller, 2003) (see also Chapter 4) we assume that the theory of Logical Incrementalism will as well be applicable to the decision-making processes concerning broadband network roll-out as well.

5.4 Conclusions

Based on theories of Simon (1960) and De Baas (1998) we can conclude that decisions regarding broadband roll-out are strategic (non-programmed) as well as complex, revealing itself in many interdependent actors with different, conflicting interests and many technological, organisational, regulatory and market-related factors. Besides non-programmed decisions, also so-called programmed decisions can be distinguished (Simon, 1960; Lin, 2004). With our qualitative research in Chapter 6, we examine whether these different types of decisions can be distinguished within practical broadband decision-making. Besides decisions, literature also offers a number of different frameworks to describe decision-making processes, not all of which can also be applied to complex strategic decision-making with regard to broadband roll-out. After comparing the characteristics of the broadband market with those of several decision-making theories, we selected two decision-making theories which seem suitable to our research model: the Garbage Can model by Cohen, March and Olsen (1972) and the theory of Logical Incrementalism (Quinn, 1980). The Garbage Can theory by Cohen et al. (1972) can be applied to organised anarchies: organisations which are characterised by unclear preferences, unclear technology and fluid participation. Although the Garbage Can theory is developed

155

for describing decision-making processes within universities, application of the theory to much larger and worldwide operating organisations like the United Nations offers the possibility to examine whether this theory is also applicable to strategic decision-making processes in telecom companies. On first sight telecom companies fit the defining criteria of an organised anarchy, to which the Garbage Can model can be expected to apply. For this reason, we proposed the hypotheses that the criteria for organised anarchies and the Garbage Can model offer a suitable theoretical framework to further explore strategic decision-making with regard to broadband roll-out. This will be further validated by qualitative and quantitative research in Chapters 6 and 7. The broadband market is characterised by rapid changes in terms of the available technologies, end-user preferences, competitive behaviour and regulatory frameworks. Irreversibility is due to the fact that most of the costs involved in investment projects in broadband infrastructure are sunk costs. These market characteristics require flexible strategies that offer possibilities for feedback loops, to incorporate environmental changes and to adapt earlier chosen technology paths. Uncertainty and irreversibility of high investments make it in many situations worthwhile to delay roll-out decisions to a moment there is more information available, for example on expected adoption or technological standards. Based on more information, the initial decision can be adapted, postponed or implemented, creating flexibility in decision-making under conditions of uncertainty (De Ruijter & Janssen, 1996; Keswani & Schackleton, 2004). The strategy of Logical Incrementalism offers possibilities to give into this need for flexibility within a dynamic market as the broadband market is. In its way of handling uncertainty by incorporating flexibility the theory shows similarities with the Real Options Analysis (see Chapter 4), of which it has been proven that it is relevant to the telecommunications market (Alleman, 2002). Based on the similarities between these two theories we assume that the theory of Logical Incrementalism can be applied to the decision-making processes concerning broadband network roll-out as well. This assumption will be validated in Chapter 6 via qualitative research.

156

5.5 References

Alleman, J. (2002). A new view of telecommunications economics. Telecommunications Policy, 26 (1) 87-92(6).

Ansell, C.K. (2001). Garbage Can Model of Behavior. International Encyclopedia of

the Social & Behavioral Sciences, 5583-5886.

Baas, H. de (1998). Hectiek en vereenvoudiging van besturingscontext- Afvalbeleid in Nederland naar een nieuwe ronde. Bestuurskunde, 5, 210-218.

Cohen, March & Olsen (1972). A Garbage Can Model of Organisational Choice. Administrative Science Quarterly, 17 (1),1-25.

Das, T.K. and Bing-Sheng Teng (1999). Cognitive biases and strategic decision processes: an integrative perspective. Journal of Management Studies, 36 (6), 757-778.

Gaffron, Philine (2002). The implementation of walking and cycling policies in British local authorities-Idea(l) into action? Appendix A- Decision-making analysis. (doctoral dissertation NAPIER university of Edinburgh).

Hayes, Sherman L., and Patricia B. McGee (1998). Garbage can decision-making in a ‘structured anarchy’ for your CWIS. Could you translate that for me please? Campus-Wide information systems, 15 (1), 29-33.

Hickson, David J. (1987), Decision-making at the top of organisations, Annual

Review of Sociology, 13,165-192.

Keijzers, Gerard (2003). Creating sustainable directions: evolving stakeholder approach in seven multinationals.(Diageo PLC, Heineken N.V., ING Bank N.V., Koninklijke Ahold N.V., Numico, Shell Chemicals U.K. Ltd., Unilever Group). The Journal of Corporate Citizenship (Summer 2003), 79-89.

Keswani, A. and Schackleton, M.B. (2004). How real option disinvestment flexibility augments project NPV. European journal of operational research 168 (1), 240-252.

Kippenberger,T. (1998). Logical incrementalism. The Antidote, 3 (6), 26-27.

KPN (2007). Annual report 2007. retrieved May 13, 2008 from: http://www.kpn.com/reporting07/report07/PDF/KPN_Annual_Report07.pdf.

Liew, Angela and David Sundaram (2005), Complex Decision-making Processes: their Modelling and Support, Proceedings of the 38th Hawaii International Conference on System Sciences, January 03-06, 2005.

Lin, Hui-Chao (2004). Decision Theory and Analysis. Futurics, 28 (1&2), 27-45.

157

Lipson, Michael (2007). A Garbage Can Model of UN Peacekeeping. Global

Governance, 13 (1), 79-97.

Miller, Kent D. and H. Gregory Waller (2003). Scenarios, real options and integrated risk management. Long Range Planning, 36 (1), 93-107.

Mintzberg, Harry (1976). The structure of “Unstructured” Decision Processes. Administrative Science Quarterly, 21(2), 246-275.

Neufville, R. (2001). Real options: dealing with uncertainty in systems planning and design, paper prepared for presentation June 29, 2001 at 5th international conference on Technology, Policy and Innovation, Technical University Delft, Delft, The Netherlands. Retrieved: January, 29 2008 from: http://msl1.mit.edu/mib/dsp/curricula.MIT.edu/~dsplan/docs/papers/Delft.htm

Neufville, Richard de (2003). Real options: Dealing with uncertainty in systems planning and design. Integrated Assessment,4 (1), 26-34.

OPTA (2005). Annual Report 2004 and Vision of the market 2005. Retrieved May 13, 2008 from: http://www.opta.nl/download/Jaarverslag%5F2004%5FENG%2Epdf.

Oslington, Paul (2004). The Impact of Uncertainty and Irreversibility on Investments in Online Learning. Distance Education, 25 (2), 233-242. Retrieved May 13th, 2008 from: http://www.unsw.adfa.edu.au/sbus/OslingtonLinks/TimingInvestEdTech.pdf

Quinn, James Brian (1980) Strategies for change-logical incrementalism, 1st Edition. Homewood: R D Irwin.

Robertson, Duncan A. (2003). Agent-Based Models of a Banking Network as an Example of a Turbulent environment: The Deliberate vs. Emergent Strategy Debate Revisited. Emergence: A Journal of Complexity in Organisations and

Management, 5 (2), 56-71

Ruijter, de Paul and Nico Janssen (1996). (Real) option thinking and scenarios. Retrieved May 12, 2005 from: www.deruijter.net/nl/artikelen/option_thinking/option_thinking.html.

Scherpereel, Christopher Michael (2001). Decision Order Theory: The semantic Dimensions of Decision-making. (Doctoral dissertation, Northwestern University, 2001).

Simon, H. A. (1959). Theories of decision-making in economics and behavioural science. American Economic Review, 49 (3), 253–283.

Simon, H.A. (1960). The new science of management decision (3rd print). New York: Harper and Row.

158

Tarter, C. John (1998). Toward a contingency theory of decision-making. Journal of

Educational Administration, 36 (3), 212-228.

Teisman, Geert (2000). Models for research into decision-making processes: on phases, streams and decision-making rounds. Public Administration, 78 (4), 937-956.

Vermaas, Karianne (2007), Fast diffusion and broadening use (Doctoral dissertation, University of Utrecht, 2007).

159

Chapter 6 Exploring the rationale

behind broadband roll-out- Qualitative

analysis

In the previous chapters, a theoretical background has been given on several subjects. Chapters 2 and 3 focused on the content-related side of broadband roll-out. Chapter 2 provided more insight into the three dimensions, relevant to decision-makers on broadband roll-out, being market, economy and policy/regulation. In Chapter 3, a Meta-analysis has resulted in a theoretical conceptual model, based on extensive literature analysis. Chapters 4 and 5 focused on the process-related side of broadband roll-out. In Chapter 4, theory on financial assessment methods and on risk and uncertainty has been described in more detail. In Chapter 5 we discussed two decision-making theories which are assumed to be well applicable to decision-making with regard to broadband roll-out. In Chapters 6 and 7, we focus on the validation of our theoretical framework and the hypotheses developed in our theoretical chapters. The validation of our theory consists of a qualitative and a quantitative part. In Chapter 7, we deal with the quantitative part of our validation. In this chapter, we describe the qualitative validation of our theoretical part based on extensive expert interviews with decision-makers from the broadband domain. We focus on the qualitative part of our validation and discuss both the content-related and the process-related side. In paragraph 6.2 we focus on the content-related results of our qualitative research. Parts of the interviews are used to illustrate these results. In this part, we focus on the three domains from the broadband market as introduced in Chapter 2. In paragraph 6.3, the first part of the theoretical framework on risk and uncertainty will be validated. After describing the content-related side, we discuss the process-related side of the qualitative validation. In paragraph 6.4 we will provide more insight into the course of the decision-making process, several types of decisions and the actors involved. We also provide insight into the preference of decision-makers for applying financial assessment methods in technological investment-related decisions. In paragraph 6.5, we discuss the validation of the applicability of the decision-making theories of Logical Incrementalism and Garbage Can. The final part of this chapter focuses on model development. After sharing our descriptive results, paragraph 6.6 contains the validation of the concepts and relationships of our conceptual model as presented in Chapter 3. In this paragraph, we discuss whether a parsimonious model on broadband roll-out can be developed on the basis of the results of our qualitative research.

160

6.1 Method

For our qualitative research we carried out in-depth expert interviews and qualitative data analysis. Interviews All interviewees were asked the same questions via a questionnaire. The questionnaire consisted of two main parts, 1) the course of the decision-making process on upgrading and developing networks and 2) validation of the conceptual model developed in Chapter 3. In part 1, around 20 questions were asked on the following subjects: technological components; financial components; policy components (policy and regulation); cooperation in the value chain and service development. The questionnaire was not sent to interviewees beforehand. On average, the interviews lasted 2.5 hours and they were all taped, typed out verbatim and sent back to interviewees for validation and to give them the possibility to mark confidential parts. We made it clearly that we would safeguard the anonymity of the interviewees. Sample In-depth interviews were conducted with 12 decision-makers within telecom companies in the Netherlands. The interviewees work at the two biggest cable companies of the Netherlands, UPC and Essent Kabelcom (@Home41), the Dutch incumbent telecom company KPN and Tele2-Versatel, the biggest Dutch alternative provider that uses the local loop networks of other operators (mainly KPN) via local loop unbundling (LLU). The interviewees are part of the middle or higher management of these companies and they are all directly involved in the decision-making process regarding the roll-out or upgrade of the telecom network of the concerning companies. With these companies, we covered the largest part of the Dutch telecom market, including operators with a copper TP, cable and ‘virtual’ local loop. To ensure internal consistency within companies and to cover our entire area of interest (economic, organisational, regulatory and technological issues) three decision-makers have been interviewed per company. All interviews were taped, transcribed and the reports have been reviewed and approved by the interviewees. As a next step all interviews are analysed with Atlas.ti v5.0, a scientific software programme for visual qualitative data analysis. This software makes it possible to create causal or concept maps. We will now explain concept mapping in greater detail. Concept mapping In this research concept maps are used as a means of representing the conceptual structures underlying the process from decision-making about network upgrade or roll-out until the moment of actual revenues.

41 At the time of these interviews, @Home was still an independent company. Since February 1, 2007, @Home is part merged with Casema and Multikanbel into a new company, owned by private equity companies Marcus Pincus and Cinven.

161

Concept mapping is a technique for visualising the relationships between different concepts. Concept maps are a graphical two-dimensional display of knowledge that is comprised of concepts (usually represented in boxes or circles), connected by directed arcs enclosing brief relationships (linking phrases, mostly verbs) between pairs of concepts (Canas et al., 2005). Together, nodes and labelled or named links define propositions, assertions about a topic, domain, or thing (Alpert & Gruenenberg, 2000). Linking phrases are not limited to a defined set and can express any type of relationship. Figure 6.1a shows the theoretical representation of a concept map. Figure 6.1b gives an example: “Digging has an impact on the roll-out costs”.

Figure 6.1a: Theoretical representation of a concept map and Figure 6.1b: example of

concept map

In our research the used relationships are "A is cause of B", "A has strong impact on B", “A has an impact on B”, "A is part of B", "A is a B", “A and B are mutually influencing, “A contradicts with B” and “A is associated with B”. This method shows similarities with network analysis as we applied to our Meta-analysis with the software tool Netminer. Concept mapping focuses, however, strongly on several, more specific, types of relationships (not just causal in general) and on direction of relationships. In this sense, concept mapping has some strong parallels with semantic web modelling. Not only merely causal relationships are allowed, but also relationships like ‘is part of’ and ‘is equal to’ can be part of concept maps. Concept maps are used to structure argument forms and express relationships between ideas (Gaines and Shaw, 1993). They are useful to understand complex issues that can have multiple consequences or multiple explanations (Limkeatcherdchoo, 2006). The ability to measure a particular person’s knowledge about a given topic in a specific context is the major strength of concept maps (Canas et al., 2005). Due to variations in mapping conceptual structures, a variety of other terminologies is used for concept maps, i.e. ‘cognitive maps’, ‘knowledge maps’, ‘network maps’, or ‘mind maps’ (Tergan et al., 2005). For consistency, we will use the term concept mapping. Novak and Gowin (1984) mention several characteristics of concept maps. Firstly, concept maps must have a structure, which is, by definition, hierarchical. The second criterion is propositional coherency, meaning that concept maps are based on propositions of which every two concepts (A and B) and their linking phrase

ConceptA

ConceptB

Linking

relationshipa Digging Roll outcosts

Has impact onb

162

must form a ‘unit of meaning’ (be propositional coherent). Thirdly, concept maps have a specific context, being a particular domain of knowledge within which all concepts and linking phrases must be interpreted. These strict criteria demanded by the concept mapping format developed by Novak and Gowin (1984), also referred to as the ‘Novakian standard concept mapping format’, are, however, questioned by other researchers. Åhlberg (2004) mentions that although Novak & Gowin (1984) and Novak (1998) argue that concept maps should always be hierarchical, this is not always logical or economic. The fact that the world is a system implies that the best presentation of the world and its subsystems are conceptual systems, which are not always hierarchical (Åhlberg, 2004), but are more characterised by network structures. Putting the concept maps in this research to these criteria, they do not meet the first criterion, due to specific characteristics of the relationships. In our research we see, besides causal, many other types of relationships between concepts. Moreover, the concepts in our maps have many interrelationships, as has become apparent from our Meta-analysis and are for that reason the concept maps are quite complex. Due to these different types of relationships and the many interrelationships between the concepts it is not possible to create clear tree structures. Concepts are related to many more concepts in many ways and relationships are, for that reason, not in definition hierarchical and causal. The high complexity and many interrelationships in this research can be explained by the fact that these concept maps do not show the concepts and relationships mentioned by one person, but by 13 persons. We consciously decided to bundle all concepts and relationships of all decision-makers we interviewed into a single dataset for concept mapping. The goal of concept mapping is to generate general, sector-wide insight into concepts and the relationships between them to create a general conceptual model. By eliminating concepts that are mentioned only a few times, individual opinions are filtered out, which leads to more generic but broadly covered concepts. A possible disadvantage of this approach is that unique relationships are filtered out, but we consciously opted in favour of communality instead of uniqueness. Before deleting, we checked for specific, special, unique relationships, but they were not found. This communality makes it, however, very hard to create structured and hierarchical maps. The many interrelationships between the concepts shows how complex the decision-making process is and how many variables are involved in this process. This complexity was seen earlier in our Meta-analysis and clearly visualised with the ‘blurb’ in Figure 3.3. The other two characteristics of concept maps, propositional coherency and being part of a particular knowledge domain, can be applied very well to the concept

163

maps involved in this research. The specific knowledge domain is ‘the upgrading and roll-out of fixed broadband networks’. Advantages of concept mapping software The scientific software programme Atlas.ti v5.0 enables segmenting a text into passages (words, (parts of) sentences, paragraphs) to be indexed with codes. The software makes it possible to visually "connect" selected passages, memos and codes into diagrams which graphically outline complex relationships. Using a software programme like Atlas.ti. offers the advantage that all interviews and the concepts (named ‘codes’ within Atlas ti.) are linked to each other. The used concepts are listed and for every word or sentence that must be coded as a new concept, this list can be used to pick an already used concept for this word or sentence. As a result, links with these concepts out of several interviews are linked automatically, because they have the same concept name. For example, when concept A is linked in interview 1 with concepts B and C and in interview 5 with concept D, the software links these concepts and shows a concept map for concept A with relationships to B, C and D. This way, the answers from 13 decision-makers are combined in concept maps, which results in concept maps with a more generic character, representing the opinion of the biggest part of the Dutch telecom market instead of the opinion of a single decision-maker from a specific company. Disadvantages of concept mapping software A major drawback of the tool is that, although it delivers clear insight into the frequency of the concepts themselves and in the number of relationships they have with other concepts, the number of times they are mentioned remains unknown. Once a relationship between two concepts has been established, the programme offers no possibility to establish this relationship a second time. The programme does provide insightful indications of the strength of relationships between concepts, indicated by accompanying sentences of the connecting arrows (see Figure 6.1a and b), but further validation of these relationships is necessary. We applied this programme in the first place to validate the concepts of the conceptual model we developed in Chapter 3 and in the second place to gain some initial insights into the strength of the relationships between the concepts. Concept aggregation Concept mapping resulted in a total of 312 codes (concepts). To provide further insight into this ‘black hole’ of variables, we conducted twelve rounds of concept aggregation, in which concepts covering the same or very close related variables were aggregated into a single code. This resulted in a final amount of 163 concepts (codes). Because the number of codes was still large, we decided to apply four criteria in relation to the frequency of the concepts, to include them in our models. The following criteria were formulated and will be taken into account in the concept maps:

164

� ‘Causal’ relationships (A is a direct cause of B) are in all cases included in concept maps, because these relationship are the strongest relationships between variables.

� ‘Strongly impacting’ variables (A has a strong impact on B) must be mentioned at least 5 times to be part of the model.

� ‘Impacting relationships (A has an impact on B) must be mentioned over 10 times to be part of the model.

Relationships not matching these criteria are, however, not ignored. For the qualitative analysis as described in the following paragraphs (6.2 – 6.5) all interview texts are included and analysed in detail. These criteria are only applied to the model validation as described in paragraph 6.6 to get more insight into important concepts and strong relationships. Reading suggestions

In this chapter, we deliberately chose to show as much empirical material as possible. Readers who are mainly interested in conclusions at a higher conceptual level are advised to merely read the analysis paragraphs at the end of each (sub)paragraph. Readers who are also interested in how we came to these analyses are advised also to read the preceding motivating parts. Due to the high level of competition in this market, we promised our interviewees to keep their answers anonymous. For this reason, text fragments will not be specified by name, function or company and, in case of competition-sensitive issues, several company specific details (like specific services or technologies) have been left out.

6.2 The content-related side of broadband roll-out: the

impact of technology, market and policy

In this paragraph, we provide more insight into the most important factors that play a role in these domains and the way these factors affect decision-making with regard to broadband roll-out. From a technology point of view, we discuss the most important technological developments and criteria used to choose between several available technologies. From a market perspective, we provide more insight into the role of competitors, end-users and services development. With regard to the policy domain, we discuss the role of the national regulatory authority (NRA) and local public broadband initiatives.

6.2.1 Technology

Several technological factors affect the decision-making process on broadband roll-out. The most relevant technological developments, criteria on which choices

165

between technologies are made, the role of standardisation and the choice in favour of evolutionary or revolutionary technology paths, are discussed. Importance of technological developments Technological developments are considered to be very important to decision-making with regard to broadband roll-out. The following interviewee provides clear insight into the process of technological deliberation…

“Another important factor is technological developments. You’re constantly checking the

status quo of developments: what is Wimax doing? What is Wi-Fi doing? What’s going to

happen in the fixed network? What are the technological innovations? You’re constantly

keeping up with the alternatives: what are the possibilities of these technologies? Are

they cheaper and should we implement them?”

The technological developments classified by interviewees as the most relevant technological developments are developments in Ethernet and IP (also VoIP), as is made clear by the following interviewees…

“.. (Important developments are) increasing possibilities and demand for more

bandwidth within access technologies, mainly Ethernet: Ethernet in combination with

VDSL as carrier, or EttH with fiber as carrier technology. It is not a choice between

VDSL/fiber and Ethernet, but for Ethernet over VDSL or FttH.”

“..Ethernet is being applied to make it easy to bring more services into homes of the end-

users without interference between these services”

“..IP has a significant impact on transportation costs. As a derivative of the fact that IP

is becoming the uniform transportation technology, Ethernet, following in the wake of

this, has become the most obvious transmission technology. (..) Our uniform transmission

technology is being replaced by Ethernet as a result of this development”

Two interviewees also mention the development in compression technologies as being of significant relevance. One of them…

“The technological developments are such that, mainly with suppliers, via compression

technologies more and more bandwidth can be gained from the same cables.”

Technology choice As discussed above, operators are constantly keeping up with the latest technological developments and are considering whether or not to implement these new technologies. Operators mention several criteria on which they choose between technologies, among other things future-proofness, scalability and costs. Many variables play a role and need to be taken into account, as illustrated by the following interviewee…

166

“..What about the available capacity, quality and realisable coverage, considering a

given budget? As a next step, you draw capacity pictures and have a look at the way the

expected capacity of, for example, radio will develop in the future. Moreover, (you look

at) how costs will develop in the future, what the costs for the end-users will be and

which uptake you need. After that, you have to weigh these factors against each other.

And of course the number of years the technology will be sufficient and the question

whether the investment can be earned back within this time span is taken into

consideration.”

The role of standards To exclude uncertainties in the field of supplier dependency and equipment provision, and to be sure of mass production resulting in lower prices, it could be expected that operators choose internationally accepted standards when choosing between technologies. The behaviour of market parties relating to this subject is, however, not unambiguous. Market parties show diverging behaviour when it concerns following international standards in their choice between technologies, as illustrated by the following interview fragments. One party clearly shows a pioneering role, although clearly realising the disadvantages and risks…

We’re the only operator worldwide experimenting with this technology on a reasonable

scale. Does it give you a good feeling compared to Docsis which is standardised, backed

up by several suppliers and has a significant installed base? You take some risk, because

the equipment will just become affordable the moment it is produced at a reasonable

scale. You want mass to be created to get the prices sufficiently low and to prevent

becoming very dependent on one supplier.”

Where another party consciously follows proven and accepted standards…

“We’re consciously not following the path of non-standardised technologies, because

the applied elements aren’t in use on a worldwide scale yet. If you move in too quickly,

you run the risk of becoming isolated. From a national point of view, this may be

acceptable, but we operate on a European scale. For that reason, we only use

internationally accepted technologies and standards.”

167

Analysis

It becomes clear that operators are constantly keeping up with the latest technological developments. The most important developments are at the protocol level of the networks, consisting of developments in Ethernet and IP. Cable operators consider the development of the Docsis42 protocol highly important. Also, the development in compression technologies is considered highly important, which appears to make sense because these technologies make it possible to transmit more and more data over the same networks. As a result, high and risky infrastructural investments can be postponed and more time is left to gain revenues from the sunk investments in the existing infrastructures. Choices between technologies are based on many characteristics of the available technological alternatives. In choosing whether to follow international standards, different considerations sometimes clearly play an important role. A reason for following standards is to avoid an isolated position as a result of being among the first to move. Financial considerations are a reason not to follow a worldwide standard. Among the tasks that have to be completed before DOSIS 3.0 roll-out can start are hardware certification and operator tests, spectrum re-allocation, support system transitioning, head end re-cabling and re-combining, and back office support (Junkus, 2008). These adjustments demand significant network investments, which must be balanced against investments in other technologies.

6.2.2 Market

In addition to technology, there are several developments in the market that also influence the choice between the available technologies and the upgrade path towards these technologies. In this paragraph, we discuss the impact of competition, end-user demand, service development and the ways operators cooperate in the value chain.

Competition As discussed in Chapter 2, the broadband market in the Netherlands is highly competitive. Decision-makers consider competition a very important driver for strategic decisions. Although market parties are following their own strategy and will not blindly copy technological choices in imitation of their competitors, they keep a close eye on each other. Two interviewees illustrate the attention paid to the behaviour of competitors…

“You have to keep an eye on what your competitor is doing. This is a highly important

factor...”

42 Data Over Cable Service Interface Specification (DOCSIS) is an international standard developed by CableLabs and contributing companies for data transmission over coaxial cable. See also chapter 2.

168

“All operators eventually reason from a ‘me too’ perspective: the competitor has it, so I

must have it too. This was also clearly visible in The UMTS world. Many operators knew

that UMTS was not financially feasible, but market parties rather become ruined together

than that somebody dares to differ from the others and do something else.”

The high level of competition in the Netherlands has had some important effects. The first effect mentioned by decision-makers is innovation in technology and services. A second effect is acceleration of network upgrades, resulting in a leading position of the Netherlands within the European Union in broadband roll-out, as stated by these interviewees…

“In case KPN would not have accelerated its ADSL roll-out, we wouldn’t have made our

network suitable for two-way traffic in the countryside so quickly and dropped our prices

so much. Connections are namely quite expensive in these regions. This also goes for

KPN.”

“The Dutch competition in the Internet field can, ultimately, be attributed completely to

two parties: KPN and ‘the cable’, which will try to outscore each other in the coming

years. The result of this competitive battle is that the Netherlands now have a leading

position in Europe in the area of broadband development.”

Another important effect is positive for end-users but less favourable for the operators themselves. All interviewees mention a “price war” as a direct result from the high level of competition. As a result of this price war, more and more bandwidth has to be offered against the same prices, which reduces the operators’ margins. Two interviewees…

“Offering more and more against the same prices is done under competitive pressure.

You cannot stay behind.”

“The pie has largely been divided. You’re upgrading your customers towards higher

speeds against the same price, hoping to take away customers from your competitors.”

The positive effect of this price war is the fact that is has clearly improved the ability to adopt broadband in the Netherlands (Vermaas, 2007, p.207). This race to offer the highest bandwidth at the lowest price is, however, not such a positive development for the operators, because it puts their margins under enormous pressure. The fact that, in spite of these decreasing margins, the price war goes on is even more striking, considering the fact that all operators recognise that the needs of the majority of end-users do not lie in high bandwidth subscriptions, but in the cheaper, lower speed subscriptions. This awareness is demonstrated by the following three interviewees, form three different companies…

169

“One rolls out ADSL2+ to be able to offer TV and higher Internet speed, but the need of

the customer still lies in lower speeds against lower prices”

“..Most demand in the market lies around 4 Mbps download speed. What we do see is

more demand for higher upload speeds. We also offer 20 Mbps. It’s the early adopters

buying this product and it is relatively expensive. The average wage earner cannot afford

these products. For him, it is important that the product is easy, well installed and that

we offer him good service.”

“The core of the problem within the decision-making around networks lies in the fact that

the needs of residential and business end-users for bandwidth and services are not

developing as quickly as many in the outside world proclaim.”

These insights are confirmed by Vermaas (2007), after extensive research among end-users in the Netherlands. All operators we interviewed, however, clearly see themselves forced to offer these higher bandwidths in the interest of a positive perception of their companies by end-users…

“Transmission speed is very important in marketing and determines the reputation of a

provider. For that reason, we closely follow the market, so consumers cannot use speed

as distinguishing factor for their choice.”

“The fact is that consumers buy a ‘brand’, in other words, buying from an operator that

is also able to offer a high-speed service. So, you’re forced to also offer a high-speed

product to convince buyers that your product is a good one. Although they take the slow

product, they look at the fast ones.”

“..For the rest, it works the same as with car brands: you have regular brands, but on top

of that you have one model that really rises above them. This brand is increasing your

reputation in such way that it arouses the association with you’re your leadership in the

field of exciting cars”.

(Expected) end-user demand Although the above mentioned aiming for customer satisfaction is, financially seen, in most cases not the most optimal business strategy all operators are participating in the price war, which is very understandable. End-users, ultimately, make sure that revenues are generated. For this reason end-user demand is considered to be a very important variable for decision-makers. An important problem, related to end-users is, however, that the behaviour of end-users is hard to predict, that they are capriciously and that there is no such thing ‘the broadband adopter/user’ (Vermaas, 2007, p.208). Decision-makers from all companies confirm this problem related to the unpredictability of end-user behaviour and the uncertainty of the moment the need for higher bandwidths will occur. This shared view is illustrated by…

170

“Insight into future demand is very limited. As a result of this, you have to make

investments which may yield returns in five or ten years. This very risky business case

makes it very difficult to attract money. You have, as an example, no clue whether HDTV

will be adopted at large scale in one, three or ten years, but it is crucial for the way you

have to further develop your network..”

Analysis

In the market domain, competition is a very important factor in deciding to upgrade their networks. The high level of competition has brought about three important effects. Firstly, innovation in technology and services is a clear result of competition. Secondly, competition has brought the Netherlands to a top position within the world in the field of broadband roll-out. Finally, it resulted in a price war for broadband access with more and more bandwidth for lower prices. The latter two results are closely related. Higher levels of broadband adoption are the result of this price war, due to the fact that the financial threshold for broadband adoption is removed. This is confirmed by many user-oriented studies, for example Vermaas (2007), who states that that ‘convenience’ at reasonable costs’ is highly important in people’s decision whether or not to adopt broadband. It can, however, be concluded that, concerning the choice between operators, another factor plays an important role, namely the image of the provider, which is determined by its highest broadband offer. Although this factor is highly important, it is not related to the actual demand of the end-user. To maintain existing subscribers and especially to gain market share at the expense of competitors, two aspects are necessary. Firstly, offering regular adopted service packages against the lowest market price and secondly, offering and marketing of a high-speed reputation-enhancing broadband product.

End-user demand is an important driver for upgrades, although the boost to a company’s image, related to high bandwidth offers, appears to be an even more important driver. Customer satisfaction and reputation-building are even more important drivers than actual and expected demand. Exactly when a critical mass for really high bandwidths (above 10Mbps symmetrical) will occur remains an important uncertainty for operators.

6.2.3 Policy and regulation

The final dimension we discuss in this paragraph is policy and regulation. In Chapter 2, the most important regulation and policy applicable to network operators have been discussed. We will now pay attention to the extent policy and regulation affects decisions related to network upgrades and roll-out. Firstly, we discuss in this paragraph the impact of the national regulatory authority (NRA), being OPTA in

171

the Netherlands. Secondly, we will pay attention to the impact of local public fiber initiatives on market players. Impact of the NRA Operators mention several ways the NRA affects their business management. OPTA is able to impose several obligations to operators that result in a reconsideration of investment-related decisions. The analysis of relevant markets as well as the possible obligation to allow competitors to use one’s network has an impact on market positions of parties. This view is substantiated by...

“OPTA is able to impose obligations as a result of which investment-related decisions

must be reconsidered.”

Another interviewee mentions the impact of the market analyses for the new regulatory framework:

“OPTA’s market analyses could have consequences for one’s market position. At the

moment, we’re not obliged yet to allow other operators on our network for Internet, but

that could happen and that has an impact on one’s market position.”

Regulation, however, is not seen as a leading factor with regard to strategy. Operators consider regulation a given, a conditional framework within which they have to operate, but they eventually follow their own course. This view was echoed throughout the interviews, for instance in the following section…

“Regulation may be a condition for our ability to follow our course, but definitely not the

other way around. VOIP, for example, has been introduced while the market analyses of

OPTA were still going on. As an operator, you will die if you have to wait for their

judgments, that’s impossible.”

Local public fiber initiatives In the Netherlands, mainly between 2003 and 2006, many local governments initiated the roll-out of public fiber networks to stimulate broadband roll-out (for more details, see Chapter 2). We asked operators what the impact of these initiatives is on their strategy. In general, operators have a negative attitude concerning fiber initiatives undertaken by local governments. The most obvious reason is that they consider them additional competitors in an already highly competitive market with low margins, which results in even more pressure on their business cases. However, operators have other concerns as well. Firstly, the inextricability of the public and private role is a crucial issue for operators. On issues where operators used to work together with local governments (for example the geography of their networks), they are now reluctant to cooperate, and in some situations they will only do so when legally obliged. The reason for this reluctance is that governments may use this information for competitive objectives. Another concern is that, when these networks are no longer viable for local governments,

172

they may be sold to a market party for much less money than telecom operators have invested in their fiber networks. As one of the interviewees remarked…

“There is only one danger: the government invests, a network is being built, the network

financially falls down and some market party buys the bankrupt property against a cheap

price. This already happened in Nuenen: there is subsidy for citizens, the network

collapses and a construction firm buys the network.”

Public involvement has several positive as well as negative effects. Firstly, the interference by the public and private role of governments causes uncertainty, which could have a negative influence on investments. As one interviewee states…

“Governments run the risk that, in case they start disrupting the market with cross–subsidies,

they could scare off foreign investors. This is what, for example, happened with shareholders in

the US. They didn’t want to invest their money anymore in the Netherlands due to the uncertain

investment climate as a result of interference of the Dutch government in the market.

Governments must be careful with these kinds of things.”

Another interviewee mentions the difficulty to compete with public parties…

“Interference by governments in the private telecommunication market via subsidised

services could result in a break off of our roll-out. In the end, one cannot compete with

public money.”

On the other hand, policy operators indicate that public involvement in the market has also had a stimulating effect on the market, as two interviewees mentioned…

“Take, for example, Surfnet. Surfnet connected universities with subsidised money. Our

company could also have connected them, but universities probably could not have

afforded it. We also learn from these initiatives: what were successful aspects we can

copy in our own initiatives and what was not successful? This way, operators keep a

close eye on these initiatives.”

“Local initiatives cause small hypes, by which all kinds of things happen in the market.

In case the government wouldn’t have done anything, we would be less far in the

Netherlands.”

One operator mentions as a positive result that the company involved is making more visible to end-users what the company is and does…

“Large telecom companies have been too much technological, introvert companies.

These fiber initiatives have made us exhibit more what the possibilities and capacities of

our company really are”

173

Analysis

We can conclude that regulation by OPTA clearly affects the decision-making process, but that this regulation does not play a leading role in the strategy of telecom companies. Regulation is considered a given with which operators have to deal. The fact that regulation only plays a minor role in decision-making processes compared to financial and technological considerations was demonstrated earlier in our literature analysis (see Chapter 3). Although public fiber projects have been initiated to stimulate broadband roll-out, the leading position of the Netherlands can be attributed mainly to the infrastructural competition between cable and copper infrastructure. Public initiatives have, nevertheless, certainly contributed to broadband development and created understanding among operators with regard to the need to change their monopolistic attitude to a more customer-oriented one. On the other hand, governments should be conscious of the possible negative side effects of their interference in the private market. Foreign market parties indicate they are considering postponing or terminating their investments in the Dutch market due to an instable and distorted market environment. Moreover, market distortion takes place due to new players buying up the networks of bankrupt broadband initiatives. Generally speaking, operators consider public fiber initiatives a distortion of the private market and have a negative attitude towards an unclear mixture regarding the public and private role of local governments.

6.3 Risk & uncertainty

In Chapter 4, we introduced a theoretical framework based on Meijer et al., 2006, Walker et al., 2003 and Courtney et al., 1997. This framework includes classifications of risk and uncertainty on three dimensions: nature, level and source. In our qualitative research, we will validate the dimension of source and whether the several sources of risk and uncertainty as introduced by Meijer et al. are experienced by strategic decision-makers on broadband roll-out and how these sources manifest themselves in practice. The classifications to nature and level are measured in our quantitative research and are discussed in Chapter 7. As a next step, in this paragraph we provide more insight into the most important risks and uncertainties, as indicated by our interviewees. After that, we focus on which methods aimed at managing risk and uncertainty, as described in Chapter 4, are applied by decision-makers in practice. Finally, we take a look at the effects of these risks and uncertainties.

6.3.1 Sources of risk and uncertainty

Based on the Five Forces of Competitive Position model by Porter (1980), discussing uncertainties such as the behaviour of new entrants, substitutes in the

174

market, buyer power, supplier power and rivalry, Meijer et al. (2006) make the following distinction with regard to sources of uncertainty: technological, resource, competitive (new entrants & rivalry), supplier, consumer uncertainty (buyer power) and political/regulatory uncertainty (see also Chapter 4). On the basis of our in-depth interviews, we examine whether these sources of uncertainty are also experienced by decision-makers in the broadband market and how they manifest themselves. Technological risk and uncertainty Uncertainty about the technology can refer to the technology itself, to the relationship between the technology and the technological infrastructure due to system changes or to the availability of alternative available and future technological solutions (Meijer et al., 2006). The first category of uncertainty, the technology itself, concerns the stability of the networks. This uncertainty is caused by the explosive growth in the number of customers or new technologies, as illustrated by two interviewees…

“..The biggest problem lies in the fact that, due to explosive network growth, IT systems

must support five times more customers than five years ago. IT systems are, however, not

scalable enough to handle this, which causes them to reach their boundaries.”

“In case of new technologies, the stability of these technologies forms a major

uncertainty. It concerns stability towards end-users, but also when replacing parts of the

network and sharing means within the network like new line cards, new chassis, new

routers, etc.”

The second type of technology uncertainty refers to whether and to what extent adaptations need to be made to the technological infrastructure to connect a new technology to the infrastructure (Meijer et al., 2006). Two interviewees illustrate the uncertainty to determine the practical impact on the infrastructure as a result of technological adaptations...

“Immature technologies come together with uncertainties in the field of standardisation,

uptake and availability. We also have (in this case) a detrimental risk because you start a

migration path overnight which closes the way back. Another risk is what it will mean for

your organisation. You put down boxes everywhere in your network, but is the technology

really future-proof, although we now think it is?”

“An ever-present risk is that your equipment does not function the way you expect it to

do. You try to copy as many available things as possible. Bu, the fact that we always try

to distinguish ourselves by means of new technologies and new services means you have

to be a frontrunner and that you can’t copy everything. This implies that sometimes you

can’t offer the right quality or the right quality against the right costs.”

175

The third type of technology uncertainty has to do with choices between technologies. Telecom operators are faced with quite some risks and uncertainties with regard to the technological choices that have to be made. Operators are constantly examining prospects of new technologies, like price, technological possibilities and performance in densely or sparsely populated areas, and must decide whether existing technologies should be replaced or not. Operators see the necessity to upgrade their networks. Although it is not known when this necessity will exactly occur and when a company should invest, the operator must be prepared for the upgrade and must have all options ready follow competitors quickly enough, when the need arises…

“Uncertainty is also shaped by the versions you choose: VDSL 1 vs. VDSL 2, ADSL 1 vs.

ADSL 2. What is the direction most parties are following? Which vendor and which

underlying choice is eventually going to give you the best price?”

Other risks associated with implementing new technologies are the risks of availability of the new technology, standards (will the technology be standardised so mass production will occur), risk of irreversibility of the chosen technology path, risk of the chosen technology getting outdated and path dependency (future technological choices depend on previous paths set in). Resource risk and uncertainty Meijer et al., 2006 define three aspects of resource uncertainty: uncertainty about the amount and availability of raw material, human resources and financial resources (see also Chapter 4). Operators indicate they are faced with all these three types of resource uncertainty. One operator clearly illustrates the uncertainty in relation to human resources…

“During network upgrade you really have to deal with priorities concerning allocation

of people: should you renew something or allocate people to fix problems in the existing

network? This process is difficult to manage. Operation has to do with people and you

are dependent on operational issues that do or do not function well. This is very

complex.”

Another interviewee states that the uncertainty in networks upgrades has to do with the future availability of equipment when a certain technology is chosen…

“..It’s the uncertainty of technological developments. In case you make a certain

technological choice, is your supplier following that development?”

Several operators mention financial risk and uncertainty as extremely important uncertainties related to upgrading networks…

176

“One of the most important risks is unforeseen additional costs as a result of technology

worse functioning than expected. Integration with the environment comprising IT

integration and optimisation with regard to maintenance, billing, supply, etc., causes

the most uncertainty. Adjustments, also in technology must be done more often than

desirable to realise this optimisation.”

“Technological developments are succeeding one another faster and faster. The result

is that the first investment has not been paid back yet or another is already presenting

itself. Your business case is based on a payback time of 5 years, but after two years a

new technology is already there, reducing your payback time to only three years.

Technologies are succeeding one another so fast that you’re forced to work with shorter

time lines all the time.”

Competitive risk and uncertainty Competitive uncertainty refers uncertainty about the actions of competitors and the possible effects of these actions on the competitive position of the own organisation (Meijer et al., 2006). As we saw in paragraph 6.2.2 of this chapter, competition is one of the main drivers for network upgrading and roll-out for telecom operators. In case a competitor starts upgrading its network, other companies must follow or fall behind. Telecom operators must always be ready to have an adequate response to what the competition does. But the comings and goings of competitors are hard to predict, as becomes clear in what the following interviewees see…

“Competition plays an important role, for example between DSL and cable. In the past,

you saw cable operators being the first parties offering broadband Internet by

gradually skimming off the market. Then the DSL providers came up and they took over

the market. At the moment, cable operators are regaining market share with TV and

VOIP. These market developments are difficult to influence and predict.”

“There’s uncertainty about the direction competitors choose: what are the others going

to do? You have to keep an eye on the moves of your competitors.”

Operators see a major risk in not following and anticipating the actions of their competitors. Because the effects of their actions lead to high levels of uncertainty about the own future market position…

“In case cable companies are starting to upgrade their networks and start offering high

bandwidth against low prices on a massive scale and your network is not up to date to

offer these capacities, you’re the one being left with the bill”

“..It is ultimately a major risk not to come along. When everybody is ruined, your

competitor will at least also will be ruined and will not run off with the first mover

advantages and the entire market.”

177

Supplier risk and uncertainty Uncertainty about the actions of suppliers refers to uncertainty about the timing, the quality and the price of the delivery (Meijer et al., 2006). As discussed earlier, suppliers play a key role in the decision-making process about network roll-out and upgrades. Choosing a technology and roll-out takes place in close cooperation with suppliers, because technological developments are changing quickly and become so complex that operators need suppliers for their specialised knowledge. This close involvement and specialised knowledge gives suppliers a significant amount of power, which increases supplier dependency, as illustrated in the following fragments…

“Supplier dependency has been increasing strongly within the last years, because

equipment becomes more and more complex and technological developments evolve

very fast. Suppliers install their equipment by themselves during a roll-out, we’re often

not even involved in this process anymore.”

“At several network layers, the power of suppliers is increasing due to the fact that

technology and equipment are becoming more complex all the time. This supplier

dependency manifests itself especially at the service layer. Multi-vendor policy is

difficult at this layer, because the service layer is tightly interwoven with customer

processes. Operating with more parties is complicated because it demands much more

time and because this could mean that the same service will show small differences

between different geographical areas.”

Consumer risk and uncertainty Meijer et al. (2006) mention three types of consumer uncertainty: uncertainty about consumer preferences, consumer characteristics and end-user demand. Uncertainty about consumer preferences is clearly expressed by the following interviewees:

“An extremely important uncertainty is the uptake: will it be successful or not?”

“The most important uncertainty is whether the standard or service will eventually be

adopted by the market...”

Also, the changing behaviour, especially the fact that end-users become more and more demanding, results in higher risks for operators…

“Another risk we see is that the services we offer are becoming more and more important

for business management of our clients. You have to deal more and more with claims and

proofs of default the moment you’re not operating according to performance agreements.

In case services of a telecom provider aren’t functioning, this often implies the revenues

of a client are on hold...”

Uncertainty about demand is a very important factor in decision-making about which technological paths to follow. Forecasting demand is extremely difficult,

178

because future demand is very unpredictable and hard to measure. Operators have to base their roll-out on a certain network occupation, but at the same time this is their biggest uncertainty…

“Influencing the market is very difficult and complex because consumer behaviour is

almost unfathomably. You’re never able to forecast whether it will evolve slowly, fast,

or not at all. What kind of new services will really be successful on a large scale?”

It is not just the unpredictable behaviour of end-users that causes high levels of uncertainty. Also, the different visions operators have on the future of this uncertain market adds an extra dimension of risk and uncertainty, as noted by the following interviewee…

“The market forms the most important uncertainty. How open are you going to make

your networks? In case you share the same opinion on future demand, it is easier to

share networks and technologies with other parties, which strongly affects your

strategy. The more you share a vision with other parties, the opener you are able,

together with the ‘telecom market’, to aim for the same solutions. The more diverse the

picture, the more secluded everybody is working and the higher the risks, because

everybody has chosen a different path. This costs much money and brings along

uncertainties.”

Uncertainty about the characteristics of end-users was not specifically mentioned during our interviews. Vermaas (2007), however, clearly proves the unpredictability of end-users, which causes this type of uncertainty for operators: “adoption triggers and thresholds are not absolute but subjective and the broadband adopter/user does not exist” (p208). Political risk and uncertainty Political uncertainty (also known as regulatory uncertainty or government-related uncertainty) stems from perceived uncertainty about government behaviour, regimes and policies (Meijer et al., 2006). Meijer et al. mention five causes of political uncertainty: existing policies, unclear or inconsistent regulation, a lack of regulation, future changes and government behaviour. All operators indicate that regulation leads to uncertainty. One interviewee clearly mentioned existing policies as a cause of uncertainty, consisting of the longer routes operators have to take before they know issues like price developments and technological migrations are approved or not…

“An important factor is that OPTA routes and the entire circus of objection and appeal

afterwards last very long. We’re still waiting for a verdict regarding a dispute we started

in 2000. The process has improved now, but in this world a year is already too long, you

already have to move on by then...”

179

Uncertainty about future regulatory changes was mentioned several times during interviews. The following interviewee indicates that, besides the question whether certain regulation will be implemented, an even more important question for operators is when…

“We experience much uncertainty regarding when regulation is going to take place. All

these issues have to do with money, investments. And where investments are involved, it

is very important to find the right moment. Timing is key.”

Another interviewee mentions uncertainty about which way the new market definitions will influence the business and market position of telecom operators and which way the regulation of emerging markets will turn to, are important issues for existing telecom operators. He remarks…

“..The phase of the highest uncertainty is the current phase, in which you’re not sure

which direction regulation will go.”

A straight-forward example of uncertainty due to government behaviour and a lack of regulation is the (absent) regulation of OPTA regarding the next generation network strategy of the Dutch incumbent KPN. The problem is that alternative providers, interconnecting on the network of the incumbent, are highly dependent on the verdicts of OPTA concerning the plans of KPN to phase out the MDF locations it now uses as interconnection points. Alternative operators clearly experience high levels of uncertainty as a result at the moment, as clearly expressed by this interviewee…

“What will OPTA’s vision be on the possibilities for phasing out the incumbent is

offering? And on the term within which the incumbent is allowed to phase out their MDF

locations? That is very determining for us. How long can we proceed with our own

MDF43 network? Until 2009 or 2010? These differences are quite important. What will

be OPTA’s vision on the development of tariffs on the 20 RAPs44 we’re now

interconnecting on? What if these tariffs will be released again?”

43 Main Distribution Frame 44 Regional Access Point/current points of interconnect

180

Analysis

In this paragraph, we illustrated that, within the telecom market, the six sources of uncertainty from the framework by Meijer et al., 2006, as introduced in Chapter 4, are clearly visible. Decision-makers are faced with technological, resource-related, competitive, supplier-related, consumer-related and political/regulatory uncertainty. Uncertainty about the technology can relate to the technology itself, to the relationship between the technology and the technological infrastructure due to system changes or to the availability of alternative existing and future technological solutions. Operators indicate they are faced with all three aspects of resource uncertainty, which are uncertainty about the amount and availability of raw material, human resources and financial resources (see also Chapter 4). Telecom operators must always be ready to respond adequately to what the competition does. However, the comings and goings of competitors are hard to predict. Operators see a major risk in not following and anticipating the actions of their competitors, because the effects of their actions lead to high levels of uncertainty about the own future market position. Supplier uncertainty has its cause in a strong supplier dependency. Operators become more and more dependent on suppliers due to their specialised knowledge which is needed by operators because technological developments are changing fast and are becoming very complex. For that reason choosing a technology and roll-out occurs in close cooperation with suppliers. The three types of consumer uncertainty (uncertainty about consumer preferences, consumer characteristics and end-user demand) are clearly recognised by operators. All operators indicate that regulation leads to uncertainty. Regulatory uncertainty is caused by too long legal processes concerning approval or disapproval of issues, uncertainty about future regulatory changes and about which way the new market definitions will influence the business and market position of telecom operators.

6.3.2 Effects of risk & uncertainty

Operators indicate that the way risks and uncertainties affect technological choices strongly depends on the urgency of the problem. When it concerns a roadmap of a service that may be successful in approximately two years, operators will investigate more specifically whether this is the right path to follow. In case of high urgency and more rigorous steps, operators are in most cases forced by their environment to choose the best path of that moment…

“ ..In case of high urgency, one has to rush aside certain uncertainties. Practice

sometimes forces one to make choices.”

181

An important effect of risks and uncertainties operators mention is that they keep open several possible development paths, as explained by the following interviewee...

“In case preliminary steps already must lead to a final decision, the risks are very high,

because in these cases path dependency and irreversibility arise. For that reason, there is

some sort of decision tree in which several paths are kept open. Without these

preliminary paths, the way towards FttH would already have been closed off, but now it

is still kept open, although the final decision has not been taken yet.”

Marcus (1981) indicates that political uncertainty can affect the rate, timing and substance of an innovation. This insight is clearly expressed by two interviewees…

“(…) because too many paths are kept open, these factors (Regulation, OPTA and

politics) have a delaying impact on decision-making for a certain option. All these

alternative solution paths demand energy and resources, which is not practicably and

reduces focus. This way, these factors are also delaying the roll-out of technologies and

services.”

“..In case one would face a large investment decision, like the roll-out of fiber, a climate

of uncertainty at the level of political regulation will make one hesitate whether one is

making a wise decision by doing that.”

Analysis

Risks and uncertainties have two important effects. The first is building in flexibility by keeping open more optional development paths. This way, different development steps can be chosen depending on the circumstances, and adjustments can be made when the environment changes. Building in flexibility and keeping open more options for the future is a clear example of practically applied real options thinking. For that reason, it is striking that Real Options Analysis is not applied as a method for financial assessments within telecom companies (see paragraph 6.4.4). Although operators in practice act according to this method, they do not incorporate it in their investment-related decisions. The second effect is delay or reconsideration of investments. Although it is difficult for regulators to keep pace with the fast developing telecom market, they should be aware of the impact the uncertainty about the outcome of regulatory processes have on the decision-making processes of telecom operators. A climate of political and regulatory uncertainty acts as a threshold to operators with regard to major infrastructural investments.

182

6.3.3 Dealing with risk & uncertainty

After discussing the several sources and effects of risk and uncertainty, we now focus on ways operators try to deal with the various risks and uncertainties and how they try to minimise the negative effects. In Chapter 4, several possible ways of dealing with risk and uncertainty were introduced (technological analysis, cost-benefit analysis, obtaining information from suppliers, benchmarks with competitors, forecasting and scenario investigation). Based on our interviews, we investigate whether these methods are applied by decision-makers to reduce risks and uncertainties. In our interviews, decision-makers were asked what methods they apply in their daily practice to handle risk and uncertainty in their decision-making processes. Most operators mention several kinds of analyses they use to reduce risks and uncertainties they face during decision-making.

Two interviewees on technology analysis… “Very attentive preliminary work is being performed by making analyses, etc. This

strongly reduces the change that a technology becomes a failure.”

“We keep an eye on all possible options. A special technology board keeps an eye on

technological developments, consults people from outside and investigates what other

companies or small start ups are doing concerning new technologies and the way our

company can use that.”

One interviewee on impact analysis… “The operations department is being asked what a new technology will mean to for them.

As a next step, we try to quantify this. Impact analysis is about the impact on your

organisation, expressing this impact in financial terms and risks and asking the question

whether these risks can be covered.”

Another interviewee on scenario analysis…

“For this technology we made three scenarios with several values for different

parameters: Assuming this is the world in five years, what will be the best solution?”

One operator mentions his company looks at comparable situations involving companies in other countries where the technology has been implemented before, to see how things have worked out there and what can be learned. The method involving gaining information from experts was mentioned by several decision-makers, as illustrated by two interviewees…

“As many information as possible is being obtained from all sides and experts to

minimise uncertainty.”

183

“We have several vendors who can provide us with information on new technologies. As

far as possible, information is obtained from several players, to prevent supplier

dependency. In addition, a special group of architects explores all kinds of technological

developments that occur everywhere. They function as radar, measure expectations and

look at which expectations actually become a reality.”

One interviewee mentions the use of benchmarking…

“The R&D department is responsible for all technological developments and looks at what

technologies are available and what other operators are doing.”

Forecasting is indicated as being an important method for building business cases and making decisions to upgrade or roll-out a network. Forecasts are made for several factors, like bandwidth development, price developments, market positions of competitors and the final position of the own company. Interviewees indicate that the most important variable with regard to forecasting is end-user demand…

“Demand forecast is extremely important. The expected demand is the most important

factor that determines whether one is going to roll-out and if so, where and how much

capacity.”

“On the one hand, there’s end-user demand, on the other hand, there’s our demand

towards suppliers. In case a decision has been taken marketing follows the market

realisation and expectations. On the other side of the value chain, we deliver forecasts to

suppliers to let their dimensioning match, to prevent large price increasing and to

prevent equipment not being available the moment it is needed."

Conducting pilots is a method that is applied regularly to gain insight into potential problems before rolling out a technology on a large scale. In case a certain technology causes many problems, the decision to roll-out this technology can still be reconsidered or delayed until the problems are solved. Two examples...

“To prevent childhood diseases during implementing new technologies, all new

equipment and IT systems are tested within the own network. Technologies do not always

perform the same in practice as they do in a lab. Only in case the technology functions

optimally within a pilot it will be rolled out within the rest of the network.”

“One wants to understand what rolling out a certain technology implies. By just taking a

decision one runs the risk getting confronted wit all kinds of unpleasant surprises like a

roll-out costing twice the price you expected. For that reason a pilot will be performed to

exclude uncertainties because the investments are enormous.”

184

Analysis

To reduce risk and uncertainty, several methods are applied by operators. Of the methods introduced in Chapter 4, the use of technological analysis, scenario investigation, obtaining information from suppliers, benchmarks with competitors and forecasting are applied by decision-makers to reduce risk and uncertainty. Of these methods, technological analysis, forecasting and pilots are applied the most. Analyses and forecasting take place early on in the decision-making process, when alternatives are investigated. Pilots are performed later on in the process to prepare for possible practical unforeseen side effects with regard to the implementation of new technology. On top of the methods introduced in Chapter 4, two more methods are frequently used: impact analysis and pilots. Cost-benefit analysis was not mentioned as an uncertainty-reducing method. This can be explained by the fact that this is more seen as a financial assessment method than a method to reduce risk and uncertainty. As can be seen in paragraph 6.4.4, cost-benefit analysis is applied by decision-makers as a method to evaluate investment options.

6.4 The decision-making process on broadband roll-out

After discussing the content-related side, the following two paragraphs focus on the process-related side of our qualitative research. In this paragraph, we provide greater insight into the course of the decision-making process, several types of decisions and the actors involved. Insight will also be provided into the preference of decision-makers for the applicability of financial assessment methods in technological investment-related decisions. In paragraph 6.5 the validation of the applicability of the decision-making theories of Logical Incrementalism and Garbage Can will be discussed.

6.4.1 Who’s in charge?

Based on the theory of Simon (1960) and Lin (2004) gives a distinction between two main types of decisions: strategic or non-programmed decisions and programmed decisions. A further distinction was made with regard to programmed decisions between short-term operational decisions and periodic control decisions (see also Chapter 5). Short-term operating control decisions are simplified decisions like stock control and are performed by middle management. Periodical control decisions occur regularly but infrequently and are performed by operational management. During interviews a clear distinction was visible between strategic and operational decisions.

185

Non-programmed decisions

According to Lin (2004), strategic decisions are primarily the concern of senior management. This statement is confirmed throughout the interviews, as underlined by this interviewee…

“Investment-related decisions are taken by the board based on business cases and

underlying information.”

“Strategic decision-making takes place at a high level: the management team fixed lines

and the board of directors with the Board of Commissionaires looking over their

shoulders to make sure the decisions are made the right way.”

Strategic decisions are also made by a small number of people as well.

“The decision-making and initiation of new things is performed by a small group, of

course with some support here and there.”

Programmed decisions

Programmed decisions are initiated by operational departments. Also, more people are involved when it concerns operational decisions. After the main strategic decision has been made, smaller, more operational decisions are made in the roll-out and exploitation phase. This clear distinction is illustrated by the following interviewees, (with the first interviewee referring to the phases in the model presented in Chapter 3)…

“The main decision is made in phase 1(e.g. decision-making phase). The smaller

decisions like whether there will be 100 Mbps or which applications will be offered are

made during the roll-out and move to phase 2 (e.g. roll-out)”

“Smaller decisions to shape the product during the process are made in the exploitation

phase. Major decisions concerning the entire budget are made before that.”

A clear example of a short-term operating decision is the decision to upgrade the network based on marketing forecasts. These forecasts take place on a weekly or monthly basis and it is based on these forecasts that a check is made to see whether enough line cards are available in the exchanges to meet demand. The routine character of these decisions is illustrated by the following interviewee…

“Decisions which have to be made when the network reaches its maximum capacity are

actually taken on autopilot. It is determined in advance which step has to be taken. At a

certain moment there is a trigger based on a forecast indicating that somewhere the

186

network reaches its capacity. Based on that a simple plan is made, a financial decision

is made based on this plan and the plan is performed. There’s little excitement in that.”

Periodical control decisions occur less frequently. These decisions are performed by the operational department and concern technological network extensions that can be performed with the existing network equipment but involve more than merely adding line cards. An example provided by one of the interviewees is the upgrade to a next version of an already implemented protocol, like the transition of Eurodocsis 1 to 2…

“The process of Eurodocsis almost takes place in the operational organisation.

Capacity extensions are initiated by the operational departments. Also, more people are

involved in the roll-out phase.”

Analysis

It becomes clear that the distinction between non-programmed and programmed decisions, as introduced by Simon (1960), as well as the further distinction between short-term operational decisions and periodic control decisions (Lin, 2004) can very well be applied to the different decisions being part of the process of broadband upgrade and roll-out. Smaller decisions concerning the implementation are moved to operational departments and involve more people. The major strategic decisions, concerning, for example, investments, are taken by senior management in small groups.

6.4.2 The duration of decision-making and roll-out

This paragraph will pay attention to the duration of the decision-making process and the roll-out of chosen technologies, as well as factors determining these durations. From first idea until strategic decision

As mentioned in the previous paragraph, decisions to roll-out or upgrade a broadband network are strategic decisions. These strategic, high-impact decision-making processes are processes of which durations can strongly differ. This is illustrated by the following interviewee…

“An average means little in this case, because durations can vary strongly between

three months and, in the most complex and risky cases, a few years. The decision to dig

a new physical network will take years. An update of systems software can be decided

upon in a few months.”

187

Two interviewees indicated durations of half a year and a year to arrive at the first strategic decision. An illustration by these interviewees…

“The period to arrive at the decision to implement ADSL2+ took half a year”

“The phase of decision-making takes 4 to 5 months, selection of suppliers included. So,

pure decision-making and making preparations.”

However, to arrive at a final decision after all the details have been worked out, the total duration is a year, as illustrated by two interviewees…

“This strategic direction has been decided upon in half a year. At the moment, details

are being worked out. It will take at least another half year to reach a final decision.”

“…such a process scan takes a year, counting from the moment suppliers come up with

an idea until the first spade in the ground...”

Which factors determine the duration of this decision-making processes? A first clear determinant is complexity, as mentioned by the following interviewee…

“The more complex and radical the change, the longer the duration of the decision-

making process.”

Another interviewee explains the impact of complexity in greater detail…

“The duration depends on all kinds of factors, like the complexity of the technology, the

scale of changes that must be carried out in the network, the level of investments, impact

on operations and the related operational costs, etc. Duration also depends on the layer

you’re operating on: decisions to replace the entire physical infrastructure or to just

replace the transmission layer have other investment dimensions and other decision-

making durations as a result of this.”

Additional determinants mentioned by another interviewee are market developments, uncertainty and financial factors. The role of financial factors is confirmed by several interviewees, as indicated by the following…

“…the magnitude of the expected investments and the returns of the invested capital are

leading in this.”

“…in case of technologies like EttH, FttC and FttH, one speaks of such large amounts

of money that decision-making processes are very lengthy.”

188

The presence of a certain ‘sense of urgency’ is an important influencing variable with regard to decision-making. The impact on the duration of the process is mentioned by several interviewees…

“…the absence of urgency can lead to the situation where no decision is made.”

“Importance: discussions can evolve very fast in case they are really important. There

must be a feeling of urgency to make the decision because why else should one take a

decision?”

From the first spade towards the entire footprint

After the decision has been made, the chosen technology can be rolled out. This roll-out is another time-consuming process. Adjustments to systems and building platforms or single locations maximally take one year time…

“VoIP has been rolled out in a couple of months. For this service, switches had to be

adjusted and aspects for network control had to be arranged.”

“…the moment a location is physically full, one has to build a new location to install or

extend equipment. Then we’re talking about a lead-time of three quarters of a year to

one year.”

On average, rolling out evolutionary upgrades takes one to three years. The following fragments clearly illustrate the roll-out time of several evolutionary steps in the DSL network…

“For ADSL, it took three years to make the entire network suitable but the process

started slowly. ADSL2+ roll-out happens in one year. VDSL probably will take three

years. ”

“In total, we worked about two years to get the entire DSL network initially extended

and after that we did some extensions.”

A revolutionary upgrade like FttH, however, will take much more time, as illustrated by two interviewees…

“In case the local loop has too little capacity, one reaches the point that one must

implement fiber. These are projects with much longer durations. Digging in the entire

area of network coverage alone will take eight to ten years.”

“..In case one uses all building capacity available in the Netherlands, one will manage

to roll-out fiber in the entire Netherlands in minimally ten years.”

189

The examples of roll-out times of several technologies illustrate the dependency of roll-out time on the amount of technological adjustments, which is also demonstrated by another interviewee…

“The duration depends on the nature of the upgrade. In other words, whether one builds

further on something existing or one is going to roll-out something completely new,

something revolutionary.”

A final determinant for the roll-out of networks is scale, as indicated by the following interviewee…

“Duration strongly depends on the scale (replacing the core network or national roll-

out?) Rolling out the core network takes about half a year, rolling out the entire

footprint depends on the volume.”

Analysis

Strategic decision-making takes between six months and a year. Factors determining the duration of the decision-making process are mainly to be found in complexity of the object of decision-making, financial factors and the availability of a clear ‘sense of urgency’ to arrive at a decision. Roll-out times vary between several months to ten years and depend on the extension of the roll-out: systems, platforms or new technologies. Duration, moreover, depends on the amount of technological adjustments, in other words on the evolutionary or revolutionary character of the roll-out. The third determinant for the duration of the roll-out of networks is scale, in other words the difference between upgrades of the core network alone or a roll-out towards the entire footprint. Costs of networks are mainly driven by IT systems and the costs involved in digging, co-location, and adjustments to homes and equipment.

6.4.3 Cooperation in the value chain

One aspect of complexity, mentioned by De Baas (1998) is interdependency between actors in reaching their objectives. Due to this strong interdependency between market players, during the various phases of broadband roll-out, operators cooperate with several parties. With regard to cooperation within the value chain, two types can be distinguished: horizontal and vertical cooperation. Horizontal cooperation refers to cooperation with companies operating at the same level(s) of the value chain (see also the layer model in Chapter 3). Vertical cooperation is cooperation with companies operating on higher or lower levels of the value chain. One interviewee clearly indicates this distinction between horizontal and vertical cooperation…

190

“To come from network to services demands horizontal as well as vertical cooperation

takes place. Early in the process more horizontal cooperation takes place, concerning the

development of standards. Later on in the process, when it concerns roll-out of services,

cooperation becomes more vertically oriented.”

Cooperation in the process of decision-making occurs with market research companies but, even more importantly, with suppliers. These ‘technology partners’ provide information, which serves as important input for the analysis of alternatives. The most usual form is via requests for information (RFI), which leads to discussions with these parties on technology developments and options. But information also flows the other way around: informing suppliers about capacity forecasts and demands in the field of capacity and services. Besides in the decision-making phase, suppliers also play a role in the roll-out of networks. The important role of suppliers was mentioned by all operators we interviewed. The importance of this is underlined by…

“Suppliers form an important part of the decision-making process. What can several

suppliers offer? The choice in favour of a certain technology takes place in close

cooperation with suppliers. During the roll-out, suppliers are even more important.

Suppliers are usually in the middle of the roll-out. Sometimes, even a part of the roll-out

is outsourced to suppliers to be able to accelerate certain parts.”

In addition to suppliers, building contractors are mentioned as the most important partner for cooperation in the operational phase of roll-out. The strong interdependency between market players reveals itself even more clearly in the fact that cooperation does not always occur voluntarily, but is necessary due to the mutual dependency between different parties. As one interviewee illustrates…

“Depending on the subject, we work together with other companies, whether or not by

necessity. The incumbent is our biggest competitor, but I’m his biggest customer. That’s

very unusual. You must be able to arrange connectivity from everybody to everybody,

which forces you to cooperate with other parties. You are, ultimately, a network company

so you have to knot networks together.”

This mutual dependency between competitors is also reflected in the behaviour of the incumbent. Although the incumbent is forced by regulation to allow competitors to make use of its access network, wholesale customers form an important source of revenue for the incumbent. As a result, the incumbent considers the cooperation with its wholesale partners very important, as is illustrated by one interviewee…

“We closely work together with our wholesale customers. We try to take them with us in

our technological upgrades to find technological solutions for them. Roll-out takes place

191

in close consultation and coherence with wholesale. Wholesale is an important source of

income which makes that is it very important to accompany these parties towards our

new technologies, so they can remain customers afterwards.”

Finally, cooperation between competing cable and alternative operators takes place for a completely different reason, namely to join forces against an even bigger competitor: the incumbent. Interviewees from cable companies mention that they work together on issues concerning the cable branch, as evidenced by the following statement…

“The Dutch cable companies only have a nationally covering network together. For that

reason, we have to work together with regard to regulatory and sector-specific issues like

digital TV.”

Also, smaller, alternative operators join forces in certain situations to strengthen their standpoints against the power of the incumbent, as one of the interviewees mentioned…

“Sometimes, we act in conjunction with our small competitors towards the regulator to

look after the interests of the alternative operators”

Analysis

In the value chain, horizontal as well as vertically cooperation between market parties is visible. Horizontal cooperation is visible in the development of standards or in lobbies for collective interests. Vertical cooperation takes place mainly with suppliers and building companies during roll-out. This close, but not always voluntary, cooperation results from a strong mutual dependency between market players. This is emphasized by De Baas (1998), who mentions this mutual dependency as one of the characteristics of complexity (see Chapter 5). This close interdependency explains the earlier discussed extreme importance of competition as factor in the decision whether or not to upgrade networks.

6.4.4 Financial assessment methods

In Chapter 4, we provided insight into several financial assessment methods that can be applied by decision-makers to compare business cases of broadband roll-out projects. We introduced discounted cash flow methods (i.e. Net Present Value (NPV) and internal rate of return (IRR)), ratio methods (i.e. Return on Investment (ROI) and the benefit/cost ratio) and Real Options Analysis (ROA). In our interviews, we asked decision-makers which methods they usually apply to assess business cases.

192

The importance of the NPV in financial decisions was reflected throughout the interviews and is, moreover, underlined by this interviewee…

“NPV plays an important role in investments decisions of all companies...”

Two interviewees also mention ROI. One of them says…

“Revenues, costs and investments are made visible with NPV calculation. We also

calculate the Return on Investment (ROI)”

The NPV method, in combination with scenario analysis, is used to compare business cases, which is specifically mentioned by two interviewees…

“Business cases are compared via scenario analyses. Scenarios are compared based on

NPV equations.”

“NPV and scenario analyses have also been applied to make the consideration between

fiber and other technologies and between new technologies for voice services.”

Another method that is applied to build business cases mentioned by interviewees is cost-benefit analysis and calculation of cash flow effects…

“Before a technology can be implemented business cases are built, cost-benefit analyses

are made, CAPEX and OPEX are calculated, etc. “

Real Options Analysis (see Chapter 4) is not being applied by decision-makers, because most of them are not familiar with this method, as is clearly illustrated by the following interviewee…

“Real Options Analysis plays no role whatsoever. Because A) I’m not familiar with the

method and B) that’s why it is not being applied in our company.”

Cost-related drivers As a next step, we asked the interviewees what they consider the most important cost-related drivers for network roll-out. Cleary, IT and physical costs are the most important. Three interviewees mention IT as an extremely important cost-related driver, which is substantiated by…

“IT is an often mistaken, major source of costs, which is quickly forgotten...”

“Adjustments to existing IT systems cost a lot of money and effort. On the other hand,

completely new IT systems also cost a lot of money and time and they must be integrated

with the old ones.”

193

The second, major source of costs is physical costs, which includes digging, co-location, adjustments to homes and equipment, as illustrated by the following…

“Digging is a really big cost-related driver “

“Costs for equipment is a very important one and the costs for co-location in the case of

DSL”

“…Physically digging and adjustments to homes and offices demand many resources.

The latter because of the enormous amount of houses...”

Analysis

We see that several assessment methods are applied to choose between different technological options. In all cases, the costs and benefits are made visible. Whether in combination with other methods, like ROI or scenario analysis, all interviewees mention the NPV as an important assessment method. This has been demonstrated earlier by Keswani & Schackleton (2004), who state that the Net Present Value rule has emerged as the dominant decision rule for investment-related decisions. From the methods introduced in Chapter 4, we see the methods of NPV, ROI and cost/benefit ratio being applied in practice. The method of IRR was not mentioned by operators. In Chapter 4, we already explained that traditional discounted cash flow methods are less suitable for environments characterised by uncertainty and a need for flexibility (Neufville, 2001; Keswani & Schackleton, 2004) and that the method of ROA is more suitable for assessing investment options in uncertain environments like communication and information technology (Oslington, 2004; Neufville, 2003). For this reason, it is striking that none of the interviewees, with the exception of one, were familiar with the method of Real Options Analysis and that the method is not applied for financial assessments by any of the decision-makers we interviewed. This supports the observation from Alleman (2002) that, although relevant to telecommunications. Real Options Analysis is not widely applied within this industry. The main cost-related drivers for broadband projects are IT and physical costs. Other cost-related drivers mentioned by interviewees as being important are costs for marketing and logistics and project management.

6.4.5 Evolutionary or revolutionary upgrades

In choosing a technology to upgrade their networks, operators can opt in favour of an evolutionary (stepwise) or revolutionary (rigorous network changes) approach (see Chapter 2). Cable providers clearly prefer the evolutionary path, as becomes

194

clear in the following two fragments involving interviewees working for cable operators…

“At the moment, Docsis offers a suitable evolutionary path, which can handle the growth

well. This is also the path we’re following at this moment. On the other hand, you see the

urgency coming up to further upgrade the network. Where this urgency will originate and

when investments can and must be done is not known yet, but one must be prepared for

the future.”

“Most end-user demand lies around 4 Mbps download speed, although demand for more

upload speed is visible. The demand for bandwidth will increase in the coming years, but

we can grow along this demand. We follow an evolutionary approach in this. (…) We

decided to grow along the demand curve. In case a critical mass for a certain amount of

bandwidth occurs, we upgrade the network to this bandwidth. Revolution is (like in

history) a bloody process. We prefer evolution above revolution.”

Also, PSTN operators opt in favour of an evolutionary approach…

As long as your market shares for the future aren’t certain, the roll-out of fiber cannot

yield a profit and you must do something else. As alternatives there are ADSL2+ and

VDSL, which are, moreover, continuously developing. As a result, it is a constant process

of deliberation when to set the next step. It is a continuous balance: Am I on time? How

much time will it take to roll-out a technology? When do end-user’s needs start to further

develop themselves?”

Another possibility is that an operator chooses different strategies for different network layers. As one interviewee illustrates…

“(The development towards our All IP network is) a revolutionary development

(compared to other telecom operators). Not just in the realisation of this new technology,

but also in the (partially forced) rigorously leaving the old network technology.”

Another interviewee from the same company…

“At the physical network layer a more evolutionary development (ADSL2+ - VDSL –

FttH) is taking place. The road via ADSL 2+ and VDSL is chosen because there clearly

is no demand yet from consumers for more network capacity than these technologies can

offer.”

The FttH issue Although, generally speaking, FttH is considered the most future-proof technology at the moment, operators in the current market are not rolling out this technology toward their residential end-users. The main problem, as

195

addressed by the interviewees, is the high number of necessary investments and the number of households needed to recover these costs.

“There is no place for more parties to compete on infrastructure besides the existing

infrastructures, meaning the copper and cable infrastructures competing with fiber. One

speaks of an investment of around € 1000 per household in case one wants to bring fiber

into these homes. Calculate that for the Netherlands…”

“The Netherlands can most likely not handle having two fiber infrastructures to the

residential homes. It involves an amount of € 6 billion to roll-out FttH on a national

scale. Two times makes 12 billion. This amount must be paid by 16 million citizens.

Instinctively this cannot be right. 3 billion (costs of FttC) divided by 6 million households

can also not be earned back. That’s the reason almost nobody is moving in the market.

How can one get a positive business case out of that? ‘

Several drawbacks limit FttH roll-out:

“The roll-out of fiber to the home takes years. To realise this within a limited amount of

time is practically impossible. There aren’t enough contractors for this within the

Netherlands and there’s not enough optical fiber cable available. Finally, there’s also no

demand for it.”

A second important problem is that, in most cases, social benefits do not end up where the costs are incurred (Expertgroep Breedband, 2002). This is a problem of the chicken and the egg: development of useful content does not start as long as the infrastructure (high-speed broadband networks) is not available and suppliers do not invest in infrastructure in case there’s no clear view on useful content. We clearly see this problem reflected:

“… One must know beforehand, i.e. at the moment one can justify the investments in fiber

to the home, which service providers will make use of the infrastructure. Otherwise one

will never get profit.”

Analysis

We can conclude that, at the moment, Dutch cable and PSTN operators favour evolutionary network upgrades. Although several operators choose different strategies concerning the use of standardised technologies, cable operators as well as PSTN operators agree they prefer evolutionary upgrades above a revolutionary path for their physical networks, sometimes combined with a more revolutionary approach on the higher protocol layer. Critical mass theory is clearly applied here (see also Chapter 2). Networks are upgraded at the moment a critical mass for a certain bandwidth is visible and operators follow an ‘invest-as-you-grow’ strategy. The first dominant cause mentioned for this preference for an evolutionary

196

strategy has to do with the fact that there clearly is no demand yet from consumers for more network capacity than the existing technologies can offer and that the existing networks are capable of handling the expected demand growth in the foreseeable future. The second reason is of a financial nature: investments are too high and the expected revenues too uncertain to build a positive business case. The number of households needed to recover the huge upfront investments leaves only space for one national operator. A third reason is related to the chicken-egg problem: when use of the infrastructure by service providers is not guaranteed beforehand, no upgrades to fiber will take place due to uncertain revenues. As a result of these problems, market parties do not opt in favour of a revolutionary approach but keep following an evolutionary approach. During our interviews, we asked operators which paths they are currently following. In the figures presented the possible migration paths from the existing networks towards GPON networks are presented (see paragraph 2.3.6 for more details). In these figures, the migration paths Dutch network operators are following so far are drawn.

ADSL 2+ADSL

VDSL

APON

EPON

BPON

GPON

Active Optic Ethernet

Etherloop

EFMC

EPON

BPON

GPON

APON

EPON

BPONGPON

Active Optic Ethernet

BPON

GPON

GPON

BPONGPON

VDSLAPON

EPON

BPON

GPON

Active optic Ethernet

EPON

BPON

GPON

Active Optic Ethernet

EPON

BPON

GPON

Active Optic Ethernet

Figure 6.2 Applied evolutionary path Telecom operators in the Netherlands

197

HFC

EttH

BPON

GPON

BPON

GPON

Narad

Docsis3.0

GPON

Docsis 2.0

APON

EPON

Active Optic Ethernet

BPON

GPONBPON

GPONBPON

Figure 6.3 Applied evolutionary paths by cable operators in the Netherlands

6.5 Applying decision-making theory to broadband local

loop upgrades

Where the previous paragraphs focused on several factors affecting the decision-making process, the coming paragraphs will focus on the course of the decision-making process itself. In Chapter 5, we introduced two decision-making theories, Garbage Can and Logical Incrementalism, which applicability will be tested in the remaining part of this chapter. This paragraph will first focus on the validation of the Garbage Can theory by Cohen, March and Olsen (1972). This validation will take place in two steps. Firstly, it will be tested whether the three criteria for organisational anarchies, being problematic goals, unclear technology and fluid participation are applicable to telecommunication companies. In case we find enough evidence in our interviews that telecom companies are organised hierarchies, it can be assumed that the Garbage Can theory can be applied. As a next step, we will try to prove, based on interview results, that the four streams of the Garbage Can, being problems, solutions and participants can be recognised and flow together in the so-called choice opportunities. As a next step, this paragraph will focus on the validation of the theory of Logical Incrementalism of Quinn (1980).

198

6.5.1 The Garbage Can: criteria for organisational anarchies

As discussed in Chapter 5, the Garbage Can theory by Cohen, March and Olsen (1972) can only be applied to organisational anarchies. Cohen et al. drew up three criteria for organisational hierarchies. Below, we discuss to what extent these criteria apply to telecom companies. Problematic goals

The first criterion for organisational anarchies is problematic goals, meaning that they may be uncertain about the nature of the problems they face and about what they hope to accomplish. A variety of inconsistent and ill-defined preferences is recognisable within organisations. In telecom companies, several conflicting goals can clearly be recognised and are echoed throughout the interviews. These inconstant preferences can be divided into three main conflict areas. Firstly, there are conflicts between technological and commercial preferences. Two examples of clear illustrations of this conflict of goals…

“The departments of sales and business development often want a quick and dirty

solution, whereas the technology department knows that such simple designs aren’t

future-proof. Problems have occurred in the sense that quick and cheap solutions

appeared to be poorly scalable, as a result of which the year after the introduction

something new had to be implemented again.”

“There is an inherent conflicting interest between Commerce/Finance and Technology.

Commercial people hold the opinion solutions must cost less and must evolve quicker

and investments must be postponed. This makes their lives easier. Technological people

want to spend more money and want more time, that’s what makes their lives easier…”

There is a second type of conflicting goals between business and residential (consumer) departments visible, as evidenced by the following interviewees…

“Different interests between (departments of) business and consumer market exist: other

needs and other time paths. A consumer service can cannibalise a business service just

by that or the other way around.”

“Differences are mainly visible between business and consumer departments. One can

imagine that the business department prefers MDF locations in business districts where

the consumer department prefers them in residential areas.”

The third conflict has to do with diverging interests between the short and long-term strategy of the operation. One interviewee illustrates this conflict…

199

“Overall objective is profit for the company, but the vision how to reach this goal is not

always the same for different people. Marketing & Sales are led by daily impressions.

They look at what they see happening today. The strategy department looks at

worldwide developments with a time horizon of ten years. These departments have

different opinions on what is necessary and good for the company.”

As another interviewee explains, the financial departments also work with short time horizons, which can lead to problems with long-term strategy…

“..These issues arise in the second line of decision-making, the financial line. Here it is

where is must be decided whether financial resources really will be released. (…) One

sees on a regular basis that tension arises when decisions are made based on another

interest (namely financial incentives), which sometimes absolutely does not take account

of long term objectives. (…) Finances are managed differently than the business plan

they are a derivative from. Plans which originally were in line with the strategy of the

board are ultimately implemented differently because the financial resources aren’t

allocated by the financial department.”

Another interviewee mentions that another source of conflict lies in the differences in the fact that the goals of separate departments, which increasingly become independently operating units, conflict with the overall strategy of the company.

Unclear technology

The second criterion for organisational anarchies is unclear technology. This refers to the fact that members within the organisation do not fully understand how their organisation works and are uncertain of the rules, structures and processes by which decisions are made (see also Chapter 5). Throughout our interviews, it became clear that decision-making processes concerning strategic issues like network upgrades and technology choices are far from structured and transparent. A first vagueness has to do with the participants being involved in decision-making processes and how they come together, as illustrated by the following interviewee…

“A decision for network upgrades, say strategic decisions, occur only a few times a

year. As a result, they are not taking place in well-defined structures. On a case-by-case

basis small groups are formed for these processes.”

A second source of unclear processes has to do with the fact that these kinds of decisions are not made on a well-defined rational basis and for that reason are partly vague. Several interviewees underline this non-rational element in decision-making…

“In the end, decision-making is a very emotional process for decision-makers.”

200

“It can be described better as an ensemble of emotions, politics and ratio. Not

everything is deliberated rationally.”

“Also, preferences of people play an important role. These kinds of major decisions

bring about emotions. Because one can’t predict demand for technology so far ahead

this brings ahead beliefs and emotions about this demand. No one can substantiate it,

which causes emotional and political tensions between several interested parties.”

Other sources for problematic goals mentioned by interviewees are the unstructured course of the process as a result of high complexity and diverging interests, which result in non-rational decision-making. In these cases, decisions are more the result of horse trading. In contrast with what other elements of a company and the outside world may think, the objective outcomes of business cases are not by definition leading in decision-making. This is moreover evidenced by the following interviewee…

“A business case based on as many objective numbers as possible does not

automatically result in the best decision. The basic component is, ultimately,

entrepreneurship, and entrepreneurship has a lot to do with feeling and vision. What

that signifies is difficult to pinpoint.”

Fluid participation

The last criterion for organisational anarchies is that of fluid participation, consisting of the involvement of different actors in different situations. As one interviewee mentions…

“On a case-by-case basis small groups are formed for these (strategic decision-making)

processes.”

In several phases of the roll-out process, many different parties are brought in. It can be assumed that their constellation is not the same in each situation, depending on the decision-making object. The large number of parties involved in decision-making is brought up by several interviewees. Two of them…

“In the phase of decision-making all kinds of parties are involved to provide

information, like suppliers, technology designers, research firms, etc. During the

upgrade, operational parties are involved like contractors. At a certain moment, the

marketing department is brought in for service promotion.”

“The moment somebody has an idea or a new technology emerges, information on this

technology is gathered by several sources among suppliers. RFIs are sent out at many

different sides.”

201

Participants also change as a result of changing functions…

“(Decision-making processes), moreover, change over time, which is connected with

(…) people taking decisions. They change when people change positions.”

The number of departments involved in the analysis of technological options is not always equal, but depends on the magnitude of the technological renewal…

“In performing the analyses for extensive technological renewal at least five

departments are involved: strategy department, also including technologically informed

people, Marketing & Sales, Business Development, Network Development and Business

Management. In case of small renewals the network development department takes the

decisions on its own.”

Another aspect, mentioned by Cohen et al. (1972), namely the fact that participants vary in the amount of time they devote to the decision-making process, is supported by the following interviewee…

“A product manager will fight for his product longer than someone more financially

oriented. An entire process between technological and financial people takes place and

finally the ‘big boss’.”

Analysis

In Chapter 5, we concluded that two decision-making theories could be applied to broadband-related decision-making: the Garbage Can theory and the theory of Logical Incrementalism. The Garbage Can theory by Cohen, March and Olsen (1972) can only be applied to organisational anarchies. Cohen et al. drew up three criteria for organisational hierarchies: problematic goals, unclear technology and fluid participation. Based on the fact that our analysis has proven that these three characteristics for organised anarchies can be applied to telecom companies, we can conclude that telecom companies can be qualified as organised anarchies as defined by Cohen et al.

6.5.2 The Garbage Can; applicability of streams and decision windows

After establishing that telecom companies can be seen as organisational anarchies, we now try to demonstrate that strategic decision-making within telecom companies evolves according to the theory of the Garbage Can. This theory by Cohen et al. (1972) distinguishes three, relative independent, streams: problems, solutions and participants and states that ‘choice opportunities’ are the result of unique and coincidental combinations of these streams of problems and solutions at a given time and place (see Chapter 5 for more details).

202

The stream of participants refers to the fact that problems and solutions change with participants coming and leaving the process. Problems are points of dissatisfaction, which may or may not lead to a solution. Solutions can exist independently of problems and the attractiveness of a solution can also stimulate the search for a problem. Searching for a problem when a technological solution is available is quite common in this market and better known as ‘technology push’. This independence between problems and solutions and technology push is illustrated by the following interviewee…

“The DSL technology was already ready (independent solution stream). However, there

must be enough market demand for roll-out (independent problem stream). But is this case

I think it went the other way around: that market demand was created by offering cheap

bandwidths.”

We see this connection of problem and solution coming together in a choice opportunity. The same interviewee continues…

“…then there are people spotting opportunities and at that moment it becomes opportune

to spend money and then a business case must be made. At that moment the entire

decision-making process starts to run…”

The relative independence of the streams of solutions and problems is clearly recognisable throughout the interviews. The following interviewee clearly expresses the independence of a solution stream and the fact that several ‘stimuli’ suddenly cause choice opportunities…

“…what is often seen is that these things are already simmering for a long time: suppliers

come to tell what kind of interesting developments are available, market research firms tell

stories, presentations are given at conferences and that simmers for a long time and

suddenly several stimuli make that at one moment it suddenly becomes urgent: which way

are we going? What must be done? After that, within a few months it is done...”

The next question, regarding which stimuli cause this sense of urgency with regard to decision-making, the interviewee continues by mentioning the fear of leaving the market to competitors…

“…people from the business say: this is a really important development! They see what

our competitors are doing, what our peers are doing and what happens at seminars. Then

the moment comes: now we also must do something. At that moment a decision-making

process accelerates and a business case must be performed very fast, all pros and cons

and market demand must be lined up and then decision-making evolves quite fast.”

203

In the interviews, many other examples were given of problems and solutions coming together and result in so-called ‘choice opportunities’. On the basis of interview fragments we will illustrate these moments. In each fragment we will indicate which problem and solution are coming together. A very clear choice opportunity, created by two problems in the field of regulation and competition and a sudden technological solution is provided by the following interviewee…

“The road towards VDSL was initially closed off because we were forced by the regulator

to roll-out ADSL2+ to the wholesale market and our competitor had bought the TV rights

(problems in problem stream). But suddenly our suppliers indicated that ADSL2+ and

VDSL could function together so implementing VDSL would not close off the way towards

VDSL (solution in solution stream). This sudden solution reopened the road towards VDSL

and made us decide to implement ADSL2+ and VDSL as a later option (choice

opportunity). Due to technological developments of the behaviour of competitors the world

can look completely different after 2 months…”

Another interviewee mentions that a coincidental combination of a problem and solution resulted in the following choice opportunity concerning the roll-out of digital TV…

“Large scale roll-out of digital TV of our company in the US significantly dropped the

prices of our set-top boxes. Our company buys these boxes in enormous amounts with

the result we could offer them for free in the Netherlands, which causes an important

competitive advantage (solution). At the same moment, we saw competition rising

caused by the roll-out of ADSL2+ by our competitor (problem). The combination of low

set-top box prices and competitive threat by ADSL2+ has lead to the decision to also

roll-out digital TV in the Netherlands.”

In the next example, an interviewee illustrates the coincidental connection of suddenly available solution for the problem of high modem prices of VOIP, blocking service roll-out of an already available service…

“VOIP came up very quickly and is put in the market very fast. Sometimes a service

already exists latently (independently available in the solution stream), but then it

suddenly becomes urgent to introduce it in the market (choice opportunity). In the US

the service became successful very fast among cable companies with the result that the

costs of modems dropped significantly (sudden solution) with the result that the service

suddenly became available to a broad public. Then you introduce the service very fast.”

The final statement probably characterises the garbage can of decision-making in broadband roll-out best …

204

“Such a decision-making process has absolutely not evolved in a rational way, but is

tied together by strokes of luck: certain people were present at certain moments. In case

other people would have been there other decisions may have been taken.”

Analysis

Our results have demonstrated that strategic decision-making with regard to network upgrades and roll-out is not a structured, rational process, but rather can be characterised by connections of available and suddenly arising problems and solutions in combination with people being available at a certain time and place. The decision-making processes concerning broadband roll-out for that reason fits the characteristics of the Garbage Can theory by Cohen, March and Olsen (1972).

6.5.3 Logical Incrementalism: decisions are made in small steps to keep

options open for feedback and adjustments

In Chapter 5, we assumed that, in addition to the Garbage Can theory, the theory of Logical Incrementalism could also be applicable to broadband-related decision-making. This assumption was based on characteristics of the broadband market and the strong parallels with Real Option Analysis, the applicability of which with regard to the broadband market has already been proven. The assumption refers to three characteristics of Logical Incrementalism, as formulated by Das & Teng (1999): 1) incremental nature, 2) broad and relatively vague objectives, and 3) options for developments and adjustments. In this paragraph, we examine all three characteristics on the basis of our interview results. Incremental nature

The incremental nature manifests itself in stepwise decision-making, with managers not making dramatic decisions (see also Chapter 5). The following fragment evidences the fact that small steps together lead to the main company objective…

“…sometimes there are several steps in between which lead together to the final

objective. After all, many small steps also make one big step.”

The clear preference for incremental decision paths above dramatic decisions is clearly evidenced by these interviewees …

“Considering the large amounts of money one cannot do else than taking decisions

stepwise. The thing one does not want is taking at this moment a very major decision for

the coming ten years. In case one looks at it separately this would be much less

complicated (clear path). However, in this situation of huge investments and large risks

205

and uncertainties, one wants to take the smallest possible steps, to keep all options open

as long as possible. This has to do with the fact that the future is unpredictable, fast

developing technologies and shareholders. Concerning the uncertain market

developments in ten years it is irresponsible to invest 2 or 3 billion at this moment.”

“In case of such large investments one cannot decide to do one thing at this moment.

One does can have the plan and take the first step, but after step 1 it will be

reconsidered again whether step 2 will also be taken or a digress from the initial route

is made. Due to these large amounts one cannot do else than stepwise decision-

making.”

Several other interviewees identify the large size of the investments in combination with market uncertainty as the underlying cause of the need for incremental decision-making. Strategic, broad and relatively vague objectives

The decision-making process is a consistent movement aimed at reaching a clear objective. The strategic objectives are, however, formulated broadly and relatively vaguely. The following overall company objectives were mentioned in our interviews and demonstrate the clearness of the final objectives, but also their broad and vague character…

“Strategic/tactical reasons (are), for example, not losing the battle with competitors,

preservation of market share on the mid-term or to secure certain products.”

“…other factors comprise the market and the position as a company you want fulfil in

this market: market leader in quality or in low prices or in access speed?”

Or, at an even higher company level…

“In the end, everybody shares the same interest: profit for the company.”

As a result of this vagueness, people may have different ideas of how the overall objective should be reached, as mentioned by one interviewee…

“The objective of course is profit for the company, but the vision on how to reach this

objective is not the same for everyone.”

Options for developments and adjustments

The reason for these relatively vague and broad objectives is to keep options with regard to developments and adjustments open in case more information becomes

206

available. The following interviewee clearly explains this relationship between incremental decision-making and keeping options open for the future and again we see the relationship with large investments…

“Because one has to decide about such large amounts which take a long time before

they are earned back and because one has to look so far into the future at what demand

will be and what one eventually will earn back, one preferably wants to keep decisions

as small as possible so that every time many options are kept open.”

According to Das & Teng (1999) the incremental approach provides organisations with the flexibility to consider emerging alternatives and future options to adjust processes to changes in the environment. This becomes clear in the following two fragments …

“…then one rolls out a network and after that it depends on the number of customers

and market expectations whether capacity is being increased or not. So, in case there

are enough customers or one expects enough demand capacity is increased at existing

locations or economically less viable locations are also connected because as a result of

increasing demand they are now becoming attractive.”

“…within the roll-out of a service also a delay can take place. This is caused by the

actual adoption. In case this is disappointing less of this service will be rolled out. Our

network is ready for 50 Mbps, but there’s already very little demand for our 20 Mbps

service so there will be a delay in the 50 Mb service. It’s very well possible that as a

result of a technological innovation the 50 Mb service is becoming much cheaper with

the result that it will be offered to the market all of a sudden.”

According to Das & Teng (1999), managers constantly develop and evaluate options, based on feedback from previous actions. The following interviewees emphasise this role of learning moments, feedback and adjustments to existing processes…

“In case implementation following on decisions does not turn out well, a new round of

decision-making follows to see where things are not working out well and why. As a

next step feedback takes place and, where possible, process adjustments are performed.

It is a matter of continuously monitoring, feedback and adjustment.”

“It’s an iterative process. In case the decision to roll-out is not taken, there is a delay

but influenced by, for example, a technological innovation a return to the decision takes

place. Everything is reconsidered all over again. Now it has been decided to roll-out

ADSL2+ everything will still be deliberated for the decision on VDSL. This way one

gains more and more understanding of all components. One gathers learning moments

which one applies further on in the process. As a result, risks are decreasing and, due to

more insights, certainty about demand is increasing.”

207

Analysis

In Chapter 5, we assumed that, in addition to the Garbage Can theory, the theory of Logical Incrementalism (Quinn, 1980) would also be applicable and that strategic decision-making with regard to broadband network roll-out is incremental in nature. Based on Quinn, we assumed that decision-making in broadband is a consistent movement aimed at reaching strategic objectives that are broad and relatively vague to keep the option with regard to future developments and adjustments open in case more information becomes available. Das & Teng (1999) formulated three characteristics of the Logical Incrementalism theory; 1) incremental nature, 2) broad and relatively vague objectives, and 3) options for developments and adjustments. Throughout our interviews various examples were provided that ground the three characteristics of incrementalism formulated by Das & Teng (1999).

6.6 Concept validation as a first step in model validation

In this paragraph, we discuss the results of our qualitative validation of the variables in the conceptual model developed in Chapter 3 (see Figure 3.6). This validation has taken place based on concept mapping. First the four phase basis model will be dealt with. Next, for every phase, the additions to the conceptual model, based on the concept maps will be discussed.

6.6.1 The four-phase base model

As a first step this paragraph discusses the basic four main phases as can be seen in the middle of Figure 3.6. The concept map shows that the four phases, which form the core of the conceptual model, are affecting each other in the way the causal model shows. All interviewees indicated that they recognised this high level basic model. The concept map shows, additional to this, two feedback loops that are not mentioned in the initial conceptual model. These are a feedback loop from end-user demand and adoption phase towards the decision-making phase and a feedback loop from revenues to the decision-making phase. These loops are easy to explain: in case end-user demand exceeds expectations, a decision must be made about further upgrading the network or in case demand is disappointing, maybe roll-out must be delayed. The same goes for revenues. Disappointing, as well as revenues above expectation, lead to reconsideration of the original roll out strategy. Disappointing revenues are an immediate cause for delaying the planned roll-out or entering a new round of the decision-making process to upgrade the network. This can be clearly seen Figure 6.4. Like in the conceptual model, the variable network upgrade delay is not connected with phase 1 and 2, but with phase 3 and 4. Also, disappointing end-user demand will lead to a delay of the

208

network roll-out. We also see the relationship between end-user demand and adoption, and the acceleration of roll-out. This is a positive relationship: in case demand and adoption exceed expectations, roll-out will be accelerated.

is cause of

Has impact on

Has impact on

Has impact on

is cause of

Has impact on

Has impact on

is cause of

Has impact on

Decision making process to upgrade/ rollout broadband {51-63}

Network upgrade delay {11-11}

End user demand & adoption {58-34}

Revenues {12-12}

Network upgrade/ roll out {39-37}

Acceleration roll out {8-6}

Figure 6.4: concept map of 4 phase model and network upgrade delay

The numbers in brackets to the right of the concepts in Figure 6.4 indicate the number of times the concept is mentioned and the number or relationships it has with other concepts. For example, the concept ‘network upgrade/roll-out’ is mentioned 35 times by interviewees and has 34 relationships (ingoing and outgoing) with other concepts. The concepts of figure 2 and their linking phrases are presented in Table 6.1.

Table 6.1: Relationships between the 4 main phases

Source concept

Relation

(linking phrase) Target concept

DM process to upgrade/roll-out broadband

Has impact on Network upgrade & roll-out

Network upgrade & roll-out Has impact on End-user demand & adoption End-user demand & adoption

Has strong impact on DM process to upgrade/roll-out broadband (feedback loop)

End-user demand & adoption

Has impact on Revenues

End-user demand &

adoption

Is cause of Broadband roll-out delay

Revenue (phase) Has impact on DM process to upgrade/roll-out broadband (feedback loop)

Revenue (phase) Has impact on Network upgrade & roll-out (feedback loop)

As a next step, we compare the concept map to the original conceptual model to validate the latter. Figure 6.5 shows the result of the adjustments to the original conceptual model (Figure 3.6), based on the concept map presented in Figure 6.4.

209

The adjusted model has three feedback loops that did not exist in the original model.

Decision Making process to upgrade/roll out broadband

Network Upgrade/roll out

End user demand & adoption

Revenues

Has impact on

Has impact on

Has strong impact on

Has impact on Has impact on

Has impact on

Figure 6.5: Adjusted base of the conceptual model (4 phases)

Analysis

It can be seen in Figure 6.5 that the feedback (results) from later phases also serves as input for earlier phases. It becomes clear that decisions can be adjusted or reconsidered based on end-user demand & adoption, and based on revenues. In case end-user demand or revenues turn out to be higher than expected, the decision can be made to accelerate the roll-out of the network. Also, in case of higher revenues, more money is available for network investments. On the other hand, in case end-user demand is disappointing, it can be decided to (temporarily) defer the roll-out. This is illustrated by the feedback loop from revenues to network upgrade/roll-out. The same applies in situations where revenues are disappointing. In such situations, it is, moreover, likely that some projects that are more vital to the business are given a priority over others, which will therefore be delayed. We can conclude from this that the decision to roll-out or upgrade a network is not a one-time decision, but one that appears to evolve in steps, as conditions in the environment are changing. This clearly supports our assumption that decision-making with regard to broadband roll-out is an incremental process in the way described by Quinn (1980) (see also Chapter 5 and paragraph 6.5.3). In the next part of this paragraph, we discuss the 4 phases and their related concepts in greater detail.

6.6.2 Phase 1: Decision-making process to upgrade/roll-out broadband

The first phase of the upgrade- and roll-out process of broadband networks is the decision-making process to upgrade or roll-out a broadband network. The concept map of phase 1 of the conceptual model (the decision of broadband roll-out) and all related variables show a non-transparent network, caused by many and highly interrelated concepts. In the concept map of phase 1, five variables are direct causes of the decision-making process; 20 variables have a strong impact on the process (and are mentioned over 5 times). These concepts and their linking phrases are shown in Table 6.2. The variables can be divided into five categories.

210

Firstly, some more general factors on a higher level of abstraction. Next, we see factors related to technology, finance, market and internal organisation.

Table 6.2: Source and target concepts and relationships phase 1

Source concept Relationship

(linking phrase) Target concept

Incoming variables General factors

Sense of urgency (6) Is cause of DM process to upgrade/roll-out bb Driver for network upgrade (6) Is cause of DM process to upgrade/roll-out bb Uncertainty (7) Has strong impact on DM process to upgrade/roll-out bb Technological factors

Technological alternatives (23) Is cause of DM process to upgrade/roll-out bb Technological developments (36) Is cause of DM process to upgrade/roll-out bb Bandwidth development (6) Has strong impact on DM process to upgrade/roll-out bb Future-proofness (12) Has strong impact on DM process to upgrade/roll-out bb Scalability (9) Has strong impact on DM process to upgrade/roll-out bb Financial factors

Business case (35) Has strong impact on DM process to upgrade/roll-out bb Cost rationalisation (9) Has strong impact on DM process to upgrade/roll-out bb Financial analysis (11) Has strong impact on DM process to upgrade/roll-out bb Financial resources (7) Has strong impact on DM process to upgrade/roll-out bb Revenues (12) Has impact on DM process to upgrade/roll-out bb Investments (24) Has impact on DM process to upgrade/roll-out bb Market-related factors

Competition (46) Has strong impact on DM process to upgrade/roll-out bb Demand forecasting (7) Has strong impact on DM process to upgrade/roll-out bb End-user demand and adoption (58) Has strong impact on DM process to upgrade/roll-out bb Expected end-user demand (developments) (13)

Has strong impact on DM process to upgrade/roll-out bb

Market developments (25) Has strong impact on DM process to upgrade/roll-out bb Regulatory issues (25) Has strong impact on DM process to upgrade/roll-out bb Service development (18) Has strong impact on DM process to upgrade/roll-out bb Internal organisation factors

Business strategy (27) Is cause of DM process to upgrade/roll-out bb Decision holding (7) Has strong impact on DM process to upgrade/roll-out bb Individual actor behaviour (7) Has strong impact on DM process to upgrade/roll-out bb Long-term vision (5) Has strong impact on DM process to upgrade/roll-out bb Vision, sense and entrepreneurship (10)

Has strong impact on DM process to upgrade/roll-out bb

Finance & control department (11) Has impact on DM process to upgrade/roll-out bb Management (13) Has impact on DM process to upgrade/roll-out bb

Outgoing variables DM process to upgrade/roll-out bb Has impact on Network upgrade/roll-out (35) DM process to upgrade/roll-out bb Is cause of Timing of upgrade/roll-out (8)

Figure 6.6 shows the concept map. It is clear that there are many concepts and that all concepts are very closely interrelated, which confirms the high level of complexity in broadband-related decision-making we discussed in Chapters 2 and 3. Besides many factors, four types of actors were also mentioned as being of the

211

utmost importance in the decision-making process: general management, the finance and control department, shareholders and suppliers. Moreover, Table 6.2 shows that five variables have a causal relationship with the decision-making process to upgrade or roll-out. Because this is the strongest relationship, these variables have the highest impact on decision-making. Two variables are of specific interest because they can provide some additional insights at a lower level. These variables are ‘sense of urgency’ and ‘drivers for network upgrade’. They are defined at a higher level of abstraction and for that reason it is interesting to examine them one level deeper. Looking into the factors affecting these two variables can show us which factors are most important in driving network roll-outs and which factors cause an immediate sense of urgency with regard to the decision to upgrade. For these reasons, we take a closer look at these two variables.

212

is associated with

Has strong impact on

is part of

Has strong impact on

is cause of

Has impact on

is associated with

is ais cause of

Has impact on

Has strong impact on

Has impact on

is part of

is cause of

Has impact on

is associated with

Has strong impact on

Has impact on

Has strong impact on

Has strong impact on

Has strong impact onis cause of

Has impact on

is part of

Has impact on

Has impact on

Has impact on

Has strong impact on

is part of

is associated with

is associated with

Mutal influence

is part of

Has strong impact on

Has strong impact on

contradicts

Has strong impact on

is part of

Has strong impact on

is associated with

is part of

Has strong impact on

Has strong impact on

Has strong impact on

Has impact on

Has strong impact on

is cause of

is cause of

is a

Has impact on

Has impact on

is associated with

Has strong impact on

Has strong impact on

Has impact on

Has impact on

Has strong impact on

Has impact on

is cause of

Has impact on

is a

is part of

Has strong impact on

is associated with

is associated with

Has impact on

is associated with

Has strong impact on

Has impact on

is a

Has impact on

Has strong impact on

Has strong impact on

is cause of

is a

is part of

is cause of

is a

is cause of

is cause of

is a

is associated with

is part of

Has impact on

Decision making process to upgrade/ roll

out broadband {51-63}

Service development {18-12}

Expected end user demand

(developments) {13-14}

technological developments {36-27}

future proofness {12-8}

Business case {35-38}

Market developments {25-20}

End user demand & adoption {58-34}

Financial analysis {11-9}

Competition {46-35}

Business strategy {27-24}

Regulatory issues {25-18}

Vision, sense & entrepeneurship {10-5}

Investments {24-23}

Finance & Control department {11-8}

Management {13-5}

Revenues {12-12}

Sense of urgency {6-9}

Driver for network upgrade {6-12}

Bandwidth development {6-6}

Cost rationalization {9-6}

Decision holding {7-5}

Demand forecasting {7-11}

Financial resources {7-6}

Individual actor behaviour {7-7}

scalability {9-6}

Uncertainty {7-18}

Timing of upgrade/roll out {8-9}

Fig

ure

6.6

: C

oncept

map p

hase

1-

decis

ion-m

akin

g p

rocess

to u

pgra

de n

etw

ork

(in

com

ing a

nd o

utg

oin

g c

au

sing a

nd (

stro

ngly

) in

fluenci

ng

vari

able

s)

213

Drivers for network upgrade

Although the concept itself is only mentioned six times, it provides clear insight into other, denser concepts which are specifically mentioned as being drivers for network upgrades. Figure 6.7 shows the factors that are drivers for network upgrade.

is a

is a

is associated with

Has strong impact on

is associated with

Has impact on

Has impact on

is a

is cause of

Has impact on

is a

is a

is a

is a

is a

is a

Revenues {12-12}

technological developments {36-27}

Costs {22-19}

End user demand & adoption {58-34}

Competition {46-35}

scalability {9-6}

Cost rationalization {9-6}

Service development {18-12}

Driver for network upgrade {6-12}

Technological network boundaries {6-3}

Figure 6.7: Concept map for “Drivers for network upgrade”

Nine drivers can clearly be distinguished. At a higher level of clustering, we can divide these nine variables into four categories: market, finance, technology and service development. We briefly discuss these four categories on the basis of interview fragments. Market-related drivers Firstly, market developments are considered an important driver for decisions. These developments include external developments like consumer and competitive behaviour, technological and economic developments, etc. One interviewee illustrates the role market developments play in decision-making…

“ It happens that for the sake of a strategic or tactical reason (for example not losing the

battle with competitors, preservation of middle-term market shares) a positive decision

will be made for a roll-out of which the cash flow will only be positive after 5, 6, or 8

years. Decisions on broadband access are not isolated decisions, but are linked with

market shares.”

Financial drivers As a second group, we distinguish financial drivers, consisting of costs, cost reduction (rationalisation) and revenues. The importance of financial aspects is underlined by two of the interviewees…

214

“The clearest driver is purely financial, meaning cost reductions and revenues.”

“A cyclically discussion is on barely decreasing investments in the existing network and

limited uptake of new infrastructures. What is the investment plan and which cost

rationalisation can we make?”

Technological drivers Thirdly, we distinguish technology drivers, consisting of technological developments, scalability of technologies and limitations (boundaries) of the existing networks, as illustrated by the following interview fragments…

“..At a certain moment it appears that what you initially built reaches its boundaries. At

that moment, you must extend, upgrade your network”

“Another trigger is a new technology, which makes it possible to perform cheaper, or

better scalable or it makes new services possible.”

Service development Finally, service development is another important driver for the decision whether or not to upgrade. Service development as well as end-user research is considered very important. In the end, an end-user buys a service, not a network. It is not the network, but the services that demand the entire process. The importance of the role of service development is explained by the following interviewee…

“A network does not sell on its own, offering services is necessary for this. As a company,

you can roll-out high bandwidths everywhere, but when there are no services to sell these

bandwidths, because customers do not want them or because they are simply not

available, it does not make much sense.”

One interviewee states that service developments also significantly influence the choice in favour of technologies…

“Introducing new services sharply reveals the hindrances of the existing network.

With the introduction of really new services the existing network turns out to be

insufficient. These services and the vision for these services are drivers for your

capital assets.”

Streaming video (movies and TV), teleworking and peer-to-peer are mentioned as services driving up the need for more bandwidth the most. The role of the Internet provider in offering services is changing. This changing, more powerful role of the end-user was mentioned in several interviews. One interviewee explains…

“Turbulence will remain and end-users become more and more demanding and their

influence on decision-making processes in telecom companies will increase. The risks

will not disappear just like that.”

215

Most customers find their own way on the Internet and buy services at other places than at the Internet provider. This behaviour has its repercussions on the market, as clearly expressed by this interviewee…

“One sees that more and more brokers are eliminated from the value chain.

Packagers like Canal+ are elapsed in the time that consumers couldn’t gather their

own services. To an increasing extent, a business model now becomes visible in which

customers themselves go to content producers and, for example, download movies

from Walt Disney studios directly. The idea of ‘I know what’s good for you’ does not

work anymore. This change in the market causes uncertainty among many

operators.”

Sense of urgency

A second important driver for decision-making and especially for the acceleration of decision-making is the experience a ‘sense of urgency’ to make a decision. The need for urgency was echoed throughout the interviews and illustrated by the following fragments…

“Innovation always takes place under urgency. As long there is not really a problem and

the evolutionary way is sufficient to keep pace with the demand, it is fine. Only then,

when the moment is there that you can’t keep up and your existing technology seizes up,

you will be forced to take a decision for another investment.”

“This decision is made quite fast, which can be explained by a high sense of urgency: this

decision means innovation. We must do this to create flexibility and to remain

competitive.”

“The roll-out takes place faster and faster, especially under increasing pressure of

competitors. The pressure to introduce something in the market becomes even bigger at

that point. It’s the drive to introduce things in the market very fast, although it does not

function optimally yet, that’s the risk you have to take.”

Because this variable appears to be very important to get decision-making started, we will look at it in greater detail to find out which variables cause this important sense of urgency. These variables are important because it is highly likely that they will accelerate the decision-making process by forcing decision-makers to choose direction within a short time span. As a result, variables causing a sense of urgency are also drivers for broadband-related decision-making, albeit with more radical consequences. Figure 6.8 shows the concept map for ‘sense of urgency’. Four variables directly causing sense of urgency to make a decision on broadband roll-out can be

216

distinguished. We see two market-related factors (competition, market developments), a technological factor (the necessity to upgrade the network) and a factor related to the internal organisation (business strategy). It is, moreover, seen that the necessity to upgrade the network is caused by the fact that the existing network reaches its technological boundaries.

is cause of

Has strong impact on

is cause of

Has strong impact on

is cause of

is cause of

is cause of

is cause of is cause of

is cause of

Competition {46-35}Business strategy {27-24}

Market developments {25-20}

Sense of urgency {6-9}

Decision making process to upgrade/ rollout broadband {51-63}

Necessity to upgrade network {6-6}

Technological network boundaries {6-3}

Figure 6.8: Concept map for “sense of urgency”

Analysis

The first phase of the upgrade- and roll-out process of broadband networks is the decision-making process to upgrade or roll-out a broadband network. Of the four phases of the model, the first phase of decision-making is clearly the most complex one. All related variables show a non-transparent network, caused by the presence of many and highly interrelating concepts. The variables can be divided into categories related to technology, finance, market and internal organisation. Nine variables are defined as ‘drivers for network roll-out’. These factors are most important in driving network roll-outs. Network upgrade and roll-out is clearly driven by specific technological, financial and market-related drivers. End-user demand and adoption, and the behaviour of competitors drive decision-makers to make decisions. Costs, revenues and the potential to reduce costs (cost rationalisation) are important drivers for decision-making in relation to finance. Decision-makers are also encouraged to make decisions in case of new technological developments, developments that influence the scalability of technologies and limitations of the existing network. In case the existing network configuration reaches its technological boundaries, operators need to decide how to solve this

217

technological problem. A final, interesting driver is service development, which again brings us to the problem of the chicken and the egg. Service developments stimulated operators to make decisions with regard to network roll-out, but these services will only be developed in case enough bandwidth on the existing networks is available for these services. Besides these drivers, five factors are of such importance to decision-makers on network roll-out that they cause an immediate sense of urgency to make a decision. These factors are very important because they will not only accelerate the decision-making process, but most likely force operators to realise a breakthrough in this process within a very short time span. Factors clearly causing such sense of urgency are partly related to the internal business and partly related to external factors. We see two market-related factors: the behaviour of competitors and specific developments in the market. As explained in paragraph 6.2.2, we see the extremely important role of the behaviour of competitors and the changes in the market they can accomplish. Also, the necessity to upgrade the network (caused by the technological boundaries of the network) can lead to breakthroughs in decision-making. This appears to make sense because we already saw the extreme importance of being able to offer the highest bandwidth in the market for the sake of the operator’s reputation (see paragraph 6.2.2). Besides these external factors, the internal business strategy can also lead to a sense of urgency to make decisions, especially in cases where the strategy needs to be revised based on competitive behaviour or developments in the market, like convergence, frequency auctions, technological developments, etc. Fast decision-making is very important to keep ahead of competitors or not to loose too much market share to them. After a decision in favour of a migration path is made, the following phases are more operational in nature and for that reason less complex. This can be attributed to the fact that decisions within phase 1 are so-called strategic or non-programmed decisions. Decisions made in following phases are more operational and are part of the category programmed decisions (see also Chapter 5 and paragraph 6.4.1. The decisions made in these phases are less strategic in nature and more planning-related. Because decision-making is less complex, phases 2, 3 and 4 will be described in less detail than phase 1.

6.6.3 Phase 2: Network upgrade/roll-out

The decision to roll out or upgrade the existing broadband network by following one or more migration paths, results in an operational roll-out. Upgrade/roll-out forms the second phase of the process and is mainly an operational phase. In this phase, several decisions have to be made surrounding the operational progress, the speed of the roll-out, the technology, the contracts with equipment suppliers, (sub)

218

contractors, etc. Concepts affecting the upgrade process include suppliers, competition, bandwidth and choice between technologies. The concept map of network roll-out is less complex than the concept map of the first phase. Table 6.3 shows the variables that are included in the concept map.

Table 6.3: Source and target concepts and relationships phase 2

Source variable Relation

(linking phrase) Target variable Incoming variables

Suppliers (15) Has strong impact on Network upgrade & roll-out Bandwidth (7) Is cause of Network upgrade & roll-out Competition (46) Is cause of Network upgrade & roll-out Choice between technologies (32)

Has impact on Network upgrade & roll-out

Decision-making process to upgrade/roll-out broadband (51)

Has impact on Network upgrade & roll-out

Outgoing variables Network upgrade & roll-out Has impact on End-user demand and adoption (58) Network upgrade & roll-out Has impact on Operational costs (13) Network upgrade & roll-out Mutual influence Customer penetration (11) These concepts and their interrelationships are visualised in the concept map of Figure 6.9.

Figure 6.9: Concept map phase 2- network upgrade and roll-out

Analysis

It becomes clear that suppliers as well as competitors play an important role in the second, operational phase of broadband roll-out. As described in paragraph 6.3.1 suppliers have large impact on technology choices of operators due to their

219

specialised knowledge. This is also visible in this concept map, which shows that suppliers affect the available equipment, but, moreover, that providers are dependent on suppliers concerning the (on time) delivery of this equipment. Also, the important role of competitors is shown in this figure. Competitors mainly affect the speed and timing of the roll-out process. Interviewees indicate that competitive behaviour has an impact on how and with what pace they will roll-out their own network. This partly explains both the delay in fiber roll-out in the Dutch market, but also the extremely high biddings on the UMTS frequencies: “It’s better to go down

all together than giving your competitor competitive advantage.” Most operators prefer to be ‘quick follower’ to being a first mover, because the first mover risks are too high. The choice of technology and the bandwidth affect the roll-out and upgrade in the sense that different bandwidths and technologies demand different infrastructures and equipment. Operational costs are a logic consequence of rolling out a network as well as customer penetration. Due to network effects, customer penetration is also a trigger for further roll-out.

6.6.4 Phase 3: End-user demand & adoption

In the next phase, phase 3, end-users are subscribing to the network and start using it for all kind of services. All interviewees mention end-user demand as being a very important aspect in the roll-out process. As noted by one on the interviewees…

“Big investments are only profitable in case you have enough coverage. A network

comes along with very high fixed costs which only can be earned back when a sufficient

amount of people is using your network. The payback period of your network highly

depends on the adoption time.”

The number of users and their bandwidth usage influence the technological network performance: when customers use more bandwidth, the network will reach its boundaries, which results in worse network performance. As we saw in Chapter 4, end-user behaviour is very unpredictable, making it an important cause of uncertainty about market developments. All interviewees clearly emphasize this uncertainty, as illustrated by the following fragment …

“The most important risks and uncertainties are related to the uptake of demand or, in

other words, the occupation of your network. One assumes a certain occupation, but this

occupation is the most important uncertainty...”

Table 6.4 shows the in- and outgoing variables related to end-user demand and adoption. Figure 6.10 shows the concept map of the concepts and their linking phrases.

220

Table 6.4: Source and target concepts and relationships phase 3

Source variable Relationship

(Linking phrase) Target variable Incoming variables

Service development (18) Has impact on End-user demand & adoption Technological developments (36) Has impact on End-user demand & adoption Market developments (25) Has impact on End-user demand & adoption Network upgrade and roll-out (35) Has impact on End-user demand & adoption Competition (46) Has strong impact on End-user demand & adoption Wholesale customers (5) Has strong impact on End-user demand & adoption

Outgoing variables End-user demand & adoption Is cause of Evolutionary upgrade (10) End-user demand & adoption Is cause of bandwidth (7) End-user demand & adoption Is cause of Bandwidth upgrade existing network (4) End-user demand & adoption Is cause of Evolutionary paths (7) End-user demand & adoption Is cause of Network upgrade delay (11)

End-user demand & adoption Is cause of Uncertainty about market developments (11)

End-user demand & adoption Is cause of Acceleration roll-out (8)

End-user demand & adoption Has strong impact on Decision-making process to upgrade network (51)

End-user demand & adoption Has strong impact on Payback time investments (11) End-user demand & adoption Has impact on Technological performance (13) End-user demand & adoption Has impact on Revenue (phase) (12) Figure 6.10 shows the concept map of the concepts and their linking phrases.

Has strong impact on

Has strong impact on

Mutal influence

Has impact on

Has impact on

is part of

Has strong impact on

is cause of

Has strong impact on

is cause of

Has impact on

Has impact on

Has impact on

is cause of

Has impact on

Has strong impact on

Has strong impact on

Network upgrade/ roll out {39-37}

Choice between technologies {32-51}

Customer penetration {11-13}

Suppliers {15-8} Bandwidth {7-6}

Competition {46-35}

Decision making process to upgrade/ rollout broadband {51-63}

End user demand & adoption {58-34}

Operational costs {13-17}

Figure 6.10: concept map phase 3: End-user demand & adoption

221

Analysis

End-user adoption and demand is influenced by five variables: developments in services, technology and market, network upgrade and roll-out, and competition ( Figure 6.10). Because end-users pay for their subscriptions and bring in money, demand and adoption affect revenues and the payback time of investments. Demand also influences the course of the roll-out: an evolutionary roll-out implies to roll-out along with the demand curve, which explains the relationship between the concepts. On the other hand, disappointing demand will result in a delay of the current upgrade and roll-out and changes in demand (increasing or decreasing) will result in a new decision-making process, balancing costs and benefits. Adoption and use by end-users is influenced by market-related, technological and service-related developments. Demand and adoption affect the way of continuation of the network roll-out and also influence the strategy to follow evolutionary paths. This has to do with the fact that end-user demand and adoption is a direct cause of uncertainty about market developments: in an uncertain environment, it is saver to opt in favour of evolutionary upgrades instead of making risky high investments in revolutionary upgrades.

6.6.5 Phase 4: Revenues

The final phase is the revenue phase. In this phase, adoption will result in actual revenues for the providers because customers will pay for their broadband subscription and services. This phase takes place in parallel with phase 3 because revenues start immediately as soon as adoption takes place. Table 6.5 shows the in- and outgoing variables of phase 4, together with their linking relationships. These concepts and relationships together result in the concept map shown in Figure 6.11.

Table 6.5: Source and target concepts and relationships phase 4

Source variable Relation

(linking phrase) Target variable Incoming variables

Broadband services (20) Has impact on Revenues End-user demand and adoption (58)

Has impact on Revenues

Operational costs (13) Has impact on Revenues Outgoing variables

Revenues Is cause of Network upgrade delay

Revenues Has impact on Decision-making process to upgrade/roll-out

Revenues Has impact on Network upgrade/roll-out

222

is cause of

Has impact onHas impact on

Has impact on

Has impact onis cause of

Has impact on

Has impact on

Has impact on

Has impact on

Has impact on

End user demand & adoption {58-34}

Broadband services {20-13}

Operational costs {13-17}

Network upgrade delay {11-11}

Revenues {12-12}

Decision making process to upgrade/ rollout broadband {51-63}

Network upgrade/ roll out {39-37}

Figure 6.11: Concept map phase 4: Revenues

Analysis

The final stage of the roll-out process is the revenue phase. In most roll-out processes, revenues start to come in the moment adoption starts. We can see in Figure 6.11 that revenues on the one hand increase by end-user demand and adoption. Also, service development increases revenues, because end-users buy service like telephony, Internet access and television from operators and pay for their subscriptions. On the other hand, revenues are reduced by operational costs (OPEX). If we look at the outgoing variables, we see that revenues cause two feedback loops, as described in 6.6.1. Revenues offer the possibility to make new investments in technological and service-related developments. Disappointing revenues, on the other hand, may result in a decision to defer ongoing roll-outs or to revise business strategy. In case revenues are less than expected, the roll-out of certain projects will be stopped or delayed, which explains the feedback loop with network upgrade/roll-out.

6.6.6 Analysis

If we compare the variables from the concept maps to the first phase of the conceptual model (Figure 3.6), we see that the variable ‘density’ has not been mentioned by interviewees, which appears to make sense, in light of the geographical closeness of most premises in the Netherlands. If we compare the variables from the second phase to the conceptual model, the following aspects stand out. The variable ‘scalability’ does not appear in phase 2, although it does in phase 1. In the model, competition appears as variable that is

223

influenced by network roll-out. It reappears in the new model, however, as an influencing variable: competition has an impact on network roll-out, not the other way around. This is, however, a matter of perspective. As far as company A is concerned, company B is competition, but also other way around. As a result, when the network roll-out of company A and company B are both influenced by competition, they both must automatically impact competition. The variable ‘roll-out delay’ does not show in phase 2 of the new model, but it does appear in phases 3 and 4. Operational costs are a logical consequence of rolling out a network as well as customer penetration. Customer penetration is, on the other hand, also a trigger for further roll-out. The incoming variables ‘competition’ and ‘service developments’ (service innovation) are also mentioned in interviews as important incoming variables in phase 3. Several adjustments and changes were made in the new model as compared to the original conceptual model. Comparing the relationships from the concept map of the last phase with the conceptual model, it can be seen that the variables ‘social and economic welfare’ and ‘strategic advantage broadband supplier’ are not mentioned in the interviews. The variable ‘operational costs’ has been validated as an incoming variable of revenues. Analysis with Atlas.ti also results in the relationship: “Revenues is a driver for network upgrade”, which implies that phase 4 also has a feedback loop with phase 2, which appears to make sense, because revenues bring in money that makes it possible to invest in network upgrades. The final model presented in Figure 6.12 is much more complex than the initial conceptual model presented in Figure 3.3. On the one hand, this can be explained by the fact that these models show more detail and have not been aggregated as much as the conceptual model. On the other hand, we gathered new information from the interviews that was not available in the literature we used as the basis for the initial conceptual model. The model provides more and better insight into the process of broadband-related decision-making and roll-out and the factors that are most important in the various phases of the process. Due to the fact that Atlas.ti does not provide the possibility to further aggregate variables based on the frequency relationships between variables are mentioned, a more aggregated model cannot be built based on our qualitative analysis. We can conclude that, based on interview analysis alone, a parsimonious model for drivers and outcomes of broadband roll-out cannot be developed. Further validation and aggregation will take place via quantitative data analysis. New concepts, as gathered by the qualitative analysis in this chapter, will be included in our quantitative analysis, which will focus on both the clustering of concepts via factor analysis and the relationships between concepts.

224

Reg

ulat

ory

issu

es

Tec

hnol

o-gi

cal

alte

rnat

ives

Vis

ion

&

entr

epre

-ne

ursh

ipC

osts

For

ecas

ting

Fut

ure

proo

fnes

sS

cala

bilit

yU

ncer

tain

tyE

xpecte

d

dem

and

Tec

hnol

ogi

caln

etw

ork

boun

darie

s

Bus

ines

s ca

se

Com

petit

ion

Mar

ket

deve

lop-

men

ts

Bus

ines

s st

rate

gy

Sen

seof

ur

genc

y

Dec

isio

nbr

oadb

and

upgr

ade

Inve

stm

ents

Fin

anci

al

reso

urce

sN

etw

ork

upgr

ade/

ro

llou

t

End

use

r de

man

d&

ad

optio

nR

even

ues

Tec

hnol

ogy

deve

lop

men

ts

Ser

vice

de

velo

pm

ents

Cho

ice

betw

een

tech

nolo

gies

Ban

dwid

th

Cus

tom

erpe

netr

atio

n

Pay

back

tim

e in

vest

men

ts

Net

wor

kup

grad

e de

lay

Acc

eler

atio

nro

llou

t

Tec

hnol

o-gi

cal

perf

orm

ance

Bro

adba

ndse

rvic

es

Unc

erta

inty

mar

ket

deve

lopm

ents

upgr

ade

exis

ting

netw

ork

Evo

lutio

nary

upgr

ade

Ope

ratio

nal

cost

s

Fig

ure

6.1

2:

Adju

sted c

once

ptu

al

model

225

6.7 Conclusions

In this chapter, we focused on the qualitative part of our research validation. We started by providing more insight into the most important factors in the three domains of broadband roll-out as introduced in Chapter 2: technology, market and policy. In addition, we examined the applicability of several theories with regard to the broadband market. Financial theory has been validated by researching the applicability of financial assessment methods (a.o. Oslington, 2004; Neufville, 2003; Keswani & Schackleton (2004). Also, we examined the applicability of critical mass and broadband usage. Moreover, we investigated the validation of the several ‘sources’ as described in the theoretical framework on risks and uncertainties earlier in Chapter 4 and investigated the preference of decision-makers in the field for financial assessment methods. We then provided more insight into two types of decisions: programmed and non-programmed decisions (Simon (1960); Lin (2004)) and the actors involved in the decision-making process. Furthermore, we examined the applicability of two decision-making theories, i.e. the Garbage Can model by Cohen, March & Olsen (1972) and the theory of Logical Incrementalism (Quinn, 1980; Das & Teng, 1999). We concluded by validating the variables (concepts) from our conceptual model developed in Chapter 3. It becomes clear that the distinction between non-programmed and programmed decisions, as introduced by Simon (1960), as well as the further distinction between short-term operational decisions and periodic control decisions (Lin, 2004) can very well be applied to the different decisions that are part of the process of broadband upgrade and roll-out. Minor decisions concerning the implementation are moved to operational departments and involve more people. The major strategic decisions, concerning, for example, investments, are made by senior management in small groups. Strategic decision-making takes between six months and a year. Factors determining the duration of the decision-making process can mainly be found in complexity of the object of decision-making, finance and the availability of a clear ‘sense of urgency’ to arrive at a decision. Roll-out times vary between several months to ten years and depend on the extension of the roll-out: systems, platforms or new technologies. Duration, moreover, depends on the amount of technological adjustments, in other words the evolutionary or revolutionary character of the roll-out. The third determinant for the duration of the roll-out of networks is scale, in other words the difference between upgrades of the core network alone or a roll-out towards the entire footprint. Costs of networks are mainly driven by IT systems and physical costs, comprises digging, co-location, adjustments to homes and equipment. In all the companies we consulted, investments in different upgrade alternatives are assessed via the NPV

226

method possibly in combination with other methods like Return on Investment (ROI) or scenario analysis. This has been argued earlier by Keswani & Schackleton (2004), who state that the Net Present Value rule has emerged as the dominant decision rule for investment-related decisions. The method of Real Option Analysis (ROA) is not being applied by any of the companies we consulted. This is remarkable considering the fact that several studies have shown that this method is more suitable for assessing investment options in uncertain environments, like communication and information technology (a.o. Oslington, 2004; Neufville, 2003). The most important technological developments are found at the protocol level of the networks, consisting of developments in the Ethernet, IP and Docsis protocols. Also, developments in compression technologies are considered highly important. By making it possible to transmit more and more data over the same networks these technologies give operators more time to gain revenues from sunk investments in the existing infrastructures and offer them the possibility to postpone new risky, large infrastructural investments. Market parties show diverging behaviour with respect to following international standards in their choice between technologies. Reasons for following standards are mass production, resulting in lower prices and avoiding an isolated position as a result of the ‘Law of the handicap of a head start’. A reason not to follow worldwide standards is that new versions of an international standard may require drastic, but necessary network adjustments, demanding significant upfront investments, which investments must be balanced against investments in other (possibly non-standardised) technologies or revolutionary upgrades like fiber to the home. In Chapter 5, we concluded that two decision-making theories could be applied to broadband-related decision-making: the Garbage Can theory and the theory of Logical Incrementalism. The Garbage Can theory by Cohen, March and Olsen (1972) can only be applied to organisational anarchies. Cohen et al. drew up three criteria for organisational hierarchies: problematic goals, unclear technology and fluid participation. Based on the fact that our analysis has proven that these three characteristics for organised anarchies can be applied to telecom companies, we can conclude that telecom companies can be qualified as organised anarchies in the way Cohen et al. 1972 meant in their theory. Our results have, moreover, grounded that strategic decision-making with regard to network upgrades and roll-out is not a structured, rational process, but can rather be characterised by connections of available and suddenly arising problems and solutions in combination with people being available at a certain time and place. For that reason, the decision-making processes concerning broadband roll-out fits the characteristics of the Garbage Can theory by Cohen, March and Olsen (1972).

227

Next, we assumed that the theory of Logical Incrementalism (Quinn, 1980) would also be applicable and that strategic decision-making with regard to broadband network roll-out is incremental in nature. Based on Quinn, we assumed that decision-making in broadband is a consistent movement aimed at reaching strategic objectives that are broad and relatively vague to keep the option for developments and adjustments open in case more information becomes available. Das & Teng (1999) formulated three characteristics of the Logical Incrementalism theory being 1) incremental nature, 2) broad and relatively vague objectives, and 3) options for developments and adjustments. Throughout our interviews various examples were given that ground the three characteristics of incrementalism as formulated by Das & Teng (1999). Moreover, it has become clear that strategic decision-making is an incremental process. Top executives work consciously towards their, often vague and broadly formulated objectives, but they are also prepared to be adaptable along the way to these objectives. Here, a clear link can be seen with theory of Real Options Analysis. The applicability of the theory of Logical Incrementalism is, moreover, supported by the fact that cable operators as well as PSTN operators agree on preferring evolutionary paths above a revolutionary strategy for their physical networks. Networks are upgraded at the moment a critical mass for a certain bandwidth is visible and operators follow an ‘invest-as-you-grow’ strategy. Sometimes this evolutionary network upgrade is combined with a more revolutionary approach on the higher protocol layer. Three causes can be pointed out for operators clearly preferring an evolutionary approach. First is the combination of the clear lack of demand from consumers for more network capacity than currently available with the fact the existing networks are capable of handling the expected demand growth in the foreseeable future. Secondly, investments are too high and the expected revenues too uncertain to build a positive business case for a fiber to the home infrastructure. The number of households needed to recover the huge upfront investments only leaves room for one national operator. A third reason lies in the chicken and egg problem: in case use of the infrastructure by service providers is not guaranteed beforehand, no upgrades to fiber will take place due to uncertainty with regard to the expected revenues. Within the value chain, horizontal as well as vertical cooperation between market parties can be seen. Horizontal cooperation is seen within the development of standards or in joint lobbies. Vertical cooperation involves mainly suppliers and building companies during roll-out. This close cooperation, which is not always voluntary, results from a strong mutual dependency between market players. This is demonstrated by De Baas (1998), who mentions this mutual dependency as one of the characteristics of complexity (see Chapter 5). As a result, competition is a very important factor in deciding to upgrade their networks. The high level of

228

competition has brought about three important effects. Firstly, innovation in technology and services is a clear result of competition. Secondly, it resulted in a price war for broadband access with more and more bandwidth for lower prices as result. Although also indicated as being of the utmost importance, actual and expected demand in practice do not seem to be as important as customer satisfaction and reputation-building as drivers for the decision whether or not to upgrade. Finally, competition has brought the Netherlands to a leading position in the world in the field of broadband (i.e. cable and ADSL) roll-out. Also, public broadband initiatives to stimulate fiber roll-out have contributed to Dutch broadband development and in bringing the Netherlands to its leading position with regard broadband connection. Another positive result of public market intervention is that it created awareness and understanding among operators about the need to change their monopolistic attitude into a more customer-oriented one. On the other hand, governments should be conscious of the possible negative side effects of their interference in the private market. Foreign market parties indicate they are considering postponing or terminating their investments in the Dutch market due to an instable and distorted market environment. Moreover, market distortion takes place due to new players buying up networks of bankrupt broadband initiatives. A climate of political and regulatory uncertainty poses a threshold to operators concerning large infrastructural investments. Meijer et al. (2006) draw the following distinction with regard to the sources of risk and uncertainty: technological, resource, competitive (new entrants & rivalry), supplier, consumer uncertainty (buyer power) and political/regulatory uncertainty (see also Chapter 4). It becomes clear from the interviews we conducted that these six sources of uncertainty are clearly experienced by broadband decision-makers. These risks and uncertainties have two important effects. The first is building in flexibility by keeping open more optional development paths. This way, different development steps can be chosen depending on the circumstances and adjustments can be made in case of a changing environment. Here, we see a clear link with the theories of Logical Incrementalism and Real Options Analysis. With regard to this effect, it is striking that Real Options Analysis is not applied in practice by telecom companies as a method for financial assessments. Although operators in practice make decisions according to this method (Logical incrementalism) they do not incorporate it into their investment-related decisions. The second effect, brought about by regulatory uncertainty, is delay or reconsideration of investments. To reduce risks and uncertainties, several methods are applied by operators. Technological analysis, forecasting and pilots are applied most frequently. Analyses and forecasting take place early on in the decision-making process, when alternatives are investigated. Pilots are performed later on

229

in the process to prepare for possible practical unforeseen side effects of implementing a new technology. After validating the concepts of the conceptual model we introduced in Chapter 3, we can conclude that most concepts of the original conceptual model have been validated by the qualitative data analysis, albeit not always in their original position within the model or at the same level of aggregation. On the other hand, we gathered new information in the interviews, which resulted in additional concepts. Most noticeable are several feedback loops in the process, which imply it is an ongoing and repeating process. Secondly, the interviews show the importance of the ‘soft side’ of the decision-making process. For that reason, market developments, the behaviour of competitors, the necessity to upgrade the network and the strategy of the business cause a high sense of urgency to make a decision about rolling out or upgrading the network are important potential accelerators of the decision-making process. We can conclude that, based on interview analysis alone, a parsimonious model for drivers and outcomes of broadband roll-out cannot be developed. Further validation and aggregation will take place via quantitative data analysis. New concepts, as gathered by the qualitative analysis in this chapter, will be included in our quantitative analysis, which will focus on both the clustering of concepts via factor analysis and the relationships between concepts. In Chapter 7, we discuss the quantitative results of this study.

230

6.8 References

Åhlberg, Mauri (2004). Varieties of concept mapping. Proceedings from The First Conference of Concept Mapping, Pamplona, Spain.

Alleman, J. (2002). A new view of telecommunications economics. Telecommunications Policy, 26 (1) 87-92(6).

Alpert, Sherman R. & Keith Grueneberg (2000). Concept Mapping with Multimedia on the Web. Journal of Educational Multimedia and Hypermedia, 9 (4), 313-330.

Baas, H. de (1998). Hectiek en vereenvoudiging van besturingscontext- Afvalbeleid in Nederland naar een nieuwe ronde. Bestuurskunde, 5, 210-218.

Cañas, Alberto J., Roger Carff, Greg Hill, Marco Carvalho, Marco Arguedas, Thomas C. Eskridge, James Lott, and Rodrigo Carvajal (2005). Concept maps: integrating knowledge and information visualization. In: S.O. Tergan & T. Keller (Eds.), Knowledge and Information Visualization: Searching for

Synergies, (pp. 205-219). Heidelberg/NY: Springer Lecture Notes in Computer Science.

Cohen, March & Olsen (1972). A Garbage Can Model of Organisational Choice. Administrative Science Quarterly, 17(1), 1-25.

Das,T.K. and Bing-Sheng Teng (1999). Cognitive biases and strategic decision processes: an integrative perspective. Journal of Management Studies, 36

(6), 757-778. Expertgroep Breedband. (2002). Nederland Breedbandland, recommendations to

the cabinet from the national broadband expert group. The Hague. Gaines, B.R. and M.L.G. Shaw (1993). Supporting the creativity cycle through

visual languages. Proceedings from AAAI Spring Symposium ‘93: AI and Creativity. Menlo Park, California, 155-162.

Junkus, Justin J. (2008) DOCSIS 3.0 Drivers and Speed Bumps. Retrieved April 22, 2008 from http://www.cable360.net/ct/sections/features/28858.html.

Keswani, A. and Schackleton, M.B. (2006). How real option disinvestment flexibility augments project NPV. European journal of operational research, 168 (1), 240-252.

Limkeatcherdchoo, S. (2006). Understanding dynamics between firm capabilities and changing environment: the case of mobile data service innovation (presented Doctoral dissertation, University of Strathclyde, 2006).

Lin, Hui-Chao (2004). Decision theory and analysis. Futurics, 28 (1&2), 27-45. Marcus, A.A. (1981). Policy uncertainty and technological innovation.The Academy

of Management Review, 6 (3), 443–448. Meijer, Ineke S.M., Marko P. Hekkert, Jan Faber and Ruud E.H.M. Smits (2006),

Perceived uncertainties regarding socio-technological transformations:

231

towards a framework. International Journal of Foresight and Innovation

Policy, 2 (2), 214-240. Miles, M.B. and A.M. Huberman (1994). Qualitative data analysis: An expanded

sourcebook (2nd ed.). London/New Delhi: Sage Publications. Neufville, Richard de (2001). Real options: dealing with uncertainty in systems

planning and design, paper presented at 5th international conference on “Technology, Policy and Innovation”, Technical University Delft, Delft, The Netherlands. Retrieved January 29, 2008 from: http://msl1.mit.edu/mib/dsp/curricula.MIT.edu/~dsplan/docs/papers/Delft.htm.

Neufville, Richard de (2003). Real options: Dealing with uncertainty in systems planning and design. Integrated Assessment, 4(1), 26-34.

Novak, J.D. (1998). Learning, creating and using knowledge: Concept Maps™ as

facilitative tools in schools and in corporations. London: Lawrence Erlbaum.

Novak, J. D. and Gowin, D.B., (1984). Learning how to learn. New York: Cambridge University Press.

Oslington, Paul (2004). The Impact of Uncertainty and Irreversibility on Investments In Online Learning. Distance Education, 25 (2), 233-242.

Porter (1980). Competitive strategy: Techniques for analyzing industries and

competitors. New York: Free Press. Quinn, James Brian (1980). Strategies for change-logical incrementalism (1st ed.).

Homewood: R D Irwin. Simon, H.A. (1960). The new science of management decision (3rd print). New

York: Harper and Row. Tergan, Sigmar Olaf (2005). Digital concept maps for managing knowledge and

information. In: Tergan & Keller (Eds), Knowledge and Information

Visualization: Searching for Synergies (pp.185-204). Berlin/Heidelberg: Springer-Verlag.

Vermaas, Karianne (2007), Fast diffusion and broadening use (Doctoral dissertation, University of Utrecht, 2007).

232

233

Chapter 7 Towards an explanatory

parsimonious model- quantitative analysis

The qualitative research described in Chapter 6 resulted in a model that is even more complex than our conceptual model presented in Chapter 3. We can conclude that, although it provides a great deal of useful information on decision-making in broadband, qualitative research does not bring us closer to our objective: a parsimonious model that provides insight into determining factors for broadband-related decision-making and outcomes of network roll-out. As a next step, we apply quantitative research to realise the parsimony for which we are looking. Our quantitative research starts from the conceptual model presented in Figure 3.3. This model is the result of an extensive Meta-analysis on broadband literature (see Chapter 3 for details on method and model development). This model is validated in two ways. Firstly, as described in detail in Chapter 6, it is validated based on qualitative research, i.e. interviews with executives of major telecommunication operators and network providers. Because this did not result in a desired level of parsimony, we carried out a second validation based on quantitative research among stakeholders and experts in the broadband domain. This quantitative validation consists of two online surveys: a pre-test and a main study. First, we conducted a small-scale pre-test to develop scales and get insight into the construction of our questionnaire and items. The results of the pre-test provide some initial insight into the validity of the conceptual model. As a second step, we conducted another online survey, which was directed more to stakeholders. While the pre-test only included Dutch participants in the telecommunication market, the final survey has an international character. This makes it possible to get results of a more general rather than a country-specific character. Also, the sample was broader and larger. The pre-test was performed to develop scales via exploratory factor analysis. These scales are used as the basis of our main study. The main study first consists of a confirmatory factor analysis to create a measurement model. Secondly, this measurement model forms the basis of a structural equation model. In this chapter, we discuss the results of our quantitative research. In paragraphs 7.1 and 7.2, we discuss the method, i.e. sample and scale development of our pre-test and main study. In paragraph 7.3 we discuss the descriptive results of our main study. As a next step, we discuss the outcomes of our confirmatory factor analysis, the measurement model and the structural regression model in paragraph 7.4. We conclude with a discussion and conclusions.

234

7.1 Exploring the field: national pre-test

In this paragraph, we provide insight into the method we applied and the development of scales of our national pre-test. These scales form the input for a confirmatory factor analysis, which we discuss in the next paragraph. We start by discussing the method we applied.

7.1.1 Method

Questionnaire To test the conceptual model from Chapter 3, we developed a questionnaire. The questionnaire contains about 19 questions, consisting of multiple items. The questionnaire was administered via the Internet, via an online survey software tool called Surveyworld. The questionnaire consisted of four parts. We began by asking several questions on the impact factors from several disciplines (technology, organisation, market) have on the decision-making process to upgrade or roll-out broadband networks. Also, questions were asked on factors that influence reconsideration of a previously made decision to delay. In the second part, we asked questions about risk and uncertainty, while the third we designed to gain more insight into variables influencing network viability and outcomes of network roll-out. The questionnaire concluded with several questions on professional background and experience to gain more insight into the characteristics of the participants. Sample As respondents we selected people who are involved in advice, research or decision-making with regard to broadband. Because there is no directory of decision-makers, researchers and/or advisors, we had to rely on our own network. A number of potential respondents were approached via e-mail, asking for their cooperation. If respondents agreed, a link to the website where the questionnaire was published was sent to them. Sixty people were invited to fill in the survey. Of the forty persons who agreed to do so, 37 actually did. We are aware of the fact that this is not a random sample and we are therefore limited with regard to the external validity of the results. Nevertheless it is important to sketch the background of our respondents. They are predominantly: 92%. On average, they have fifteen years experience in the telecommunication sectors. One person recently got involved in telecommunications, while another respondent has a working experience of over 32 years. Most respondents have more than ten years experience. Table 7.1 describes the job titles of the respondents.

235

Table 7.1: Professional background of respondents

1. Decision-maker for a telecom infrastructure provider 3%

2. Decision-maker for a cable provider 5%

3. Decision-maker for another type of infrastructure provider 3%

4. Decision-maker for a DSL provider

5. Decision-maker for a content provider

6. Decision-maker for a equipment supplier 8%

7. Advisor, consultant 60%

8. Regulator, or policy-maker

9. Researcher 16%

Other 5%

N= 37 100%

7.1.2 Development of scales

Technology items Based on the model, we developed a questionnaire in which the concepts and relationships could be measured. Due to the fact that no standard scales are available, we developed a number of new scales. We used explanatory factor analysis, principal component analysis, and varimax rotation in SPSS 14.0. We will first discuss a scale for technologies as such. The outcomes of our factor analysis are presented in Table 7.2.

Table 7.2: Factor Analysis relating technologies

I II

1.6 New hybrid radio-optical technologies, like Hybrid Fiber Radio .86

1.5 New wireless optical-based technologies, like free space optics .74

1.4 New radio-based technologies, like Wimax, or microwave or millimetre wave distribution systems

.65

1.7 Development of new compression technologies .62

1.2 New coax-based technologies, like Narad networks .90

1.1 New copper-based technologies, like xDSL, etherloop, Ethernet

first mile over copper (EFMC)

.80

Eigen value 2.209 1.664

Explained Variance 37% 28%

KMO measure of sampling adequacy .57

Cronbach’s α .70 .71

It is striking that two scales for influence of emergent technology emerge. The first scale discusses all kind of new- mainly wireless and radio-optic-related technologies, while the second scale is more related to more familiar technologies: cable and copper. The first scale has a Cronbach’s α of .70. Scales with α between .70 and .80 are acceptable. The more traditional new broadband technologies load also on one factor and have an acceptable Cronbach’s α, even for the very small sample. Seen the fact that only a limited number of items are included and the low number of observations, we may conclude, based on the factor analysis and the α,

236

that the items as used can be incorporated into a scale for the importance of traditional technologies and a scale for the relevance of emerging technologies. The factor scores are saved using the Anderson-Rubin method (avoiding correlations between factor scores).

Technology characteristics Factor analysis of items regarding specific characteristics of technologies that may play a role in the decision-making result in a three factor solution. KMO is mediocre (.66). The first factor concerns scalability, flexibility, maturity and compatibility issues, while the second scale has to do with standardisation and the third refers to the trade-off between capacity and reach. With the exception of the last scale (not in the .70 - .80 range), Cronbach’s α is acceptable. See Annex A, Table A1 for factor scores. Market items Factor analysis of various market-related items does not result in satisfying factors with acceptable values for Cronbach’s α. Also, the KMO index is below the level of acceptance. Organisational items comprise some financial items like revenues and CAPEX, marketing-related items and competition. See Annex A, Table A2 for results of the factor analysis. Organisational items Within organisational items, three factors with acceptable alpha values are the result of factor analysis (see Annex A, Table A3). The first factor clearly addresses (the impact on) internal business processes. Factor II refers to the power of different parts of the company and the last factor concerns marketing.

Network viability What are the factors that really demand the viability of broadband roll-out? Four factors clearly have the highest impact on viability. The first factor comprises items related to the offering by competitors in the market. The second factor relates to end-user bonding and satisfaction. Critical mass of broadband users and services make up the third factor. The fourth factor refers to items related to end-user services. Factors III and IV show values for Cronbach’s α which are below the level of acceptance. The results of our factor analysis are shown in Table 7.3.

237

Table 7.3: Items influencing network viability

I II III IV

12.2 Competitive offerings of other parties .92

12.1 Competitive pricing .91

12.3 Marketing campaigns focused on broadband .60

12.4 Satisfaction of lead users .87

12.5 Brand name (image) .78

12.6 Critical mass of available ‘real’ broadband

services

.82

12.7 Critical mass of consumers using broadband .79

12.8 Triple play concepts .91

12.9 3rd party end-user service promotion .59

12.10 The time it takes to change to higher

bandwidths (conversion rate)

.52

Eigen value 2.192 2.114 1.663 1.597

Explained Variance 22% 21% 17% 16%

KMO measure of sampling adequacy .58

Cronbach’s α .78 .70 .55 .61

Outcomes of Broadband roll-out The last factor analysis involves the items representing several tangible and non tangible outcomes of broadband roll-out. We deleted item 13.9 (Increased communication between people) form the analysis. This item originally loaded both on factor I and factor II. As a result, the KMO increased from .53 to .61. Deleting item 13.11 (Decreased digital divide) would result in a slightly lower score for the KMO and a four factor-solution. The factor loadings hardly differ. Interpreting the factors, the first factor indicates increased Know-how and experience with broadband roll-out, the second factor deals with more socio-cultural consequences, including an increased digital divide. The third factor has to do with improved customer attraction and satisfaction, while the last factor refers to direct as well as indirect economic benefits. With the exception of the last factor (economic benefits), we can include these scales in our analysis. The numeric outcomes of our factor analysis are presented in Annex A, Table A4.

238

Table 7.4 summarises the number of corresponding items and the most important values of the scales developed in our pre-test.

Table 7.4: Scales, number of items and main values pre-test

Scale Number

of items

Cronbach’s α

Technological 1:Techn_charact 5 .83

Technological 2:Standardisation 2 .73

Organisational 1:Internal processes 4 .71

Organisational 2:External_ineterests 3 .70

Organisational 3:Marketing2 2 .77

Viability 1:Offer competitors 3 .78

Viability 2:Custumer satisfaction 2 .70

Outcomes 1:Next step roll-out 5 .89

Outcomes 2:Social & cultural effects 4 .81

Outcomes 3:Customer value 3 .70

7.2 International survey

After conducting the pre-test, a second online survey was conducted with an international character and a broader and larger sample. In paragraphs 7.2 and 7.3, we describe the outcomes of this international main study in greater detail. In this paragraph, we first will give more insight into our sample, after which we discuss the construction of scales. To build the scales, we used the scales and items of our pre-test as a basis. However, based on feedback from the participants of our pre-test and the outcomes of expert interviews, some changes were made compared to our pre-test. We discuss these changes below.

7.2.1 Method

Questionnaire Based on the conceptual model, a questionnaire was developed in which concepts and the relationships between them were made measurable. Like our pre-test, the questionnaire was administered via the Internet via Surveyworld. The questionnaire contained about 20 questions, some of which contained multiple items. The structure of our second questionnaire was almost identical to the questionnaire we used in our pre-test, with some adjustments. Firstly, in the first part of our questionnaire we added a question on financial factors influencing the decision to upgrade or roll-out a broadband network. Secondly, in the third part on outcomes and viability of network roll-out, we added some questions to gain more specific insight into numeric values for success of broadband projects. We used two ways to measure this success: firstly, the percentage of expected ROI after two and five years of a network roll-out; secondly, the percentage of the footprint that should be connected to broadband one or five years after initiation of the roll-out project.

239

Sample The target group for our main study was the same as in our pre-test: people who are involved in advice, research or decision-making with regard to broadband. We wanted to have as broad a sample as possible with equal numbers of participants with a background in academics, consultancy and decision-making in telecommunication companies. This way, the survey would give a combined view from people who are making decisions regarding roll-out themselves, people who advise them and people who conduct research in the telecommunication sector. To make sure that only professionals with relevant experience would fill out the survey, we drew up a profile. Only people that matched the profile were asked to fill out the survey. When we contacted a larger group at once, like the ITS membership list, people were asked only to react if they matched the profile. The survey targeted persons with the following profiles: 1. Managers and experts in international telecom companies who are directly

involved in the decision-making process regarding the upgrade and/or roll-out of fixed broadband networks (network strategy);

2. Consultants, researchers and other experts who have an understanding of the technological, financial and organisational aspects that play a role in network strategy.

Potential respondents were approached via e-mail, asking for their cooperation. If respondents agreed they were sent a link to the website where the questionnaire was published. Besides a personal network, the membership list of the International Telecom Society (ITS) was approached to fill in the survey. Moreover, several respondents were brought to our attention via persons we contacted to fill out our survey. Eventually, 87 professional respondents filled out the survey, originating from fourteen countries. With 60%, the majority of the respondents are Dutch (48 respondents). 31% or 27 respondents come from other European countries, namely Finland (5), Spain (4), Germany (3), Norway (3), Denmark (1), France (2), Belgium (2), and United Kingdom (2). The other respondents are from other continents: United States (6), Australia (1), and Asia, represented by Japan (1), and Malaysia (1). The remaining seven respondents filled out the survey anonymously, so their country of origin could not be traced. Figure 7.1 shows the overview of the professional backgrounds of the participants. The biggest group consists of consultants/advisors (47%), followed by decision-makers in the telecommunication industry (39%). With 26%, the group of researchers is somewhat smaller. Many researchers who initially agreed to fill in the survey later indicated that the questions required such specific practical knowledge that they were not able to complete the survey.

240

5%

6%

26%

39%

47%

0% 10% 20% 30% 40% 50%

Other:

Regulator or policy maker

Researcher

Decision maker in telecommunication

industry

Advisor, consultant

Figure 7.1: Professional backgrounds of respondents

Figure 7.2 visualises the years of experience in telecommunications the respondents have. 65% of the survey respondents have over 10 years experience in the telecommunication market, of which 26% over 20 years. We can conclude that this amount of experience represents a professional sample.

26%

20%

20%

20%

15%

0% 5% 10% 15% 20% 25% 30%

>20 years

15-20 years

10-15 years

5-10 years

1-5 years

Percentage respondents

Years

of

exp

eri

en

ce i

n T

ele

co

m in

du

str

y

Figure 7.2 Years of professional experience in telecommunication

Although our sample seems somewhat small, it should be realised that the number of people working in broadband-related decision-making at a strategic level is not large, most of the times less than 5 people per telecom company. Moreover, scientific research on broadband-related decision-making is not represented by a very large research community. For that reason, 87 respondents can be considered an acceptable result. Moreover, the high level of experience of our respondents implies a very professional sample, which increases the value of the answers to the survey questions.

241

Figure 7.3 shows a comparison between the profession of respondents from our pre-test and out international survey. As can be seen, the largest part of our national sample consists of consultants and researchers in the telecom market. Only a small part involves immediate stakeholders. The sample of our international survey is more balanced. 39% are decision-makers from the industry (immediate stakeholders) against 19% in our pre-test. In both samples, the average experience in the industry is approximately 15 years.

0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60%

Other:

Decision maker for a content provider

Decision maker for an equipment supplier

Decision maker for a cable provider

Decision maker for other type of infra provider

Regulator or policy maker

Decision maker for DSL provider

Decision maker for telecom infra provider

Researcher

Advisor, consultant

National pre-test

International survey

Figure 7.3 Professions of respondents pre test versus international survey

7.2.2 Construction of scales

For our main study, we used the scales we developed in the pre-test as a starting point. Some adjustments were made to these scales by adding or removing items and, in one case, adding a new scale. These adjustments were in the first place based on the outcomes of our qualitative analysis (see Chapter 6). In interviews with decision-makers on broadband roll-out, several items were mentioned as important, which had not been included in our pre-test. Secondly, some adjustments were made based on the comments from the participants of our pre-test. In our pre-test we asked respondents to give feed-back on the questionnaire. We also received several e-mails with some feedback remarks. The biggest adjustment we made was to develop an additional scale for financial items. In the pre-test, financial items were incorporated into the organisational scale, but this

242

resulted in factors that were all below the acceptable level of Cronbach’s α. From this we concluded that the organisational and financial items differed too much to load on the same factors. For that reason, we developed a new scale with financial items. Broadly speaking, items from the main study load on the same factors as they did in the pre-test. Some differences can, however, be noticed. These differences can be attributed to several factors. Firstly, as described above, some changes were made with regard to the items and a scale was added. Secondly, the sample of our main study differs somewhat from our pre-test. The second sample represents more decision-makers in the field and more nationalities. The sample of our pre-test consisted mainly of consultants and researchers, and all the respondents were Dutch. Table 7.5 contains an overview of the scales, items and values for Cronbach’s α for the scales that emerged from our main study.

Table 7.5: Scales, items and main values main study

Scale

Number of

items Cronbach’s α

Technological 1: Techn_charact 8 .87

Technological 2:Development BW & services 2 .71

Market 1:Demand 4 .74

Market 2:Service development 3 .76

Organisational1: External_interests 4 .79

Organisational 2: Internal processes 2 .80

Organisational 3:Vision & strategy 2 .77

Financial 1:Costs & revenues 5 .73

Financial 2:Cost reduction 2 .72

Viability 1:Marketing &services 6 .75

Viability 2:Offer competitors 2 .76

Viability 3:Critical mass 2 .74

Outcomes 1: Broadband roll-out 4 .84

Outcomes 2: Social & cultural effects 5 .83

Outcomes 3:Customer value 5 .79

Compared to the scales from our pre-test, we see some differences (compare Table 7.4). We see a new second scale for technological factors. The scale standardisation is being replaced by the development of bandwidth and services. For organisational factors, the first two scales remain the same, but the third is replaced by the scale vision & strategy. These softer aspects were not included in our pre-test, but were added based on qualitative results. Also, the third scales of

243

factors determining broadband viability and outcomes differ from our pre-test. All other scales remain the same. To validate the scales developed with exploratory factor analysis, we conducted a confirmatory factor analysis. Because of our relative small sample, we were obliged to exclude some more scales from our measurement model during confirmatory factor analysis, due to conditions that have to be met in order to be able to carry out the analysis. Before describing the results and development of our explanatory analysis, we first present the descriptive results of our exploratory research in the next paragraph.

7.3 Exploring broadband roll-out: descriptive results

In this paragraph, we present the descriptive results of our main quantitative study. Thos paragraph will start with discussing the results of the content-related part of our research. First the impact of technological, market-related, organisational and financial factors on the decision-making process to roll-out or upgrade broadband networks will be described. As a next step, the importance of several factors determining the outcomes of broadband roll-out is described in greater detail, after which we focus on the assessment of risk and uncertainty and the preference of experts and decision-makers in the field for several methods to compare investment options. After discussing the content-related results, we focus on the process-related results of our research, looking at financial assessment methods, differences between evolutionary and revolutionary upgrade strategies and factors that impact reconsidering a previous decision to delay network upgrades or roll-out.

7.3.1 The content-related part: Factors influencing the decision to

upgrade or roll-out broadband

In this paragraph, we describe the content-related descriptive results of our quantitative analysis. The decision-making process to roll-out or upgrade broadband networks is affected by several factors. The impact of technological, market-related, organisational, policy and financial factors on this decision-making process will now be described.

Technological factors

The respondents were asked two questions on the impact of technology-related factors. Firstly, what is the impact of several emerging technologies on roll-out decisions and secondly, to what level do various technological characteristics affect these decisions? Both will be discussed.

244

Emerging technologies In the decision-making process to roll out new broadband networks or to upgrade the existing network, decision-makers are faced with many different possible technologies, varying from incremental upgrades in the existing fixed networks like coax and copper twisted pair to revolutionary upgrades like fiber to the home. Decision-makers can also opt in favour of several wireless technologies (see also Chapter 2 for more details on broadband development paths). The first survey question investigates the influence of several innovative broadband technologies on the decision-making process of telecom operators to upgrade or roll-out their networks. In Table 7.6 the relative influence of emerging technologies on the actual decision to roll-out or to upgrade the network is presented.

Table 7.6: Relevance of emergent technologies

% of respondents, N=87

No influence

at all

Little influence

Influence Large influence

Decisive influence

Mean Std. dev

New optical fiber technologies (e.g. APON,

EPON, Gigabit Ethernet)

3.5 7 23.3 38.4 27.9 3.80 1.038

New copper-based technologies (e.g. xDSL,

etherloop, Ethernet first

mile over copper (EFMC))

4.7 9.4 28.2 32.9 24.7 3.64 1.100

New radio-based

technologies(e.g. Wimax, WIBRO, microwave or

millimetre wave

distribution systems)

7.1 29.4 30.6 23.5 9.4 2.99 1.096

Development of new

compression technologies

8.4 34.9 27.7 16.9 12 2.89 1.158

New coax-based

technologies (e.g. Narad

networks, Ethernet over coax)

10.8 26.5 33.7 22.9 6 2.87 1.079

New hybrid radio-optical technologies (e.g. Hybrid

Fiber Radio)

14.1 45.1 23.9 15.5 1.4 2.45 .986

New wireless optical-

based technologies (e.g. free space optics)

15.6 54.5 18.2 9.1 2.6 2.29 .930

New copper and fiber technologies have the largest influence with 58% and 66% of the respondents opting in favour of the answers large or decisive influence respectively. With 70% opting in favour of no or little influence and the smallest standard deviation, new wireless optical technologies have the smallest impact on the decision-making process and respondents are most unanimous about this result. They are the most divided about the influence of the development of new compression technologies, as presented by the highest standard deviation. 43%

245

hold the opinion that compression technologies play no or only a little role, while 29% on the other hand consider these technologies as having large or decisive influence on the decision-making process. Technology characteristics Before a decision can be made about which technology to roll-out and how, several technological characteristics of the available technologies have to be evaluated. (See Chapter 2 for a comparison of broadband technologies on several characteristics). Fifteen technological characteristics and their influence on the decision-making process to upgrade or roll-out a network are shown in Table 7.7. The results show that all characteristics presented to the respondents in this question have at least a medium-sized influence on the decision-making process. With 52% of the sample opting in favour of the answer large influence and 29% for decisive influence and the smallest standard deviation, we can conclude that transmission speed is the most important technological variable in choosing for a certain technology. Also, future-proofness, reach and technological scalability have a very strong impact on the choice of technology, with decisive influence percentages of 29%, 27% and 27% respectively. Based on the standard deviation, respondents agree the least on the influence of the development of broadband services, although many more believe that broadband services have large/decisive influence (67%) rather than little/no influence (20%).

Table 7.7 Technological characteristics that play a role in decision to roll-out

% of respondents, N=87

No

influence

at all

Little

influence

Influence Large

influence

Decisive

influence

Mean Std

dev

Transmission speed

(expressed as the speed in

Mbps for a typical

subscriber)

0 3.4 16.1 51.7 28.7 4.06 .768

Technological scalability 1.2 7.1 16.5 48.2 27.1 3.93 .910

Reach (expressed as a

maximum number

distance between user

premises and POP

guaranteeing the desired

speed)

1.2 7.0 24.4 40.7 26.7 3.85 .940

Future-proofness of

technology 2.4 6.0 29.8 32.1 29.8 3.81 1.012

Compatibility with existing

network technologies 7.1 25.9 48.2 18.8 3.79 .832

246

% of respondents, N=87

No

influence

at all

Little

influence

Influence Large

influence

Decisive

influence

Mean Std

dev

Bandwidth development

on existing copper and

coaxial networks

2.3 8.1 25.6 45.3 18.6 3.70 .946

Standardisation of

technology (either formal

or de facto)

1.2 5.9 35.3 37.6 20.0 3.69 .900

Development of broadband

services 3.5 14.1 15.3 43.5 23.5 3.69 1.091

Geographical scalability 2.4 9.4 23.5 47.1 17.6 3.68 .954

Openness of standards 2.4 4.8 44.0 31.0 17.9 3.57 .922

Development of new

broadband technologies 3.5 5.9 32.9 47.1 10.6 3.55 .893

Maturity of technology 1.2 8.2 37.6 41.2 11.8 3.54 .853

Flexibility of technological

solution 2.3 5.8 51.2 31.4 9.3 3.40 .830

Possibility to phase out old

parts of the existing

network

4.7 15.1 37.2 30.2 12.8 3.31 1.032

Path dependency of choice

of technology 3.9 15.6 40.3 35.1 5.2 3.22 .912

Market-related factors

As described in Chapter 2, in addition to technological characteristics, variables related to development in the market also play a role in the choice between roll-out strategies. Besides the behaviour of competitors and service development market characteristics enclose all kinds of end-user-related variables like density, demand development, forecasts, clustering, etc. The most striking result is the unanimity (43% of the sample and smallest standard deviation) about the decisive influence of density (the number of - potential - customers per square kilometre) on the decision-making process. Other market-related variables that have a decisive influence on the upgrade/roll-out decision are the broadband offering by competitors, expected end-user demand and actual market demand (resp. 31%, 29% and 24% for the answer decisive influence). See Annex B, Table B1 for detailed numbers. Demand and adoption As described in Chapters 2 and 6 end-user demand and adoption are an important factor in broadband roll-out decisions. It is, ultimately, the end-users who bring in the money. Respondents were asked to indicate to

247

what extent end-user demand and adoption influence several important results of migration in broadband networks. The financial impact end-users have is clearly the most important effect of demand and adoption. Impact on the payback time of investments and on the revenues is considered highly important. Over 80% of the respondents share the opinion that demand and adoption have a major or decisive influence on payback time and revenues. This would appear to make sense, because adoption represents money coming in via end-user subscription. But acceleration or delay in network upgrade or roll-out is also very much influenced by demand and adoption. More demand will accelerate roll-out and has a major impact on the bandwidth upgrade of the existing network; lower demand will delay roll-out. It is interesting to see that respondents relate demand and adoption directly to choosing an evolutionary strategy (invest-as-you-grow model). This means that operators gear their roll-out strategy to the level of demand and choose flexibility in their roll-out strategy to roll-out parallel to the demand curve. The results are shown in Table 7.8.

Table 7.8 Impact of end-user demand and adoption on results of broadband roll-out

% of respondents, N=87

No

influence

at all

Little

influence

Influence Large

influence

Decisive

influence

Mean Std

dev

Payback time of

investments 4.6 11.5 51.7 32.2 4.11 .784

Revenues 3.4 16.1 49.4 31.0 4.08 .781

Acceleration of network

upgrade/roll-out 4.6 33.3 46.0 16.1 3.74 .784

Delay of network

upgrade/roll-out 11.5 34.5 41.4 12.6 3.55 .859

Choosing an evolutionary

path strategy (invest-as-

you-grow model)

1.1 5.7 37.9 49.4 5.7 3.53 .745

Bandwidth upgrade of the

existing network 1.1 11.5 33.3 43.7 10.3 3.51 .874

Policy-related factors

There is least consensus about the influence of (local) governments offering broadband. 44% of our sample agrees that public broadband offers have some kind of influence on their decision-making process. Although 28% hold the opinion that public broadband offering has no or little influence, 29% of the respondents see public broadband offers as having a major or even decisive influence on the decision-making process. This difference can probably be explained by the fact

248

that broadband roll-out by local governments does not have the same impact and characteristics everywhere in the world. In some countries, the local offers take the form of small pilots, while in other countries local governments are serious competitors for private telecom operators. In the latter situation, the perceived impact of public broadband offers on private companies also depends on the market position and geographical presence of the company involved, as well as on the public initiative.

Financial factors

Relevant financial issues Probably the most important part of the decision-making is the financial side, including all possible costs, revenues, potential cost savings, expected revenues from new upgrades, etc. Annex B, Table B2 shows the results of the influence of financial characteristics on the decision-making process to roll-out new networks and to upgrade existing ones. The respondents are unanimous in stating that expected revenues are the most important variable within the decision-making process. 59% of all respondents state that the influence of expected revenues is decisive. Another 31% ascribes large influence to this variable. After expected revenues, the expected CAPEX is the most important variable. Almost half of the respondents are of the opinion that expected CAPEX also has a decisive influence on the decision how to roll-out or upgrade a network. All other financial variables show high percentages for major and decisive influence, from which we can conclude that all these variables will play an important role in the business case of the possible migration paths. There is only one variable that shows deviating results: impact of stakeholders. The high standard deviation indicates that opinions are divided here. While 34% of the respondents rated its influence as high or decisive, another 21% argued that stakeholders have little or no influence on the decision-making process. This can be explained by the fact that not all telecom parties are listed at the stock exchange and for that reason have to deal with stakeholders. Organisational factors

Next to technological, financial, policy and market-related characteristics we assume that organisational issues also have a major impact. Basically, we are interested in resources, capabilities and the possible consequences for internal processes that may hinder the roll-out of broadband. With this group of items, the central question is: Which organisational factors play a role in the decision to roll out new broadband networks or to upgrade existing networks? The quantitative descriptive results of this question are presented in Annex B, Table B3.

249

The organisational factor that has the biggest impact on the decision-making process is long-term strategy. 80% of all respondents indicate that this factor has high or decisive influence (resp. 53% and 28%) on the decision-making process. Respondents are also most unanimous about this impact, which is indicated by the lowest standard deviation (.737). What is remarkable is the equally high decisive impact (28% of respondents) attributed to regulation by national regulatory authorities, in combination with the clear disagreement on this impact, indicated by the highest standard deviation. Several variables are seen as influencing, but at the same time respondents have different opinions. With regard to these factors, the percentage of respondents indicating that the variable has no/little influence is almost equal to the percentage indicating that the variable has a large/decisive influence on the decision-making process. This concerns the following variables: (effect on) internal business processes (26% vs. 27%), company size (36% vs.35%), power of holding/mother company (32% vs.38%) and availability of human resources (29% vs.32%). Finally, it is noticeably that 43% of the sample holds the opinion that participation in the telecom market by (local) governments has no or only little impact on the decision-making process. Viability and outcomes of broadband roll-out

After having described the factors that affect the decision-making process (incoming variables), we now look at the viability and outcomes of broadband roll-out (outgoing variables).

Viability of broadband roll-out What are the factors that really determine the viability of broadband roll-out? Four factors clearly have the highest impact on the viability. Critical mass of consumers as well as that of ‘real’ broadband services has a major impact, at 70% and 62% respectively. Also, competitive pricing is considered highly important. What is striking is the importance that is attributed to the broadband offering by competitors. Apparently, the offering by competitors (probably relative to the own offer) determines the viability of the own roll-out. Also, the highest level of agreement can be seen with regard to this factor, as indicated by the smallest standard deviation. Most controversial is the impact of the brand name, as indicated by the highest standard deviation. Third party end-user service promotion and conversion rate have a small impact on viability. The numeric descriptive results are presented in Annex B, Table B4. Outcomes of broadband roll-out Finally, we discuss the importance of the effects of network upgrade and broadband roll-out. With respect to the tangible and intangible benefits and/or (social) effects that are a result of broadband roll-out/network upgrade, it is clear that economic benefits play a crucial role, directly or indirectly, as mediated by expanding customer base, a good image and customer

250

satisfaction. The strategic advantage over competitors as a direct result is also a very important effect. Know-how, experience and realising a next step are considered the least relevant. Factors related to realising the next step or acceleration in roll-out take a middle position. Underlying figures are presented in Annex B, Table B5.

Analysis: decision-making environment is multidimensional and highly complex

Based on our quantitative results, we can conclude that operators have to incorporate many factors from several dimensions into their decision-making process. This makes their decision-making processes very complex. This complexity is, moreover, increased due to the fact that all factors are constantly changing as a result of fast developments in the market. Measuring the impact of technological, organisational, financial and market-related factors on the decision-making process yields the following results. From all emerging technologies, new copper and fiber technologies have the largest influence on the decision-making process. New wireless optical technologies like FSO have the smallest impact. Transmission speed is the most important characteristic with regard to choosing a network technology. Also, future-proofness, reach and technological scalability have a very strong impact on technology choice. Considering market-related factors, density, as presented by the number of - potential - customers per square kilometre, has a decisive impact on the decision-making process. Other important market-related factors are the offering by competitors and expected and actual market demand. The financial impact end-users have is clearly the most important effect of demand and adoption. Impact on the payback time of investments and impact on the revenues are considered highly important. Demand and adoption are directly related to choosing an evolutionary path strategy (invest-as-you-grow model). This means that operators gear their roll-out strategy to the level of demand and choose flexibility in their roll-out strategy to roll-out parallel to the demand curve. The organisational factor mostly affecting the decision-making process is long-term strategy. Dissention between respondents is noticeable on the impact of regulation and the provisioning of broadband by local governments. Expected revenues and expected CAPEX are clearly the most important financial variables within decision-making. Determining factors for network viability are critical mass (of consumers and broadband services), competitive pricing and the broadband offering by competitors. Apparently, the offering and prices of competitors (probably relatively to the own offer) determine the viability of the own roll-out. As a result, the strategic advantage over competitors as a direct result is seen as a very important effect of broadband roll-out. Moreover, economic benefits play a crucial role, directly or

251

indirectly, as mediated by expanding customer base, a good image and customer satisfaction.

7.3.2 Risk and uncertainty

A relevant issue in the decision-making process concerning broadband roll-out is risk and uncertainty and the question as to how to deal with these issues. In Chapter 4, we introduced a theoretical framework to categorise risk and uncertainty based on theories by Meijer et al. (2006), Walker et al. (2003) and Courtney et al. (1997). We saw that risk and uncertainty can be categorised into source, dimension and scale. In our quantitative research, we measured two categorisation aspects: dimension and scale. In this paragraph, we first describe the division of risk and uncertainty in the dimension of ‘nature’, measured by two scales: knowledge or epistemic and variability risk & uncertainty. Secondly, we discuss the assessment of the Level of unpredictability of several types of risk and uncertainty. Finally we discuss several methods that can be applied to reduce risk and uncertainty and the preference of decision-makers and experts with regard to these methods. Nature of risk and uncertainty Risks and uncertainty can exist due to imperfect knowledge (which can be reduced by gaining more knowledge on the subject) or they can be inherent in the environment (which cannot be reduced by gaining more knowledge about the subject) (see also Chapter 4). We measured this by asking respondents to classify the various types of risk and uncertainty into two groups: uncertainties that are due to imperfect knowledge (knowledge uncertainty) and those that are inherent in the environment (variability uncertainty). It is striking that there is a great deal of disagreement between respondents on the categorisation of risk and uncertainty based on their nature. They agree most that changes in regulation are an uncertainty due to changes in the environment and, for that reason, cannot be reduced by more knowledge. The most convincing scores on categories of risk and uncertainty that can be solved by gaining more knowledge are related to technological and operational network-related aspects, including supplies and the availability of financial resources: a proper functioning of intelligence within the network, manageability of QoS of service delivery and network, Quality of necessary supplies, Impact of new broadband networks on operational systems and Availability of financial resources. Gaining more information on these subjects will reduce the uncertainty associated with them.

252

Market-related issues, like general market developments, developments in end-user services, end-user demand and regulation, are the risks and uncertainties that are least likely to be solved by gaining more knowledge: Changes in regulation, Development of broadband end-user services, Market developments, Unclear/inconsistent regulation, Development of demand. Table 7.9 summarises the five most important types of uncertainties and the percentage of respondents agreeing on these categorisations for the two dimensions of nature: epistemic and variability uncertainty. See Table B6 in Annex B for detailed descriptive numbers.

Table 7.9 : Five most important types of risk and uncertainty per type of nature and

percentage of respondents, agreeing on categorisation

Epistemic uncertainty % resp Variability uncertainty % resp

Proper functioning of intelligence

within the network 69% Changes in regulation 77%

Manageability of QoS of service

delivery and network 69%

Development of broadband

end-user services 75%

Quality of necessary supplies 67% Market developments 75%

Impact of new broadband

networks on operational systems 62%

Unclear/inconsistent

regulation 71%

Availability of financial resources 59% Development of demand 68%

Predictability of risk and uncertainty A second aspect of risk and uncertainty is their level of predictability. Risks and uncertainties can be predictable at several levels. Measuring categorisation with regard to level was done by asking respondents to indicate for every type of risk or uncertainty how predictable the future of this type is on a four point scale (4 levels: future is predictable; a limited number of possible future scenarios; a broad range of possible future scenarios; future is completely unpredictable). Here, it becomes clear that the level of unpredictability is the highest for end-user-related aspects, like consumer preferences and characteristics, end-user service development and the development of demand. We can conclude that the behaviour of regulators as well as end-users clearly are uncertainties that cannot be reduced by gaining more knowledge on their behaviour, but regulators are more predictable than end-users in their comings and goings. 10% of the respondents still hold the opinion that, with regard to changes and inconsistency in regulation, the future is completely unpredictable, which can have a restricting influence on network upgrades. Annex B, Table B7 provides more insight into the underlying quantitative data results.

253

Dealing with risk and uncertainty The question that logically follows after measuring the level of risk and uncertainty is how to handle and minimise these risks and uncertainties. As described in Chapter 4, several methods can be applied to reduce risk and uncertainty. These methods can be used to make the future more predictable for the types of risk and uncertainty that are not completely due to characteristics of the environment and thus to some extent lie within the span of control of decision-makers. Scenario analysis is seen as most suitable method. Cost-benefit analysis, forecasting and technological analysis are also considered very suitable methods to reduce risk and uncertainty. Obtaining information from other parties, like suppliers and competitors, is considered less suitable. The results are presented in Table 7.10.

Table 7.10: Preferences for methods to reduce risk & uncertainty

% of respondents, N=87

Not

suitable

at all

Not very

suitable

Some

what

suitable

Mostly

suitable

Very

suitable Mean

Std

dev

Scenario analysis 1.1 1.1 21.8 44.8 31.0 4.03 .828

Cost-benefit analysis 8.0 28.7 42.5 20.7 3.76 .876

Demand & revenue

forecasting 5.7 32.2 43.7 18.4 3.75 .824

Technological analysis 1.1 9.2 39.1 40.2 10.3 3.49 .847

Benchmarks with

competitors 14.9 36.8 33.3 14.9 3.48 .926

Obtaining information from

suppliers 9.2 54.0 31.0 5.7 3.33 .726

Analysis: the behaviour of end-users and regulators is highly unpredictable

Although there is considerable disagreement between respondents on the categorisation of risk and uncertainty into nature, risks and uncertainties related to technological and operational network-related aspects are the most clear epistemic or knowledge uncertainties. Uncertainties regarding supplies and the availability of financial resources are expected to be reducible by gaining more knowledge on the concerned issues. Aspects related to end-user behaviour, like consumer preferences and characteristics, end-user service development and development of demand, are considered highly unpredictable. The behaviour of regulators as well as end-users are clearly uncertainties that are considered being not reducible by gaining more knowledge, but as uncertainties that are inherent in the environment (variability uncertainties). Regulators are, however, more predictable than end-users in their comings and goings. From the fact that the development of broadband end-user services is so unpredictable the famous ‘chicken-egg’ problem

254

of the broadband market (see also Chapter 1) could partly be explained. Because infrastructure providers do not know whether there will be sufficient bandwidth-demanding services to make their networks profitable, they do not roll-out high-speed networks on a large scale.

7.3.3 The process-related side of network upgrades

After discussing the content-related side of our quantitative validation, we now focus on the process-related side. We will discuss the preference for several financial assessment methods that are applied to support the decision-making process involving several available technological options. Moreover, we provide more insight into the network upgrade strategy (evolutionary or revolutionary) being followed. Finally, we discuss factors that affect the decision to restart the process of roll-out that was earlier delayed due to environmental circumstances, like disappointing revenues and unexpected costs. Preference for financial assessment methods

In the content-related part of this chapter, we discussed several issues that influence the decision-making process to invest in a certain migration path to upgrade or roll-out a network. Another important issue that influences the decision whether or not to invest is the investment assessment method being used to support the decision to roll-out a new broadband network or to upgrade the existing network. In Chapter 4, several investment assessment methods have been described in detail. Cost-benefit analysis, Return on Investment (ROI) and the Net Present Value (NPV) method are considered to be mostly or very suitable, where Cost-benefit analysis clearly is most preferred (33%: very suitable). The real options method is apparently a less familiar tool and opinions about the suitability of this method are highly divided. 31% of the respondents hold the opinion that the ROA method is not very suitable or not suitable at all, whereas 37% consider ROA mostly or very suitable. The fact that people do not consider real options suitable may be due to a lack of knowledge of this method. The fact that the neutral category is the modus is typical. See Table 7.11 for detailed results.

255

Table 7.11 Preferences for financial assessment methods

% of respondents, N=87

Not

suitable

at all

Not very

suitable

Some

what

suitable

Mostly

suitable

Very

suitable Mean Std

dev

Cost-benefit analysis

(assessment of CAPEX

(Capital expenditures) and

OPEX (operational

expenses) and revenues)

1.2 7.1 17.6 41.2 32.9 3.98 .951

Net Present Value

approach (NPV) 8.5 32.9 34.1 24.4 3.74 .927

Return on Investment

method (ROI) 2.4 8.4 10.1 34.9 24.1 3.70 1.009

Real Options Analysis

approach (ROA) 13.2 17.6 32.4 25.0 11.8 3.04 1.202

Besides the assessment methods mentioned in Table 7.11, respondents could opt in favour of other, additional methods. Several assessment methods were mentioned in the category ‘other methods’. Cash flow analysis methods, including DCF (discounted cash flow) and overall cash development were mentioned four times as a preferred method. Sensitivity analysis was mentioned three times. This covers, on the one hand, sensitivity to critical success factors and their accuracy when forecasted and, on the other hand, EBIT & Net Income, Margin (before and after network cost) and OFCF (operating free cash flow). Risk analysis, scenario analysis and what if's (which approximates real options approach) were mentioned twice. Other methods, like annuity (constant payment methods), decision tree analysis and payback period Gap analysis (analysing the gap between financial results - in case of negative NPV - and strategic benefits), were mentioned once. Also, combinations of methods are preferred like a real options approach on top of an IRR valuation. As one respondent commented: “All available net revenue and

options forecasting methods will be used and compared due to the size of the required

investments.” Evolutionary versus revolutionary network upgrade strategies

In their roll-out strategy, operators can choose between evolutionary and revolutionary strategies. A revolutionary upgrade strategy refers to large-scale network adjustments in which operators change their physical networks at once to the most future-proof network, which comes together with drastic network adjustments. These revolutionary adjustments demand very high upfront investments and long payback times. An evolutionary strategy refers to smaller upgrades, mostly implying adjustments in local or central offices. These adjustments demand fewer high upfront investments and have, as a result, shorter

256

payback times. Taking technological, financial, organisational factors as well as risk and uncertainty issues into consideration, a large majority of 70% would opt in favour of or advise others to adopt an evolutionary roll-out approach, which is characterised by small incremental steps and an upgrade strategy that follows the demand curve (see Figure 7.4).

70%

30%

0% 20% 40% 60% 80%

Evolutionary upgrade/ roll out strategy

Revolutionary upgrade/ roll out strategy

Figure 7.4 Preference for revolutionary or evolutionary roll-out strategy

Preferred technology for investments Although a majority prefers an evolutionary strategy over a revolutionary approach, the question regarding the most preferred technology to invest in shows interesting results. Most respondents indicate that they would (advise others to) invest in new optical fiber technology, which is striking, because this would imply a revolutionary roll-out strategy. This result contradicts the results of the previous question, in which the majority was opting in favour of evolutionary network roll-out. Although 70% opts in favour of an evolutionary approach, 49% of this group would choose to invest in a revolutionary network technology, namely extensions of new optical fiber. 51% would opt in favour of an evolutionary technology, which could be copper, coax or radio. From the people who indicated that they would go for the revolutionary roll-out strategy, a majority of 61% remarkably enough would advise others to invest in conventional technologies, which are characterised by evolutionary upgrades, against 39% which would actually invest in optical fiber. Figure 7.5 presents the preferred technologies to invest in and Table 7.12 gives insight into preferable upgrade strategies and advised technologies in which to invest.

33%

6%

41%

9%

0% 10% 20% 30% 40% 50%

New copper based technologies (e.g. xDSL, etherloop, Ethernet first mile over copper (EFMC))

New coax based technologies (e.g. Narad networks, Ethernet over coax)

New optical fiber technologies (e.g. EPON, APON, Gigabit Ethernet)

New radio based technologies, (e.g. WIMAX, WIBRO, microwave or millimetre wave distribution systems)

257

Figure 7.5: Preferred technology to invest in

Table 7.12 : Chosen upgrade strategy versus preferred technology to invest in

New

copper New coax

New

fiber

New

radio

Evolutionary 33% 7% 49% 11%

Revolutionary 48% 4% 39% 9%

In this case, it is useful to consider the comments provided by the respondents regarding this question. The respondents indicated that their advice to invest in a certain technology depends on the specific situation. Fiber optic is considered the most future-proof at the moment and it is probably for that reason that most respondents favour it. Other factors that must be taken in consideration are the installed base and market focus, the specific business case and population density: the lower that density, the more radio is to be involved. Upgrade scenarios could also be a mix of various options. Because of the fact that optical fiber networks are now most future-proof and that all operators are working towards a fiber network could explain the lack of coherence between the upgrade strategy and the preferred technology in which to invest. Another plausible explanation for this contradiction is the fact that the question did not mention a time path for realising the roll-out of this technology, which means that an evolutionary roll-out strategy towards fiber is not excluded. Willingness to invest 41% of the respondents are not willing or able to spend more than 25% of their free available budget on the chosen technology. There is considerable disagreement with regard to the willingness to invest more than 25% in the chosen technology, as can be seen in Figure 7.6.

41%

18%

17%

16%

0% 10% 20% 30% 40% 50%

0-25%

25-50%

50-75%

75-100%

Percentage respondents

Perc

en

tag

e o

f fr

eely

availab

le b

ud

get

Figure 7.6: Percentage of freely available budget to invest/advise to invest in advised

technology

Comparing the chosen technologies and the willingness to invest in these technologies, Table 7.13 shows that people who choose the most future-proof

258

technology (optical fiber) are willing to invest the largest part of their freely available budget in this technology (45%). Willingness to invest in new copper, new coax, and wireless technologies results in weighted averages of around 30% of the freely available budget.

Table 7.13 : Willingness to invest versus preferred technology to invest in (weighted

averages)

New

copper New coax

New

fiber

New

radio

Percentage of freely

available budget to

invest in chosen

technology

33% 28% 45% 32%

Analysis: operators prefer evolutionary paths towards the most future-proof technology

To assess investment options in different technologies, several assessment methods can be applied (see Chapter 4). Cost-benefit analysis, Return on Investment (ROI) and the Net Present Value (NPV) method are considered to be mostly or very suitable, where Cost-benefit analysis clearly is most preferred. The method of Real Options Analysis is apparently a less familiar tool and opinions about the suitability of this method are highly divided. The fact that people do not consider real options as suitable may be due to a lack of knowledge of this method. The fact that the neutral category is the modus is typical. Taking technological, financial, organisational factors as well as risk and uncertainty issues in consideration, a large majority would opt in favour of or advise others to adopt an evolutionary roll-out approach, which is characterised by small incremental upgrade steps and an upgrade strategy that follows the demand curve. It is striking is that optical fiber is nevertheless the technology most respondents are willing to invest in. This technology is clearly considered to be most future-proof. We can conclude that people are most willing to invest their available budget in the most future-proof technology. This can be explained by the reasoning that the most future-proof technology offers the longest possible period to earn money from the network after the initial investments are made. However, our results show that most respondents in our sample prefer an evolutionary approach to realise this technology.

Broadband roll-out projects: success or delay?

259

After discussing several factors that eventually determine which broadband technology will be selected to be rolled out, the question arises when such a roll-out project can be considered successful. Success can be measured in different ways. We choose to measure success via Return on Investment (ROI) and the percentage of successful broadband connections within two specific time frames. Firstly, we discuss the ROI. We asked respondents what return on investment (between 0% and 100%) they would expect after two years, for a broadband network upgrade project to be considered successful. After considering the financial returns as a measure for success, we discuss a second measure of success: the percentage of the potential target groups that should have a connection to the upgraded broadband network after one and five years. Insight into the required ROI and connected footprint is important, because these requirements restrict operators in their possible choices of technologies. In case the ROI and connected footprint after five years should be 100%, revolutionary projects like FttH will not be possible, because initial investments and necessary operational efforts demand roll-out and payback times of at least 10 to 15 years. The higher the required ROI and connected footprint after one or five years is, the smaller the steps in a network upgrade must be to realise successful projects.

Required Return on Investment after 2 years Because of the strategic character of this question, not all respondents were able or allowed to answer it, which resulted in an N of 73. Eleven respondents answered 0% to the question on the ROI percentage making a roll-out successful after five years. Because the realistic value of this answer is very unlikely, we can conclude that these 0% answers were given from a strategic point of view. For this reason, we also removed the answer 0% from the question on ROI after two years form the same respondents, assuming that these respondents would most likely give these strategic answers to both questions. The remaining 0% answers for the question on ROI after two years were maintained, because an ROI form 0% after two years is in principle not unlikely. The results of the necessary Return on Investment after two years are shown in Figure 7.7. Our results show clear disagreement between the respondents. Fifty percent of the respondents agree that an ROI between 0 and 20% is reasonable after two years, of which 30% demands an ROI of no more than 10%. On the other hand, another 45% hold the opinion that this should be between 20 and 40%. 15% consider an ROI percentage of at least 40% necessary for success of a roll-out project. The maximum mentioned expectable ROI after two years is 80%. A clear explanation for these differences is a different strategy and accompanying time path for roll-outs. Respondents demanding only 0-20% ROI after two years follow revolutionary technology upgrades which require a payback time of 10 to 20 years. In case as much as 40% or more should be earned back after 2 years, payback times lie below five years, which points to evolutionary network upgrades. The question did not specify the character of the roll-out project,

260

which could be the explanation of this disagreement. Nevertheless a clear distinction is visible, which can likely be explained by this reasoning.

30%

20%

19%

15%

4%

5%

1%

4%

1%

0% 5% 10% 15% 20% 25% 30% 35%

0-10%

10-20%

20-30%

30-40%

40-50%

50-60%

60-70%

70-80%

>=80%

Perc

en

tag

e R

OI

aft

er

2 y

ears

Percentage respondents

Figure 7.7 Percentage Return on Investment after 2 years to consider a broadband network

upgrade project successful

Required Return on Investment after 5 years The percentage ROI after five years to consider a broadband roll-out project successful varies strongly from that of ROI after two years. Where no respondents expected an ROI percentage above 80% after two years, 30% does so after five years, while 21% wants to see their complete investment being earned back after five years. What is striking is that, on the other hand, 20% of respondents consider a return of investment between 10% and 20% sufficient for a successful project. The first conclusion that can be drawn from this high diversion between respondents is that they do not have a clear notion of what the ROI after five years should be. A second explanation is, however, more likely. This second explanation can again be found in different strategies with regard to network upgrades. In case the upgrade concerns a small incremental upgrade, a return of investment of above 80% can be considered reasonable within five years. However, in case an upgrade involves a drastic network change or revolutionary roll-out of a new infrastructure, an ROI of only 10% can be considered reasonable after five years, due to the very high upfront investments. 30% of the respondents hold the opinion that after five years between 80% and 100% of the investments should be earned back, which clearly points

261

towards evolutionary network upgrades. 25% hold the opinion that a maximum ROI between 0 and 20% is acceptable, which points towards a revolutionary upgrade strategy. Details are shown in Figure 7.8.

5%

20%

13%

9%

8%

5%

8%

1%

9%

21%

0% 5% 10% 15% 20% 25%

0-10%

10-20%

20-30%

30-40%

40-50%

50-60%

60-70%

70-80%

80-100%

100%

Percentage respondents

Pe

rce

nta

ge

RO

I a

fte

r 5

ye

ars

Figure 7.8 Percentage Return on Investment after 5 years to consider broadband network

upgrade project successful

Comparing the answers to the two questions, we see that 42% of the respondents who indicated that the ROI after two years should be between 0 and 10% indicates that a maximum ROI of 10% after five years still meets the criterion of success. 33% of this group increased the success criterion to a range between 10 and 20%. With one exception, all respondents considering a maximum of 10% ROI after two years successful do not increase this demand for success above a value of 50% for a period of five years after roll-out. This group remains very careful in their expectations regarding Return on Investment and clearly answered this question with the idea of relatively large network upgrades, which demand around 10 years of payback time. For the group respondents who required an ROI percentage of above 40% after two years, the trend is clearly seen that this percentage should according to them be 100% in case the roll-out is five years ago. These respondents clearly have more evolutionary network upgrades in mind, with payback times of five years. Details are shown in Table 7.14.

262

Table 7.14 Return On investment required 2 versus 5 years after roll-out

ROI after 5 years

0-10% 10-

20% 20-

30% 30-

40% 40-

50% 50-

60% 60-

70% 70-

80% 80-

90% 90-

100% 100% 0-10% 42% 33% 14% 6% 3% 3% 10-20% 7% 14% 29% 21% 7% 14% 7% 20-30% 7% 21% 7% 21% 14% 7% 21% 30-40% 9% 9% 9% 9% 9% 9% 9% 36% 40-50% 33% 67% 50-60% 100% 60-70% 100% 70-80% 100%

RO

I afte

r 2

year

s 80-90% 100%

Required connected end-users after 1 year Figure 7.9 shows the results of the question “What percentage of the potential target groups should have a connection to the

upgraded broadband network after the first year in order to consider a broadband

upgrade project successful?” Respondents do not show much disagreement on this subject. The largest part of respondents (72%) holds the opinion that coverage up to 30% is sufficient after the first year to consider the roll-out successful. Eighteen percent demands that only between 0 and 10% of the footprint should be connected after the first year, which indicates a roll-out time of at least 10 years. Operational work and availability of human resources are limiting factors, mainly in large infrastructure roll-out projects.

263

18%

29%

24%

8%

9%

1%

8%

3%

3%

0% 5% 10% 15% 20% 25% 30% 35%

0-10%

10-20%

20-30%

30-40%

40%

50%

60%

70-80%

90-100%

Percentage respondents

Perc

en

tag

e b

road

ban

d c

on

necti

on

s a

fter

1st

year

Figure 7.9 Percentage connections after 1st year to the upgraded broadband network to

consider a broadband upgrade project successful (N=80)

Required connected end-users after 5 years The results of the question as to which percentage of the footprint should have a broadband connection after five years in order to speak of a successful project are shown in Figure 7.10. We see a large mediocre group of 24%. 46% of respondents share the opinion that at least 60% of the footprint should be connected to consider the roll-out project successful. Twenty percent of them demands coverage of over 80%. It is likely that these respondents answered the question from an evolutionary point of view. On the other hand, 34% considers a 40% connection a satisficing outcome after five years. In case a roll-out of less than 40% is considered acceptable, this indicates that a roll-out towards the entire footprint of at least 12.5 years is still considered a successful broadband roll-out project, which clearly indicates a more revolutionary technological upgrade.

264

0%

4%

8%

14%

8%

24%

16%

6%

9%

11%

0% 5% 10% 15% 20% 25% 30%

0-10%

10-20%

20-30%

30-40%

40%

50%

60-70%

70-80%

80-90%

90-100%

Percentage respondents

Pe

rce

nta

ge

bro

ad

ba

nd

co

nn

ec

tio

ns

aft

er

5 y

ea

rs

Figure 7.10 Percentage connections after 5 years to the upgraded broadband network to

consider a broadband upgrade project successful (N=80)

Comparing the results after one and five years does not provide greater clarity. Comparing the answers given for the connection percentages after one and five years, we see a general shift among respondents towards a higher required percentage after five years as compared to the figures for one year. Respondents are, however, still highly divided on how high this connected percentage should be after five years. With regard to the group of respondents demanding the lowest percentage of connected end-users (0-10%), however, clear parallels are visible with the figures for required connection percentages: the 36% of respondents demanding only a percentage of below 10% of connected end-users after one year still demand this low percentage after five years. 77% of this group of respondents do not increase this demand for connection percentage above 50% after five years. It is assumable that this points to a revolutionary upgrade scenario. Bivariate correlation between connections after one and five years is two-tailed significant (p=.000). Pearson correlation is significant at the 0.01 level (2-tailed). Testing the correlations between both criteria for success for two time periods shows results as expected. There is a significant coherence between the number of connections after one year and the Return on Investment after two years. The same goes for the number of connections and the Return on Investment after five years. This appears to make sense because returns directly depend from the number of connections.

265

Reconsidering a decision to delay a network upgrade

In case roll-out projects do not turn out to be successful, in other words, in case the required ROI and number of connections are not realised, it can occur that operators decide to delay the upgrade process. Besides disappointing revenues (too low ROI due to too little end-user subscriptions) also other factors like technological complications, sudden developments in the market or in the behaviour of competitors, a reduction in the available budget or unexpected costs can make operators decide to (temporarily) stop the upgrade or roll-out project. In case a decision for upgrading or roll-out has been delayed at a certain moment, a number of issues can play a role in reconsidering this decision and continue with the roll-out project after some time. We asked respondents to indicate to what degree several factors impact reconsidering an earlier taken decision to delay initially started network upgrades. Increasing customer demand clearly is an important driver for reconsidering a prior decision to delay upgrade or roll-out. Other high-impact drivers for reviewing a previous delay of roll-out are increasing revenues, technological innovations and internal business strategy. 21% of all respondents think of internal business strategy as having decisive impact on reconsidering a previous decision to delay network upgrades. Remarkable is the high importance of service development. A potential breakthrough of a killer broadband service as well as innovations in high bandwidth requiring broadband services have high to decisive impact on restarting roll-out after a previous decision to delay. Changes in the behaviour of government and suppliers have small impact on reconsidering the roll-out decision. See Table B8 in Annex B for the underlying quantitative results.

Analysis: assumed success of broadband projects depends on following an evolutionary

or revolutionary strategy

Success of initiated broadband projects can be measured via several variables. In this research the realised Return on Investment and the percentage connected end-users at two moments in time were chosen. Clear differences are noticeable between the required ROI and number of connections between revolutionary and evolutionary upgrade strategies. For this reason, high disagreement between respondents is visible. In case of evolutionary upgrades, the Return on Investment should be between 20% and 40%. After five years, this number is expected to have increased to a value between 70% and 100%. In case of revolutionary upgrades the criteria are less demanding. After two years, the Return on Investment is expected to lie between 0% and 20%. After five years, still no more than 30% of the initial investment is expected to be earned back in case of the revolutionary scenario. With regard to the required percentage of connected end-users, we see that within

266

the evolutionary scenario the connected percentage should be above the 30% after the first year and between 80% and 100% after five years. In case of revolutionary roll-out, after one year it is required that between 0% and 20% of the footprint is connected. In these cases a roll-out towards 40% of the end-users is sufficient after five years. The decision to delay an already initiated roll-out can be made based on several considerations like, for example, disappointing end-user demand, technological complications, a reduction in available budget or unexpected costs. Changes in the environment can make decision-makers reconsider this initial decision and make them decide to restart the roll-out. Increasing customer demand, a potential breakthrough of a killer application and innovations in broadband services have high to decisive impact on restarting roll-out after a previous decision to delay. This high importance of service development is important, because we concluded earlier that the development of broadband end-user services is a source of high levels of uncertainty.

7.4 Towards a parsimonious explanatory model for

broadband roll-out

In the previous paragraph, we have described the results of our exploratory analysis. As a next step, we now describe the development and results of our explanatory research based on confirmatory factor analysis. To validate our scales as developed with exploratory factor analysis (see paragraph 7.2.2), we conducted a confirmatory factor analysis. Due to our relatively small sample, it is impossible to include all scales in confirmatory factor analysis. For that reason, only the scales (see Table 7.5) with a relatively high value of Cronbach’s alpha and/or with relevant theoretical explanatory value are included in the measurement model for confirmatory factor analysis. The measurement model was built with the following scales (see Table 7.5): Technological 1: Technological characteristics; Market_1:Demand; Market_2:Service development; Organisational_2: Internal processes; Organisational_3:Vision & Strategy; Financial_1:Costs & Revenues; Viability_1:Marketing & Services; Viability_2:Offer of competitors; Viability_3:Critical mass; Outcomes_1: Broadband roll-out; Outcomes_2: Social & Cultural Effects; Outcomes_3:Customer value. See for a detailed construction of these scales Tables C1-C6 in Annex C. Refining the measurement model

The initial measurement model did not fit with p=.000 and χ2 (968) =1399.447. For model refinement we used an imputed dataset using expectation maximisation in SPSS 14.0. Because of a Heywood case, the variance of the error terms of the

267

items Vision_1 and Vision_2 were constrained to 0.005 (Bagozzi and Yi, 1988). Because this still did not solve model problems and because item Vision_2 causes very high modification indices, it is expected that deleting this item will improve model fit. However, because a minimum of 2 items is required to load on a measure, we had to delete the entire measure ‘vision_entrepreneurship’ to test whether model fit will improve. After deleting the item, problems related to Heywood cases were solved. Based on modification indices and standard regression weights, items were subsequently removed from the model presented in Table 7.15.

Table 7.15: Refining the measurement model

Item removed Df p-value

Vision & strategy

Vision_1

Vision_2

1233.245 890 .000

Financial_5 1127.443 847 .000

Customer_value_4 1051.348 805 .000

Broadband upgrade_1 981.771 764 .000

Financial_4 912.772 724 .000

Social&cultural effects _1 852.951 685 .000

Social&cultural effects_5 792.669 647 .000

Customer_value_1 753.490 610 .000

Technological_3 698.423 574 .000

Financial_1 663.348 539 .000

Technological_2 620.958 505 .000

Marketing&services_6 573.599 472 .001

Marketing&services _3 526.314 440 .003

Marketing&services _1 467.967 409 .023

Demand_3 427.956 379 .042

Marketing&services _2 396.738 350 .043

Offer competitors

Competition_1

Competition_2

344.002 305 .061

Robustness Testing the null hypothesis that the model is correct, Bollen-Stine bootstrap results in a value for p of .736, which indicates the overall model is robust.

Tests for normality and outliers The data set contains no outliers and most of the variables are non-significantly divergent from normality.

268

Convergent validity The items that are indicators of a specific construct should converge or share a high proportion of variance in common, known as convergence validity. It can be estimated by factor loadings, variance extracted and construct reliability (Hair et al., 2006). The standardised factor loadings should be above .5, which is the case for all items except one, which a factor loading just below the threshold (0.46). Deleting this item does, however, not improve model fit. For this reason, the item has not been deleted from the model. Variance extracted of some constructs is somewhat below the .5 benchmark (Fornell & Larcker, 1981). Construct reliability for all constructs is, however, reliable, indicated by their composite reliability values exceeding the .6 benchmark (Hair et al., 2006). Financial- and organisational factors and critical mass have the highest construct reliability (>.90).

Model fit Finally, we refit the measurement model with the original dataset (with missing values). Running the model with this dataset results in a lower model fit. The resulting p value of .028 indicates that the model fit is no longer acceptable. Given the small sample size (87), the number of constructs needs to stay limited and requires choices with regard to the concepts to be included in our model. Deleting the only item with a factor loading of below .5 leads to an even worse model fit. For that reason we decided to delete one of the factors with a variance extracted below the .5 benchmark. Four constructs are eligible for removal: technological factors (VE= 0.43); Expected and actual end-user demand (VE= 0.45); social and cultural effects (VE=0.48); and marketing & services (VE=0.47). Deleting the construct of marketing & services results in a striking increase in model fit from 0.028 to 0.090. Deleting the other constructs results in lower gains in model fit. For that reason, we deleted the construct Marketing & services from the model and again computed the variance extracted and composite reliability. After deleting the construct, the model shows acceptable fit: χ2(263) = 294.201; p= .090; TLI= .942; CFI= .956; RMSEA= .037; LO90 (RMSEA) = .000; HI90 (RMSEA) = .058. An overview of the remaining items of our measurement model, their standard factor loadings, variance extracted and composite reliability is presented in Table C7, Annex C. The final measurement model is presented in Figure C1, Annex C.

Towards a parsimonious model for decision-making with regard to broadband roll-out

The initial conceptual model presented in Chapter 3 was a result of Meta-analysis, content analysis and network analysis. This conceptual model provides us initial insight into clusters of related variables, such as uncertainty around market developments (risks), cost-related issues around the technological configuration, the density of the network and expected revenues. This model is, however, purely theoretical and had to be validated with empirical results. A next step was to

269

validate the conceptual model by qualitative data analysis via in-depth interviews with strategic decision-makers in Dutch telecom and cable companies. The second model presented in this research (Figure 6.12) provides a more structured insight into the complexity of several technological, market-related and financial aspects. Qualitative data analysis has shown to be insufficient to realise the desired level of parsimony. In both models, however, many detailed hypotheses could be found. These hypotheses include factors influencing the roll-out of broadband and the interrelationship between these factors, outcomes of broadband roll-out and theory on diffusion of innovations (critical mass). On an academically higher level of abstraction, these detailed hypotheses can be encapsulated into the following four main hypotheses: Hypothesis 1: The complex interrelation of technological, process, financial and market-

related factors attributes to an evolutionary instead of a revolutionary roll-out of

broadband.

Hypothesis 2: An increase in end-user demand will lead to a critical mass that will

motivate network operators to invest in broadband roll-out. When this critical mass is

reached, more end-user services will be developed and become available in the market.

Hypothesis 3: When broadband is rolled out this will bring about social as well as cultural

effects

Hypothesis 4: When broadband is rolled out this will result in an increase in customer

satisfaction and the development of end-user services.

To test these hypotheses and the model translating these hypotheses in causal relationships, we apply structural equation modelling using Amos 7.0. We now discuss the results of our analysis. Results

The a priori model presented in Figure C1, Annex C has an acceptable model fit (χ2 (277) =299.994; p=.146. No suggestions were given for adding items based on modification indices. In favour of parsimony, all non significant paths will be deleted from the model. Based on standard regression weights several paths were sequentially deleted. The resulting paths are all significant based on their standardised regression weights (< .05). We fixed the estimates of the errors to the endogenous constructs and one of the loadings for each latent variable.

Normality and outliers The data set contains no outliers, based on Mahalanobis d-squared (all values meet the criterion of p2 >.001). In terms of normality, critical ratio for skewness was found acceptable for most variables.

270

Robustness A non-parametric bootstrap was performed with 2000 iterations. Bootstrapping indicates a robust overall model fit, as the p-value for the Bollin-Stinen statistic equals .760. Bias-corrected 95% confidence intervals indicate that all coefficients in the model are robust, which is a good result for our small sample.

Model fit Running the model with its original dataset results in a somewhat worse

model fit. But the model fit still is acceptable (χ2 (287) = 313.995, P=.123, CFI=.961; TLI=.952;RMSEA =.034). The explained variance of the endogenous constructs is reasonable. The explained variance of the variable ‘progress broadband roll-out’ is somewhat low (13%), which can be attributed to an also somewhat low path coefficient (.36) compared to the other coefficients.

Our final model is presented in Figure 7.11. In the model the explained variances and path coefficients are also made visible.

Technological factors

actual & expecteddemand

R2=.68

Availability & developmentbroadband services

Internal processes

Financial factors

R2=.12

Broadbandupgrade & rollout

R2=.33

Social & cultural effects

R2=.35

Customer value

R2=.30

Critical mass

.68***

.35*

.28*

.75***.55***

.46***

.31***.51***

.58***

.31*

.41**

.41***

.28***

Figure 7.11: structural model (measurement model and error terms omitted)

* p<.05; **p<.01; ***p<.001

Acceptance of main hypotheses Paths with acceptable standardised regression weights (p <.05), are considered to be significant. As a result, the hypotheses, corresponding with significant paths between variables can be accepted. Based on our findings we will now discuss the acceptance of our hypotheses. Hypothesis 1 In the model of Figure 7.11 a clear distinction is visible between variables that impact broadband upgrade and roll-out and variables which are results of this roll-out. At the left side of the model four variables, i.e. technological factors, internal processes, actual and expected demand and financial factors are mutually interdependent and interacting in a complex way. Finally, intermediated by

271

technological factors, these multi-dimensional factors affect the way broadband is rolled out. This unclear and complex relationship can be explained by the Garbage Can decision-making theory by Cohen, March & Olsen (1972), which describes a complex, interrelatedness of several streams before a decision is made (see also Chapter 5). All interrelationships have high levels of significance (< 0.001). Direct, causal relationships between separate factors and broadband roll-out are, however, not significant. Technology apparently is dominant in the choice whether or not and how to roll-out broadband, shown by the significant, direct causality between technological factors and broadband upgrade & roll-out. However, this relationship shows a somewhat lower significance (p<0.05). Based on the former we consider the assumption that there is a ‘complex interrelation of technological,

process, financial and market-related factors’ accepted and clearly proven based on our modelling. This complex interrelatedness of technological, financial, market and internal business factors is, moreover, affirmed by our qualitative results. In Table 6.2 we provided insight into the variables affecting the decision-making process on upgrade and roll-out. In these variables we have already clearly seen the four categories as distinguished in our final model. The fact that these variables are interrelated in a very complex way was, moreover, illustrated in our qualitative analysis (see Figure 6.6). Although the fact that this complex interrelation also attributes to evolutionary rather than revolutionary upgrades cannot be proved based on our model, it was shown earlier by our descriptive results (see paragraph 7.3.3). We concluded that, taking technological, financial, organisational factors as well as risk and uncertainty issues in consideration, a large majority would opt in favour of or advise an evolutionary roll-out approach, which is characterised by small incremental upgrade steps and an upgrade strategy parallel with the demand curve. Based on the former we can conclude that: “Hypothesis 1: The complex interrelation of technological, process, financial and market-

related factors attributes to an evolutionary instead of a revolutionary roll-out of

broadband” can be accepted.

Hypothesis 2 The second hypothesis is based on the Diffusion of Innovations theory, as explained in Chapter 2 (see paragraph 2.2.2). The role of the critical mass as intermediate factor between actual and expected end-user demand, and the availability and development of broadband services is clearly supported by the results of our structural equation modeling. Availability and development of services is positively impacted by the critical mass of broadband services and users. This relationship is significant at the 0.001 level. This critical mass is directly positively influenced by actual & expected end-user demand. Also, this positive, causal relationship shows a strong significance at the 0.001 level. As in hypothesis 1 we see a clear empirical evidence for a causality that is, moreover, grounded by theory. Based on this explanation we consider the following hypothesis accepted:

272

“Hypothesis 2: An increase in end-user demand will lead to a critical mass that will

motivate network operators to invest in broadband roll-out. When this critical mass is

reached, more end-user services will be developed and become available in the market”. Hypothesis 3 The importance of broadband roll-out for (inter)national governments has to do with the assumed economic and social effects. As discussed in Chapter 1, these effects have been demonstrated in a large number of studies, and they have persuaded governments to stimulate the roll-out of real broadband in the local loop via research projects and FttH pilots. Based on our quantitative results, we can, moreover, argue that there is a direct, significant and positive causal relationship between broadband upgrade and roll-out and social and cultural effects. The relationship is strong, indicated by the fact that it is significant at the 0.001 level, which means that we can accept “Hypothesis 3: When broadband is rolled

out this will bring about social as well as cultural effects.”

Hypothesis 4 Our final hypothesis includes two more outcomes of broadband roll-out: customer value and service development. Although at first sight these factors do not seem to (financially) benefit private operators directly, financial benefits can easily be derived. In a market that is characterised by a very high level of competition as the broadband market is, one of the biggest problems for these companies is how to distinguish themselves from their competitors. As a result, churn (switching from one provider to another) is one of the main problems for telecommunication companies. Because the underlying infrastructures are not visible to end-users, quality of service, image, and end-user services are determining factors for end-users to choose between several telecom providers. Bearing this in mind, end-user satisfaction and broadband services development are two key factors determining the eventual customer base (and thus revenues) for a telecom provider. The importance of these factors for telecom operators to distinguish themselves from their competitors explains the fact that, instead of a variable covering merely revenues, these variables and their causal relationship with broadband roll-out lead to significant paths in our model. This high importance of reputation was already clearly demonstrated in our qualitative analysis. Our model further shows that customer satisfaction is, moreover, increased by the availability and development of end-user services. As clearly proven by the descriptive results of our research, it is clear that economic benefits play a crucial role, directly or indirectly, as mediated by a good reputation and customer satisfaction. Based on the former we conclude that “Hypothesis 4: When broadband is

rolled out this will result in an increase in customer satisfaction and the development of

end-user services” can be accepted. Explaining non-significant paths Paths with p values higher than 0.05 for standardised regression weights were deleted from the model and underlying

273

assumptions had to be rejected. Rejection of relationships between variables can in general be explained by three possible causes. Firstly, the measurement contained errors, second there were statistical problems and third there’s a content-related problem. One main statistical related problem has risen. This problem is not specific for one of more hypotheses, but goes for our entire structural equation modelling. The problem can be found in the fact that many of our measures contained only 2 or 3 items. Considering the fact that 2 items is the absolute minimum, more items per construct would have increased construct stability. As mentioned, this relative weakness of our model is of a generic character and cannot explain the rejection of specific hypotheses. Rejection of direct relationships between internal processes, demand, financial factors and broadband upgrade & roll-out can be explained based on the decision-making theory of the Garbage Can. As explained in Chapter 5, roll-out most often is a coincidental combination of technological, financial, internal business and demand-related factors instead of a clear decision based on one single driver. The fact that all correlating relationships between these incoming factors are significant found this statement. Technological factors apparently have such strong influence (roll-out decisions have, ultimately a strongly technological character) that this direct link is significant. The fact that the relationship ‘progress in broadband roll-out leads to reaching the critical mass’ shows no significance can be found by the explanation that end-users do not directly experience broadband network upgrades. These network upgrades are not visible for them. Visible aspects for end-users are services and quality, helpdesks, etc. The fact that also a positive relationship between service development (which is a visible factor for end-users) and reaching the critical mass (must be rejected is for that reason striking. This probably can be found in a remarkable result we earlier found in our qualitative data analysis: the image of a provider is more important than the actual services, adopted by the end-users, which leads to reaching the critical mass! The relationship between customer value (containing customer satisfaction and image) and critical mass is, however, not been tested, but really deserves to be investigated in further research. The same goes for the relationships between critical mass and social & cultural effects. This relationship is interesting to test because it could give an explanation for the rejected direct relationship between customer value and social & cultural effects. This rejection could be found in the reasoning that customer value is not an end result, but a situation (perhaps) leading to a critical mass and via this variable to social and cultural effects.

274

Although the positive relationship between the availability & development of end-user services and broadband roll-out have been clearly indicated by decision-makers in our interviews (see also Chapter 6), this rejection can be attributed to the fact that this relationship is indirect. In the end, services and service development do not directly lead to roll-out, but this relationship occurs via actual and expected demand. The relationship between these two variables has, however not been tested. Testing this relationship is recommended for further research. The rejection of the relationship ‘reaching the critical mass of users and services influences broadband upgrade & roll-out’ could be explained by the fact that this relationship already - partly- exists in the model. The relationship between demand and broadband roll-out is, inherently available in the model although not direct. The relationship is indirect and exists via the correlation with technological factors. For this reason, this rejection probably can be attributed to a statistical problem. The relationship between development and availability of end-user services and social & cultural effects is seen all over in society and for that reason the fact that this relationship is not significant is hard to explain. The development of services like social networking sites, teleworking, e-health, e-learning, digital service numbers and all kinds of peer-to-peer services are only a hand full of examples of services having clearly proven their significant effects on society. Because the questionnaire was developed from the supply perspective it is possible that this link is not significant because the focus of our questionnaire was not on social and cultural effects and their drivers and results, but on broadband roll-out. A measurement problem could be an acceptable explanation for this content wise not explainable lack of significance.

7.5 Limitations and discussion

Limitations Our model is the result of quantitative research with a relatively small sample of 87 respondents. Although the sample contains experts from a broad range of countries and includes decision-makers in the field as well as researchers and experts, our results would have been more robust with a larger sample. Probably due to our somewhat small sample, many constructs after refinement contained only 2 or 3 items, which is not a very strong basis. A larger group of respondents will also benefit construct robustness and for that reason model stability. Our model proves that broadband upgrade and roll-out, as well as the availability and development of broadband services, increase customer satisfaction. Because our research was carried out from the perspective of telecommunication companies, our sample did not contain end-users of broadband services. As a

275

result, the causal relationships we have identified have not been validated by end-users themselves. The model, as it is now, shows the operators’ perspective on this relationship. Although the underlying hypotheses are based on a theoretical background, it is recommended to validate these causal relationships with a sample that also includes end-users. Another limitation of working with a sample consisting of competing market parties is that some information may not be provided (or be accurate) for competitive reasons. An example is the – unrealistically – high number of 0% values given for expected return on investments five years after broadband roll-out (see paragraph 7.3). However, because the questions underlying our model measures were not sensitive from a competitive perspective, we may assume that these considerations are not reflected in our final model. Discussion A more general subject for discussion is the question as to what the added or predicting value is of striving for a parsimonious model for a decision-making process taking place in a complex, highly dynamic and uncertain environment. Striving for parsimony always contains a choice between parsimony and explicability. In our model development, we followed the rule of structural equation modelling that, to reach a parsimonious model, all relationships that are not significant should be deleted (unless model significance is reduced). In our qualitative analysis we saw that, although having a high level of explicability, the resulting model was far from parsimonious. We strived for a model with an ‘optimal’ balance between parsimony and explicability, although at first sight this may seem a contradiction in terms. Moreover, the 'optimal' balance between parsimony and explicability always depends on the research focus and on the personal point of view of the researcher.

7.6 Conclusions

To validate factors relating the content and process-related side of our, a quantitative research has been executed, consisting of a questionnaire among 87 decision-makers, experts and researchers in the broadband field, originating from several countries worldwide. Descriptive content-related results Measuring the impact of technological, organisational, financial and market-related factors on the decision-making process yields the following results. From all emerging technologies, new copper and fiber technologies have the largest influence on the decision-making process. New wireless optical technologies like FSO have the smallest impact. Transmission speed is the most important characteristic for choosing a network technology. Considering market-related factors, density, as presented by the number of -

276

potential - customers per square kilometre has decisive impact on the decision-making process. Other important market-related factors are the offering by competitors and expected and actual market demand. The organisational factor mostly affecting the decision-making process is long-term strategy. Expected revenues and expected CAPEX are clearly the most important financial variables within decision-making. The financial impact end-users have is clearly the most important effect of demand and adoption. Impact on the payback time of investments and impact on the revenues are considered highly important. Demand and adoption are directly related to choosing an evolutionary path strategy (invest-as-you-grow model). This means that operators adapt their roll-out strategy to the level of demand and choose flexibility in their roll-out strategy to roll-out parallel to the demand curve. With regard to the outcomes of broadband roll-out, determining factors for network viability are critical mass (of consumers and broadband services), competitive pricing and the broadband offering by competitors. The offer and prices of competitors (probably relatively to the own offer) determine the viability of the own roll-out. As a result, the strategic advantage over competitors as a direct result is seen as a very important effect of broadband roll-out. Risk and uncertainties related to technological and operational network-related aspects are expected to be reducible by gaining more knowledge on the concerned issues (epistemic or knowledge uncertainties). Aspects related to end-user behaviour are considered highly unpredictable. The behaviour of regulators as well as end-users clearly are uncertainties that are considered being not reducible by gaining more knowledge, but as uncertainties that are inherent in the environment (variability uncertainties). Regulators are, however, somewhat more predictable than end-users in their comings and goings. Descriptive process-related results To assess investment options for broadband roll-out, cost-benefit analysis, Return on Investment (ROI) and the Net Present Value (NPV) method are considered to be mostly or very suitable, where cost-benefit analysis clearly is most preferred. The real options method is apparently a less familiar tool and opinions about the suitability of this method are highly divided. Success of initiated broadband projects can be measured via several variables. In this research the realised Return on Investment and the percentage of connected end-users at two points in time were chosen. There are clear differences between the required ROI and number of connections for success in case of revolutionary and evolutionary upgrade strategies. In case broadband projects turn out not to be successful, operators can decide to delay the initial upgrade projects. Changes in the environment can make decision-makers reconsider this initial decision and make them decide to restart the roll-out. Increasing customer demand, a potential

277

breakthrough of a killer application and innovations in broadband services have a high to decisive impact on restarting roll-out after an earlier decision to delay. Explanatory results The constructs used in our measurement model are developed earlier in our pre-test. These constructs and the underlying items are based on concepts from our conceptual model resulting from extensive Meta-analysis on broadband literature and validated in our qualitative analysis of several expert interviews. Structural equation modelling has resulted in a robust, explanatory, parsimonious model for the influencing and resulting factors of broadband roll-out. With our quantitative results we have proven that the complex interrelation of technological, process, financial and market-related factors attributes to an evolutionary instead of a revolutionary roll-out of broadband. This conclusion, moreover, supports our conclusion that the Garbage Can decision-making theory (Cohen et al., 1972) can be applied to this type of decision-making. We grounded, moreover, that an increase in end-user demand will lead to a critical mass that will motivate network operators to invest in broadband roll-out. When this critical mass is reached, more end-user services will be developed and become available in the market. This conclusion once more offers clear proof of the applicability of the theory on Diffusions of Innovations. With regard to the outcomes of broadband roll-out, we can conclude, based on our model outcomes, that when broadband is rolled out, it will bring social as well as cultural effects. It will, moreover, result in an increase in customer satisfaction and the development of end-user services. As clearly proven by the descriptive results of our research, economic benefits play a crucial role, directly or indirectly, as mediated by a good image and customer satisfaction.

Starting from a situation with no clear conceptual relationships and no available theories or models to validate (referring to the ‘black hole’ in Chapter 3) this research offers a unique contribution to the broadband research field. Although our model is not perfect (see paragraph 7.5 on limitations) at least there is now a testable model which is open for validation and adjustments by other researchers. This model can function as a starting point for further research projects and provides some stability in this multidisciplinary, unpredictable and highly dynamic research field.

278

7.7 References

Bagozzi, R. P. and Yi, Y., (1988). On the Evaluation of Structural Equation Models. Journal of the Academy of Marketing Science,16 (1), 76-94.

Bouwman, H., C. Carlsson, F. J. Molina-Castillo & P. Walden (2008). Trends in Mobile Services in Finland 2004-2006- From Ringtones to Mobile Internet. INFO, 10 (2), 75-92.

Cohen, March & Olsen (1972). A Garbage Can Model of Organisational Choice. Administrative Science Quarterly, 17 (1), 1-25.

Courtney, H., J. Kirkland and P. Viguerie (1997). Strategy under uncertainty (business forecasting). Harvard business Review, 75 (6).

Fornell, C. and D. F. Larcker (1981). Evaluating Structural Equation Models with Unobservable Variables and Measurement Error. Journal of Marketing

Research, 18(1), 39-50.

Hair, Joseph F., Bill Black, Barry Babin, Rolph E. Anderson, and Ronald L. Tatham (2006). Multivariate data analysis. Upper Saddle River, NJ: Pearson Education, Inc.

López-Nicolás, C., F. J. Molina-Castillo & H. Bouwman (2008). An Assessment Of Advanced Mobile Services Acceptance: Contributions From Tam And Diffusion Theory Models. Submitted to Information & Management.

Meijer, Ineke S.M., Marko P. Hekkert, Jan Faber and Ruud E.H.M. Smits (2006). Perceived uncertainties regarding socio-technological transformations: towards a framework. International Journal of Foresight and Innovation Policy, 2 (2), 214-240.

Reuver, de Mark (2008). Capturing value from mobile business models: design issues that matter. Accepted for publication in The International Journal on

Electronic Business.

Van de Wijngaert, Lidwien (2001). Internet in Context: Fysieke en affectieve toegang, geschiktheid; vraag, aanbod en context. In H. Bouwman (Ed.), Communicatie in de Informatiesamenleving (pp. 51-70). Utrecht, the Netherlands: Lemma.

Walker, W.E., P. Harremoes, J. Rotmans, J.P. van der Sluijs, M.B.A. van Asselt, P. Janssen & M.P. Krayer von Krauss (2003). Defining Uncertainty- A Conceptual Basis for Uncertainty Management in Model-Based Decision Support. Integrated Assessment, 4(1), 5-17.

279

Chapter 8 Conclusion

Shedding light on the black hole

8.1 Introduction

In this chapter, we present the conclusions of this research. We begin by providing an overview, after which we answer the main research question by answering the intermediate research questions formulated in Chapter 1. We also discuss the limitations of this thesis and provide recommendations for further research. This chapter closes by discussing the implications of this thesis for relevant stakeholders.

8.2 Overview of the research

In this paragraph, we provide an overview of this thesis. In Chapter 1, we introduced the research area and questions. This research contains both a content-related and a process-related side. In Chapters 2 and 3, we discussed the content-related part of our theoretical framework and focused on the main drivers and outcomes of broadband roll-out in the technology, market and policy dimension. In Chapters 4 and 5, we addressed the process-related part of our theoretical framework, focusing on the decision-making process with regard to broadband roll-out in access networks and the impact of risk and uncertainty on decision-making. In Chapters 6 and 7, we focused on the validation of both sides of our theoretical framework, via qualitative and quantitative data analysis. Chapter 1 introduced our main research question and the underlying intermediate research questions, providing insight into the social and economic relevance of broadband roll-out and the national and European public initiatives aimed at stimulating the roll-out of broadband in the local loop. Notwithstanding the public attention and various resulting initiatives, the roll-out of FttH access networks in the Netherlands is still in its starting phase. A symmetrical data connection of 10 Mbps, which is how the term broadband is defined in this research, is not yet available in the Dutch residential market. The delay in roll-out of these broadband networks is considered a problem by national as well as European governments, because important economic and social benefits will not be realised. Network investments are very risky due to specific characteristics of the infrastructure, in particular the local loop, and uncertain and unpredictable end-user demand. Things become

280

even more complex due to the multi-actor, multi-disciplinary environment in which telecom operators find themselves. Existing policy initiatives do not have the expected and desired effect of private operators rolling out broadband access networks. For that reason, gaining more insight into the complex relationships between technological, financial, market and policy/regulation-related factors and the way they affect the broadband roll-out by private network operators is important. Moreover, insight into the decision-making process is needed if we are to understand how several driving factors eventually lead to final decisions. As a result, the objective of our research is to develop a scientific integrated

multidisciplinary model that provides insight into the drivers and outcomes with regard to

local loop broadband roll-out by telecom operators and to understand and explain the way

decision-making with regard to broadband roll-out evolves.

The accompanying main research question is: What content and process-related

aspects determine broadband local loop upgrades, the outcomes of broadband roll-out and

the decision-making process regarding the upgrading of broadband access networks?

Chapter 2 focused on the content-related side of this research and provided more insight into the broadband domain. We described the domain and main actors involved. There are three main dimensions in the environment in which broadband decision-makers operate: technology, market and policy & regulation. The market perspective refers to the supply side (competitors and equipment suppliers) and the demand side (end-users). The Dutch market is characterised by a strong (duopolistic) infrastructural competition between upgraded copper (ADSL) and cable (HFC) networks. Nevertheless, on both networks there are several operators providing access services to these networks to end-users. The strong competition between the copper and coaxial networks has three main causes: access regulation, the coexistence of two fixed networks with a national coverage and the convergence of data, voice and images. The technological dimension of Chapter 2 refers to technologies that can be used to upgrade existing networks to next generation broadband networks. We compared the technologies in question based on relevant criteria. Moreover, we discussed possible migration paths towards Fiber-to-the-Home (FttH) networks. The most likely development paths are the separate developments of evolutionary paths in copper TP (via VDSL) and coaxial networks towards fiber to the curb (FttC). Broadband Fixed Wireless Access, for instance Wimax, is seen as a complimentary or alternative solution to high-speed cable and DSL versions like ADSL2+ and VDSL and as a cost-effective solution in areas that are beyond the reach of DSL and cable. FWA networks like Wimax, LMDS and MMDS can, moreover, function as back-up lines in case high-speed fixed networks have become viable and rolled out or at the moment the geographical roll-out has reached the areas in which FWA access lines functioned as temporary solutions. Between 2002 and 2005, many pilots and research have

281

been executed by (local) government to stimulate FttH roll-out. Nevertheless, in the Netherlands the roll-out of fiber in the local loop clearly lags behind compared to countries like Japan and Korea. The penetration levels regarding the fiber access networks of these countries can be attributed mainly to high levels of public investment and financial incentives. In Europe, state aid is a potentially problematic issue. The policy and regulation domain has to do with (access) regulation and legislation, broadband stimulation projects and the regulation of public fiber initiatives. Chapter 3 provided insight into our analysis of existing literature on content-related factors affecting broadband roll-out decision-making. We examined available broadband literature to see whether a framework or theory for broadband roll-out is available in existing literature. In Chapter 3, we first described our applied methodology, consisting of Meta-analysis, content analysis and network analysis. Furthermore, we discussed the results of these analyses and the development of our conceptual model. As a result of network analysis we found that most of the concepts are strongly and closely related, indicated by a densely populated Euclidean space. This ‘black hole’ more or less represents the lack of coherent and focused conceptualisation. It can, as a result, be concluded that, at the moment, there is no integrated multidisciplinary scientific framework or unifying theory that describes the way technological, financial, market-related and policy factors affect the migration of broadband networks in the local loop. It became clear that there is no theory that combines complex decision-making theory with broadband technology development and innovation. We concluded that, although technological, regulatory and user-related aspects play a role, the overall emphasis in the studies we examined is highly economic in nature. We were able to present an initial conceptual model regarding broadband roll-out. Chapter 4 focused on the process-related side, providing insight into the financial character of broadband network investment-related decisions and the risk and uncertainty associated with these decisions. There are several methods available to operators for assessing the business value of corporate investment options in the local loop infrastructure. We discussed the traditional, commonly applied methods of discounted cash flow (i.e. NPV and RoR) and ratio (i.e. ROI and B/C analysis). We also provided greater insight into a method known as Real Options Analysis (ROA). Although the relevance of ROA to the decision-making process in the telecom market has been demonstrated in literature, it is a method that has not yet been applied in practice on a serious scale. Literature shows that ROA may be a more suitable assessment method for network investments than discounted cash flow methods in the case of major technological investments in situations characterised by considerable technological and market uncertainties. Due to its different approach to assessing risk and uncertainty, ROA may have a significant

282

impact on decision-making processes regarding broadband upgrades and the choice between technologies. The risks and uncertainties telecom operators face during their decision-making process concerning the upgrading and roll-out of their broadband networks have a number of causes. On the one hand, they are in intrinsic to the nature of large infrastructural projects, while on the other hand being caused by (unknown) competitor behaviour, (unknown) end-user demand, rapid technological development and different development paths available to operators. We have demonstrated that a combination of the uncertainty typologies from Meijer et al. (2006), Walker et al. (2003) and Courtney et al. (1997) provides a suitable framework for categorising risks and uncertainties based on their nature, level and source. Finally, we discussed a number of strategies telecom operators can use to reduce risks and uncertainties in their decision-making process. Chapter 5 provided greater insight into the types of decisions and the decision-making process regarding broadband roll-out. Based on theories suggested by Simon (1960) and De Baas (1998), we concluded that decisions regarding broadband roll-out are strategic (non-programmed) as well as complex in nature. This complexity shows itself in the many interdependent actors with different, conflicting interests and in the many technological, organisational, regulatory and market-related factors that play a role. Several decision-making theories are available to describe strategic decision-making processes. After comparing the characteristics of the broadband market with those mentioned in several decision-making theories, we selected two decision-making theories that appear to be suitable frameworks for our research model. The first theory is the Garbage Can model first mentioned by Cohen, March and Olsen (1972). At first sight, telecom companies fit the defining criteria of an organised anarchy to which the Garbage Can model can be expected to apply. Based on similarities between Logical Incrementalism and Real Options Analysis, we assumed that the theory of Logical Incrementalism (Quinn, 1980, Das & Teng, 1999) can also be applied to explore the decision-making processes concerning broadband network roll-out. Chapter 6 focused on the qualitative part of our validation. Our qualitative research consisted of interviews with 12 executives of major telecommunication and cable operators in the Netherlands. Chapter 6 included the validation of the content-related as well as the process-related aspects.

Process With regard to the process-related validation, we tested several theories that were introduced in Chapters 4 and 5. We concluded that the distinction between non-programmed and programmed decisions, as introduced by Simon (1960), as well as the distinction between short-term operational decisions and

283

periodic control decisions (Lin, 2004) are very much applicable to the various decisions involved in the process of broadband upgrade and roll-out. We validated financial theory by examining the applicability of financial assessment methods. We concluded that, within all companies we consulted, investments in different upgrade alternatives are assessed via the NPV method, possibly in combination with other methods. None of the companies we consulted applied Real Option Analysis (ROA). The first part of our theoretical framework surrounding risk and uncertainty was validated by measuring the applicability of the six ‘sources’ of risk and uncertainty suggested by Meijer et al. (2006). The interviews showed that broadband decision-makers did indeed recognise these six sources of uncertainty. Technological analysis, forecasting and pilots are the most commonly applied methods to reduce risks and uncertainties. We examined the applicability of two decision-making theories, i.e. the Garbage Can model (Cohen, March & Olsen. 1972) and the theory of Logical Incrementalism (Quinn, 1980; Das & Teng, 1999). Our analysis demonstrated that telecom companies can be qualified as organised anarchies in the way Cohen et al. 1972 intended in their theory. Our results also indicate that the decision-making processes concerning broadband roll-out fit the characteristics of the Garbage Can theory by Cohen et al. (1972). Throughout our interviews, various examples were provided that reflected the three characteristics of Logical Incrementalism formulated by Das & Teng (1999). It also became clear that strategic decision-making is an incremental process.

Content With regard to content validation, we measured the impact of technological, organisational, financial and market-related factors on the decision-making process. It became clear with regard to the technological dimension that IP, Ethernet and compression technologies are seen as the most important technological developments. Choices between technologies are based on a variety of characteristics of these alternatives. In the market domain, competition is a very important factor in the decision whether or not to upgrade networks. Strong interdependency between operators results in close cooperation, which is not always of a voluntary nature, between these parties. End-user demand is an important driver for upgrades, although the boost to a company’s image associated with the fact that it is able to offer high bandwidth appears to be an even more important driver. Regulation clearly affects the decision-making process. It is, however, not the deciding factor in the strategy of operators, but rather perceived as the framework within which operators have to function. As a final part of the validation of content-related factors, we presented a validation of the variables (concepts) from our conceptual model (see also Chapter 3). We demonstrated that it is not possible to develop a parsimonious model for the drivers and outcomes of broadband roll-out based on qualitative research alone.

284

Chapter 7 addressed the quantitative part of our validation of the theoretical framework. We validated the process-related aspects via descriptive results. The quantitative validation of the content-related part started from the conceptual model that resulted from the Meta-analysis (Chapter 3). The quantitative validation consisted of two online surveys: a pre-test and a main study. The main study consisted of a confirmatory factor analysis aimed at creating a measurement model, which formed the basis of a structural equation model.

Process We tested several theories in Chapter 7. We validated financial theory is validated by examining researching the applicability of financial assessment methods. Cost-benefit analysis, Return on Investment (ROI) and the Net Present Value (NPV) method were considered to be reasonably or very suitable. Cost-benefit analysis is clearly the most preferred option. The method of Real Options Analysis (ROA) is a less familiar tool and opinions about the suitability of this method were highly divided. The remaining part of our theoretical framework on risk and uncertainty (Chapter 4) was validated by measuring nature and level of various types of risk and uncertainty. Risk and uncertainties related to technological and operational network-related aspects are the most clear epistemic uncertainties (reducible by gaining knowledge). The behaviour of regulators and end-users clearly are variability uncertainties (not reducible by gaining more knowledge). Regulators are, however, more predictable than end-users in their comings and goings. Aspects related to end-user behaviour like consumer preferences and characteristics, end-user service development and development of demand are considered highly unpredictable. Content As a first step, we measured the impact of technological, organisational, financial and market-related factors on the decision-making process. The descriptive results indicated that, of all the emerging technologies, new copper and fiber technologies have the biggest influence on the decision-making process. New wireless optical technologies, like FSO, have the smallest impact. Transmission speed is the most important factor in the choice of network technology. With regard to market-related factors, density, as indicated by the number of – potential – customers per square kilometre, has a decisive impact on the decision-making process. The organisational factor that most affects the decision-making process is long-term strategy. There is clear disagreement between the respondents with regard to the impact of regulation and involvement of local governments. Expected revenues and expected CAPEX are clearly the most important financial variables in the decision-making process. Finally, we focused on developing a parsimonious model for broadband roll-out. The constructs and underlying items were based on concepts from our conceptual model and validated in our qualitative analysis of several expert interviews. The model combines the most important influencing and resulting factors of broadband

285

roll-out with adoption theory (van de Wijngaert, 2001). Structural equation modelling resulted in a robust, explanatory, parsimonious model for the influencing and resulting factors of broadband roll-out. In this chapter (Chapter 8), we provide an overview of our results and present our general concluding remarks. Using the results of our Meta-analysis, expert interviews and online surveys we answer the research questions that were presented in Chapter 1. After answering the research questions, we discuss the limitations of this research, present recommendations for further research and address the implications of this research for several stakeholders.

8.3 Answering the research questions

This paragraph contains an overview of the results of this research. Using the results of the online questionnaire and expert interviews we answer the research questions presented in Chapter 1. The individual research questions are presented with an overview of the results and concluding remarks.

8.3.1 The impact of technological, market-related and regulatory

factors on broadband roll-out

Telecom operators find themselves in a multidimensional field. Due to the behaviour of competitors, end-users, suppliers and regulators, the optimal technological solution is not always the best practical solution. Choices regarding the upgrade of access networks are the result of a complex interplay between technological, market-related and regulatory factors. This makes the outcomes of technology decisions not always predictable and understandable for policy-makers. A more effective use of policy instruments demands more insight into the way factors out of these different domains influence broadband network upgrades. Literature on these domains combined with descriptive results of our qualitative and quantitative data analysis enabled us to provide answers to the following research question:

RQ1 What are the technological, market-related and regulatory factors that play a role

with regard to the upgrading of access networks by telecom network operators towards

next generation broadband?

Technology Measuring the impact of technological, organisational, financial and market-related factors on the decision-making process yields the following results. Operators are constantly keeping up with the latest technological developments. The most important developments we found in our qualitative research are located at the protocol level of the networks and include developments in Ethernet and IP.

286

Cable operators consider the development of the Docsis45 protocol highly important. Also, the development in compression technologies is considered highly important, which appears to make sense because these technologies make it possible to transmit more and more data over the same networks. As a result, high and risky infrastructural investments can be postponed and more time is left to gain revenues from the sunk investments in the existing infrastructures. The quantitative results show that from all emerging technologies, new copper and fiber technologies have the largest influence on the decision-making process. New wireless optical technologies like FSO have the smallest impact. Market Both the quantitative and the qualitative results show that competition is a very important factor in the decision whether or not to upgrade networks. The high level of competition has brought about three important effects. Firstly, innovation in technology and services is a clear result of competition. Secondly, it resulted in a price war for broadband access with more and more bandwidth for lower prices as a result. Although also indicated as being of the utmost importance, actual and expected demand in practice do not seem to be as important as customer satisfaction and image-building as drivers for the decision whether or not to upgrade. Finally, competition has brought the Netherlands to a top position within the world in the field of broadband (i.e. cable and ADSL) roll-out. Density, as presented by the number of - otential - customers per square kilometre has a decisive impact on the decision-making process. Another important market-related factor is expected and actual market demand. The financial impact end-users have is clearly the most important effect of demand and adoption. Impact on the payback time of investments and the impact on revenues are considered highly important. Expected revenues and expected CAPEX are clearly the most important financial variables within the decision-making process. Regulation From our qualitative research, we can conclude that regulation by the Dutch national regulatory authority (NRA) OPTA clearly affects the decision-making process. Regulation is, however, not a dominant factor in the strategy of network operators. It is considered a given, a framework in which operators have to function. The fact that regulation only plays a minor role in decision-making processes compared to financial and technological considerations was demonstrated earlier in our literature analysis (see Chapter 3). In quantitative analysis, there is disagreement between the respondents with regard to the impact of regulation and the provisioning of broadband by governments. There is the least consensus about the impact of (local) governments offering broadband. Although one third of the respondents is of the opinion that public broadband offering has no or little influence, an equal 45 Data Over Cable Service Interface Specification (DOCSIS) is an international standard developed by CableLabs and contributing companies for data transmission over coaxial cable. See also chapter 2.

287

percentage of the respondents consider public broadband offers having large or even decisive influence on the decision-making process. This difference can probably be explained by the fact that broadband roll-out by local governments does not have the same impact and characteristics everywhere in the world. In some countries, they take the form of small pilots, while in other countries local governments operate as serious competitors for private telecom operators. In the latter situation, the perceived impact of public broadband offers on the private company also depends on the market position and geographical presence of the company involved as well as the public initiative. In our qualitative analysis, some clear effects of public intervention in the market were mentioned. A first effect is that public broadband initiatives aimed at stimulating fiber roll-out have contributed to bringing the Netherlands to its top position of broadband connections. Another positive result of public market intervention is that it created awareness and understanding among operators about the need to change their monopolistic attitude into a more customer-oriented one. On the other hand, governments should be conscious of the possible negative side effects of their interference in the private market. Foreign market parties indicate they are considering postponing or terminating their investments in the Dutch market due to an instable and distorted market environment. Moreover, market distortion takes place due to new players buying up networks of bankrupt broadband initiatives. A climate of political and regulatory uncertainty has an inhibitory action on operators concerning large infrastructural investments.

Variables determining technology choice

With regard to broadband upgrades in the access network, there are many available technologies from which to choose. To understand the drivers of decisions regarding broadband upgrades, it is important to understand the complexity related to these technological alternatives. This leads us to the first sub question of RQ1: a) What are important variables determining the choice between several migration paths?

The attractiveness of technologies is determined by how well they score on relevant attributes. When a technology has to be selected, there are many things that have to be taken into account. Examples are the existing outside plant, the network location, the cost of deploying the network, subscriber density and the return of investment (ROI) (Rui, 2005). Moreover, every technology has its advantages and disadvantages, and any decision will depend on what a provider needs and on the context in which the network operator has to operate. From our quantitative research, we can conclude that the decision which technology to choose is a very complex one, as becomes clear when we see that all the characteristics presented to the respondents in this question have at least

288

mediocre influence on the decision-making process. Transmission speed is the most important characteristic when it comes to selecting a network technology. Also, future-proofness, reach and technological scalability have a very strong impact. Migration paths towards next generation networks

Starting from state-of-the-art technology, several migration paths can be followed to reach (what is now considered) the most future-proof access networks of GPON. This leads us to the last sub research in this section: b) Which possible migration paths can be and are followed to reach next generation

broadband networks?

In Chapter 2, we assessed the broadband technologies ADSL2 (+), Etherloop, EFMC, Narad networks, EttH, EPON, APON, Active Optical Ethernet, LMDS, MMDS, FSO and HFR. We concluded that the most likely development paths are the separate developments of evolutionary paths in copper TP (via VDSL) and coaxial networks towards fiber to the curb (FttC). Broadband Wireless Access above 20 GHz, like Wimax, is seen as a complimentary or alternative solution to high-speed cable and DSL versions such as ADSL2+ and VDSL and as a cost-effective solution in areas beyond the reach of DSL and cable. FWA networks like Wimax, LMDS and MMDS can, moreover, function as back-up lines in case high-speed fixed networks have become viable and rolled out or at the moment the geographical roll-out has reached the areas in which FWA access lined functioned as temporary solutions.

In the Netherlands, network operators are not applying fixed wireless technologies in their networks. The possibilities of Wimax are, however, being explored. The main reason for not using wireless technologies is the fact that two fixed networks with national coverage have already been rolled out in the Netherlands. Moreover, weather conditions in the Netherlands (lots of fog and rain) are far from ideal for a broad application of FSO and microwave technologies. In the Netherlands, operators are following evolutionary paths. Cable operators show developments based on Docisis technology (Docsis 2 towards Docsis 3) and the incumbent follows the paths via ADSL 2+ and VDSL towards Active Optical Ethernet. In some pilot areas, the VDSL step is left out (see also Figure 6.2 and Figure 6.3). The qualitative as well as quantitative results show that cable operators as well as telecom operators clearly prefer evolutionary upgrades over revolutionary ones. Taking technological, financial, organisational factors as well as risk and uncertainty issues in consideration, a large majority would opt in favour of or advise an evolutionary roll-out approach, which is characterised by small incremental

289

upgrade steps and an upgrade strategy parallel with the demand curve (see Chapter 7 for details). A striking result is, however, that most respondents indicate that they would (advise others to) invest in new optical fiber technology, which is striking, because this would imply a revolutionary roll-out strategy. This contradicts the previous statement of a majority opting in favour of an evolutionary network roll-out. Although a large majority opts in favour of an evolutionary approach, half of this group would advise to invest in a revolutionary network technology, namely extensions of new optical fiber. The other half would opt in favour of an evolutionary technology, which could be copper, coax or radio. Remarkably enough, a majority of the respondents who indicated that they would go for the revolutionary roll-out strategy would advise others to invest in conventional technologies, which are characterised by evolutionary upgrades, against a minority which would actually invest in optical fiber. The decision to invest in a certain technology depends on the specific situation. At the moment, fiber optic is most future-proof, which is probably why most respondents would (advise to) invest in this technology. Other factors that must be taken in consideration are the installed base and market focus, the specific business case and the population density: the lower the density, the bigger the role played by radio. Upgrade scenarios could also be a mix of various options. Because optical fiber networks are now most future-proof and all operators are working towards a fiber network could explain the lack of coherence between the upgrade strategy and the preferred technology in which to invest. Another plausible explanation for this contradiction is the fact that the question did not mention a time path within which to realise the roll-out of the technology, which means that an evolutionary roll-out strategy towards fiber is not excluded. Development paths that are evolutionary in nature are considered more suitable given the existing market-related and economic conditions. They entail fewer technological risks because they offer infrastructure providers feedback and an opportunity to change their (technological) strategy as the infrastructure is being developed. Moreover, the very long implementation times of new networks force operators to follow an evolutionary approach, at least at a geographical level. Rolling out FttH to nine million households in the Netherlands will take at least ten years. Even when an operator would opt in favour of the revolutionary FttH roll-out, it will be an evolutionary process that takes a decade. During this period so many changes in technologies and market can be expected that it makes sense to take new developments into account along the way and to adjust the pace of roll-out to actual demand. New promising and cheaper technologies can be implemented in the network during the upgrading process. Sometimes, an evolutionary network upgrade is combined with a more revolutionary approach on the higher protocol

290

layer. Although some operators make revolutionary, high-impact adjustments on the protocol layer, their physical network roll-out continues to follow the evolutionary model. Viability, outcomse and success of broadband roll-out

Viability The descriptive results of this research show that four factors clearly have the highest impact on the viability of the roll-out of broadband access networks. Critical mass in terms of consumers as well as ‘real’ broadband services has a high impact. Also, competitive pricing is considered highly important. What is striking is the importance that is attributed to the broadband offering by competitors. Apparently, the offering by competitors (probably relatively to the own offer) demands the viability of the own roll-out. Outcomes With regard to important outcomes of broadband roll-out, it is clear that economic benefits play a crucial role. The role of economic benefits can be a direct or indirect one, as mediated by expanding customer base, a good image and customer satisfaction. The strategic advantage over competitors as a direct result is also considered a very important effect of broadband roll-out. Success It is, moreover, possible to conclude in general that, in order for a broadband roll-out project to be considered successful, an ROI percentage between 0 and 30% must be achieved after two years and coverage up to 30% is sufficient after the first year. After five years, at least 40% of the investments of a broadband roll-out project should be earned back and coverage of at least 50% of the footprint is required for success. Optical fiber is considered to be most future-proof and is the technology most respondents are willing to invest in, although our results show that the path to reach this technology is preferred to be evolutionary by a large majority of our sample.

8.3.2 Exploring available multidisciplinary frameworks on broadband

roll-out

An integrated multidisciplinary scientific framework or unifying theory, describing the way technological, economic, market-related and regulatory factors affect migration of broadband networks in the local loop, would be relevant to scientists, strategic decision-makers in broadband and regulators. Although, as described under RQ1, many technological, regulatory and market-related factors are relevant, it is unclear how these factors are interrelated. The research question that will be clarified in this section is:

RQ 2 What multidisciplinary frameworks or conceptual models on broadband roll-out are

currently available?

291

To gain insight into the existing theoretical concepts related to broadband roll-out, we carried out a systematic literature analysis, with the aim of developing a conceptual model that could be used in the research project. This conceptual model was intended to provide insight into the relationships between the factors involved in broadband evaluation paths. To arrive at this conceptual model, we used a combination of Meta-analysis, content analysis and network analysis. The exact, accurately stated question to be answered in our literature analysis was:

“What are the variables and relationships with regard to the decision-making and roll-out

process concerning migration paths towards broadband networks that were found in

earlier studies in this field?”

The starting point of the Meta-analysis was the collection of relevant research projects as published in academic peer-reviewed journals and conference proceedings in the broadband evolution domain. The final selection consisted of fifty articles for content analysis. Based on extensive literature search, we can conclude that the literature we analysed is representative of the domain of studies into broadband roll-out and evolution in the local loop. As a next step, the articles were subjected to content analysis. Although normally the focus would have been on testable hypotheses, it was interesting to see that hardly any of the papers contained clearly formulated hypotheses. For that reason, all relevant causal relationships were extracted from the papers included in our analysis. Figure 3.3 in Chapter 3 gives an impression of what a conceptual network would look like on the basis of these primary findings. This black hole more or less represents the lack of coherent and focused conceptualisation. Most of the concepts are strongly and closely related, indicated by the fact the Euclidean space is densely populated and there are only a few outliers. Based on the extensive literature analysis, we can conclude that, within existing broadband literature, no coherent high-level conceptual model that explains broadband roll-out is available. A large number of usually closely related micro-economic concepts related to specific investment methodologies like Net present Value or Internal Rate of Return are used. There is noticeable emphasis on cost-benefit approaches, focusing on the costs involved in the implementation and roll-out of broadband networks, while more general objectives, such as strategic value, innovation in technology and services, economic and social welfare, are ignored. In general, we can conclude that a middle range theory is missing.

8.3.3 Assessing investment options under risk and uncertainty

A problem related to infrastructure investments is the necessary high upfront investment, in combination with long payback periods and an uncertain market environment. These circumstances affect the investment behaviour of network

292

operators. The fact that, ultimately, it is the investment-related costs and revenues that determine the payback time of network investments - and, by extension, the annual revenues of operators – makes the financial perspective highly important. Additional insight needs to be gathered in the financial aspects of broadband investments. Also, the high-risk character of these investment-related decisions needs to be understood. This brought us to the following research question: RQ 3 How do network operators assess several investment possibilities concerning

upgrading and rolling out broadband networks in the local loop?

The qualitative research shows that the costs of networks are mainly driven by IT systems and physical costs, including digging, co-location, adjustments to homes and equipment. In all the companies we consulted, investments in different upgrade alternatives are assessed via the NPV method, possibly in combination with other methods like Return on Investment (ROI) or scenario analysis. This has been shown earlier by Keswani & Schackleton (2004), who argue that the Net Present Value rule has emerged as the dominant decision rule for investment-related decisions. The method of Real Option Analysis (ROA) is not applied by any of the companies we consulted. This is remarkable, considering the fact that it has been shown in various studies that ROA is considered more suitable for assessing investment options in uncertain environments, like communication and information technology (Oslington, 2004; Neufville, 2003, among others). There are several methods telecom operators can use to assess network investment options. Remer et al. (1995) define five basic methods to assess investment projects: Net Present Value methods, rate of return methods, ratio methods, payback methods and accounting methods. Via qualitative and quantitative data analysis, we examined the practical use of several financial assessment methods by telecom and cable operators. From the qualitative (Chapter 6) results we can conclude that Net Present Value (NPV), Return on Investment (ROI) and cost/benefit analysis (CBA) are clearly most frequently applied in practice. All interviewees indicated the NPV as applied important assessment method, whether or not in combination with other methods like ROI or scenario analysis. Moreover, all interviewees indicated that costs and benefits are made visible via cost-benefit analysis (CBA). We found the same results in our quantitative analysis (Chapter 7). Although the important role that is played by the NPV method is clearly supported by our results, they do not completely underline the conclusion by Keswani & Schackleton (2004) with regard to the applicability of the NPV method, who argued that the Net Present Value rule has emerged as the dominant method for investment-related decisions. Both the qualitative analysis and the quantitative

293

analysis show that the CBA is considered the most suitable method for assessing investment-related decisions concerning broadband roll-out. Earlier studies (Neufville, 2001; Keswani & Schackleton, 2004) indicate that traditional discounted cash flow methods are less suitable for environments that are characterised by uncertainty and a need for flexibility. Existing literature (Oslington, 2004; Neufville, 2003), moreover, states that the ROA is more suitable for assessing investment options in uncertain environments, like communication and information technology. For this reason, it is striking that both our qualitative and quantitative results indicate that this method is not applied in practice. All interviewees, with one exception, are not familiar with the method of Real Options Analysis and do not apply this method in practice for their financial assessments. The unfamiliarity with the ROA method is, moreover, proven by our qualitative results and reveals itself by the fact that opinions about the suitability of this method are highly divided. Based on our qualitative results we can conclude that this can be attributed to a lack of knowledge concerning this method. The fact that the neutral category is the modus is typical in these situations. Our results support earlier observations from Alleman (2002), who argued that, although it is a relevant method as far as telecommunications are concerned, ROA is not yet widely applied within this industry.

8.3.4 Dealing with risk and uncertainty in investment-related decisions

An important factor operators must take in consideration with regard to investing in a specific technology for their network upgrades is the level of risk and uncertainty associated with the various technological alternatives in relation to aspects like price development, expected demand, future-proofness, etc. The risks and uncertainties involved affect the chance investments will be earned back within the expected timeframe. As a result of this, risk and uncertainty assessment is an important aspect of network investment-related decisions. To be able to minimise the undesired effects of risks and uncertainties on broadband roll-out investment-related decisions, three aspects of these factors are relevant: the characteristics of risk and uncertainty themselves, their effects on broadband roll-out and methods operators can apply to assess them. This leads to the following research question: RQ 4 Which risks and uncertainties are affecting these investment-related decisions, what

are their effects on broadband roll-out and how can risks and uncertainties be assessed and

minimised?

Risks and uncertainties are inherent characteristics of large infrastructural projects. Economic, political, social and other related risk issues have been recognised as crucial criteria for investment-related decision-making (Piyatrapoomi et al., 2004). Factors causing uncertainty are the (unknown) behaviour of competitors, the

294

(unknown) development of end-user demand, the rapid development of new broadband technologies and the different development paths available to operators (Fijnvandraat & Bouwman, 2006). Telecom operators are faced with several types of risks and uncertainties during their decision-making process concerning the upgrading and roll-out of their broadband networks. These risks and uncertainties are in one respect intrinsic characteristics of large infrastructural projects, and on the other hand they are caused by (unknown) competitor behaviour, (unknown) end-user demand, rapid technological development and different development paths available to operators. A combination of the uncertainty typologies from Meijer et al., (2006), Walker et al., (2003) and Courtney et al., (1997) forms a suitable framework to assess the risks and uncertainties involved in the decision-making process concerning the upgrading and roll-out of broadband networks. This framework comprises classifications of risk and uncertainty on three dimensions: nature, level and source. In our qualitative research (see Chapter 6), we have validated the dimension of source and investigated whether the several sources of risk and uncertainty as introduced by Meijer et al., (2006) are experienced by strategic decision-makers on broadband roll-out and how they manifest themselves in practice. The classifications to nature and level are measured in our quantitative research (Chapter 7)

Source of risk and uncertainty Based on the Five Forces of Competitive Position model by Porter (1980), discussing uncertainties such as the behaviour of new entrants, substitutes in the market, buyer power, supplier power and rivalry, Meijer et al. (2006) draw the following distinction between sources of uncertainty: technological, resource, competitive (new entrants & rivalry), supplier, consumer uncertainty (buyer power) and political/regulatory uncertainty (see also Chapter 4). Based on the results of our in-depth interviews, it becomes clear that, within the telecom market, the six above-mentioned sources of uncertainty are clearly visible. Decision-makers are faced with technological, resource-related, competitive, supplier-related, consumer-related and political/regulatory uncertainty. Nature of risk and uncertainty Risks and uncertainty can exist due to imperfect knowledge (which can be reduced by gaining more knowledge on the subject) or risk and uncertainty that is inherent in characteristics of the environment (which cannot be reduced by gaining more knowledge about the subject) (see also Chapter 4). We measured the categorisation by asking respondents to classify the various types of risk and uncertainty into two groups: those that are due to imperfect knowledge (knowledge or epistemic uncertainty) and those that are inherent in the environment (variability uncertainty). It is striking that there is a great deal of disagreement between respondents with regard to the categorisation of risk and uncertainty on the basis of their nature. Most respondents agree that changes in regulation are an uncertainty due to

295

changes in the environment, which means that they cannot be reduced through increased knowledge. Most categories of risk and uncertainty that can be solved via increased knowledge have to do with technological and operational network-related aspects, including supplies and the availability of financial resources. Increasing knowledge about these subjects will reduce the uncertainty associated with them. Market-related issues, like general market developments, development of end-user services, end-user demand and regulation, are the risks and uncertainties that are least likely to be solved through increased knowledge. Predictability of risk and uncertainty A second aspect of risk and uncertainty is their level of predictability. Risks and uncertainties can be predictable at several levels. We measured the level of risk and uncertainty by asking respondents to indicate for every type of risk or uncertainty how predictable the future of this type is on a four point scale (future is predictable; a limited number of possible future scenarios; a broad range of possible future scenarios; future is completely unpredictable). Risk and uncertainties related to technological and operational network-related aspects, including supplies and the availability of financial resources are expected to be reduced by gaining more knowledge on the concerned issues (epistemic or knowledge uncertainties). Aspects related to end-user behaviour like consumer preferences and characteristics, end-user service development and development of demand are considered highly unpredictable. The behaviour of regulators as well as end-users clearly are uncertainties that are considered being not reducible by gaining more knowledge, but as uncertainties that are inherent in the environment (variability uncertainties). Regulators are, however, more predictable than end-users in their comings and goings. 10% of the respondents still hold the opinion that for changes and inconsistency in regulation the future is completely unpredictable, which can have a restraining influence on network upgrades. The fact that the development of broadband end-user services is so unpredictable may help explain the ‘chicken and egg’ problem of the broadband market (a lack of high-speed broadband roll-out results in a lack of broadband services development and the other way around). Because infrastructure providers do not know whether there will be sufficient services to make their networks profitable, they do not roll-out high-speed networks on a large scale. Effects of risks and uncertainties Operators indicate that the way risks and uncertainties affect technological choices strongly depends on the urgency of the problem. When it concerns a roadmap of a service that may be successful in approximately two years, operators will investigate more specifically whether this is

296

the right path to follow. In case of high urgency and more rigorous steps, operators are mostly forced by their environment to choose the best path of that moment. Based on interview analysis we can conclude that risks and uncertainties have two important effects. The first is building in flexibility by keeping more optional development paths open. This way, different steps are available, depending on the circumstances, and adjustments can be made when changes occur in the environment. Building in flexibility and keeping open more options for the future is a clear practical example of real options thinking. For that reason, it is striking that, as we concluded in paragraph 8.3.3, Real Options Analysis is not applied as a method for financial assessments by telecom companies. Although operators in practice act according to this method they do not incorporate it in their investment-related decisions. The second effect of regulatory uncertainty is delay or reconsideration of investments. Although it is difficult for regulators to keep pace with the quickly developing telecom market, they should be aware of the impact of the uncertainty about the outcomes of regulatory processes on the decision-making processes of telecom operators. A climate of political and regulatory uncertainty poses a threshold to operators with regard to major infrastructural investments. Reducing risk and uncertainty In Chapter 4, we introduced several ways to reduce risk and uncertainty: technological analysis, cost-benefit analysis, obtaining information from suppliers, benchmarks with competitors, forecasting and scenario investigation. These methods can be used to make the future more predictable for these types of risk and uncertainty that are not completely inherent in the environment and thus to some extent within the span of control of decision-makers. Based on both the qualitative and the quantitative analysis, we investigated whether and to what extent these methods are applied by decision-makers to reduce risks and uncertainties. Decision-makers were asked which methods they apply in their daily practice to handle risk and uncertainty in their decision-making processes. The conclusion of the qualitative analysis is that technological analysis, forecasting and pilots are most frequently applied methods to reduce risk and uncertainty. Analyses and forecasting take place early on in the decision-making process, when alternatives are investigated. Pilots are performed later on in the process to prepare for any unforeseen side effects with regard to the implementation of a new technology. Our quantitative analysis yields more or less the same results. Also, technological analysis and forecasting are mentioned as highly important in reducing risk and uncertainty. However, an important difference with our qualitative results is that, although scenario analysis was mentioned during the interviews, it was not mentioned specifically as an important method to reduce risk and uncertainty, whereas in our quantitative study, scenario analysis was mentioned as the number one method. Finally, cost-benefit analysis was also

297

mentioned as an important reducing factor. The fact that this method was also mentioned as being highly important in our financial analysis underlines the importance network operators ascribe to this method within their decision-making processes regarding the roll-out of broadband networks.

8.3.5 The course of the decision-making process

In addition to more insight into the multidisciplinary factors influencing broadband investment-related decisions, and into the methods applied to choose between several investment-related options, more insight is needed into the decision-making process that eventually results in these investment-related decisions. This brings us to the next intermediate research question: RQ 5 How does the decision-making process by telecom network operators regarding the

migration of broadband access networks evolve?

Insight into the duration and evolution of the decision-making process with regard to broadband roll-out in the local loop is gained via qualitative research (Chapter 6). It becomes clear that the distinction between non-programmed and programmed decisions, as introduced by Simon (1960), as well as the further distinction between short-term operational decisions and periodic control decisions (Lin, 2004) can be applied to the different decisions that are part of the broadband upgrade and roll-out process. Smaller decisions concerning the implementation are moved to operational departments and involve more people. The major strategic decisions, for example with regard to investments, are made by senior management in small groups. Strategic decision-making takes between six months and a year. Factors determining the duration of the decision-making process are mainly to be found in the complexity of the decision-making object, the financial aspects and the availability of a clear ‘sense of urgency’ to arrive at a decision. Roll-out times vary from between several months to ten years, and depend on the extension of the roll-out: systems, platforms or new technologies. Duration, moreover, depends on the amount of technological adjustments, in other words the evolutionary or revolutionary character of the roll-out. The third determinant is scale, in other words the difference between upgrades of the core network only or the roll-out of the entire network.

The applicability of two decision-making theories, i.e. the Garbage Can model by Cohen, March & Olsen (1972) and the theory of Logical Incrementalism (Quinn, 1980; Das & Teng, 1999) have been validated. The Garbage Can theory can only be applied to organisational anarchies. Cohen et al. have drawn up three criteria

298

for organisational hierarchies: problematic goals, unclear technology and fluid participation. Although the Garbage Can theory and the criteria of an organised anarchy were originally applied to universities (Cohen et al., 1972; Hayes et al., 1998), the theory has since also been applied to other types of companies. Lipson (2007) has demonstrated that the criteria governing organised anarchies and, by extension, the Garbage Can theory, is even applicable to global organisations like the United Nations. Based on this broad applicability and on the insights described in this and earlier chapters, we assumed that telecommunication companies could also be characterised as organised anarchies. The results of our interviews demonstrated that the three characteristics for organised anarchies can clearly be applied to telecom companies. As a result, we can conclude that telecom companies can be qualified as organised anarchies in the way Cohen et al. (1972) meant in their theory. Our results have, moreover, convincingly shown that strategic decision-making with regard to network upgrades and roll-out is not a structured, rational process, but can be characterised by connections of existing and suddenly emerging problems and solutions in combination with the availability of people at a certain time and place. For that reason, the decision-making processes concerning broadband roll-out for that reason matches the characteristics of the Garbage Can theory. A next assumption had to do with the applicability of the theory of Logical Incrementalism (Quinn, 1980). Based on a clear similarity between the Real Options Analysis and Logical Incrementalism and the fact that it was shown in literature that Real Option Analysis can very well be applied to uncertain markets such as the telecom market (a.o. Neufville, 2001 & 2003; Oslington, 2004; Miller & Waller, 2003), we assumed that the theory of Logical Incrementalism would be applicable to the decision-making processes concerning broadband network roll-out as well. The assumption comprises three characteristics of Logical Incrementalism, as formulated by Das & Teng (1999): 1) incremental nature, 2) broad and relatively vague objectives, and 3) options for developments and adjustments. Throughout our interviews, various examples were given that underline the three characteristics of incrementalism as formulated by Das & Teng (1999). Based on this, we assume that the theory of Logical Incrementalism is applicable. Moreover, it has become clear that strategic decision-making is an incremental process. The applicability of the theory of Logical Incrementalism is, moreover, demonstrated by the fact that both cable operators as well as telecom operators prefer evolutionary paths above a revolutionary strategy with regard to their physical networks. Networks are upgraded when critical mass for a certain bandwidth has been reached and operators follow an ‘invest-as-you-grow’ strategy. An important link between the two above-mentioned theories can be made. While the Garbage Can theory can be seen more as a descriptive theory, the theory of

299

Logical Incrementalism is a theory that offers a way to handle uncertainty and unpredictable situations. The power of Logical Incrementalism lies in the fact that managers gain access to broad options and subsequently reduce the range of relevant options over time. This is necessary, because the market in which they operate is a highly uncertain one. Because so called ‘choice opportunities’, moments where problems, solutions and participants come together, can occur at any moment, decision-makers must be able to adjust in a flexible way. For that reason, we can conclude that the applicability of the theory of Logical Incrementalism can be seen as a logical response to the uncertainty and unpredictability resulting from the way problems and solutions interact in the way the Garbage Can model describes.

8.3.6 Towards a parsimonious model for broadband-related decision-

making

An integrated multidisciplinary scientific framework or unifying theory would be relevant to scientists, strategic decision-makers in broadband and regulators. Although many technological, regulatory and market-related factors are relevant, it is unclear how these factors are interrelated. Related to the content-related side of this research, we argued in Chapter 1 that it is important to gain more insight into the complex relationships between technological, financial and market-related factors. We recapture the objective of the content related side of our research: ‘To

develop a scientific integrated multidisciplinary model that provides insight into the drivers

and outcomes with regard to local loop broadband roll-out by telecom operators.’ The accompanying research question is:

RQ 6 Can we develop a parsimonious model that provides insight into the determining

factors with regard to broadband-related decision-making by telecom network operators

and the outcomes of broadband roll-out?

The initial conceptual model is the result of Meta-analysis, content analysis and network analysis (Chapter 3). The clusters found in this analysis, such as uncertainty around market developments (risks), cost-related issues around the technological configuration and density of the network, and expected revenues, are located in the first phase of the decision-making process. The development of attractive and innovative services can be seen as a condition for the actual adoption and use. Scalability and competition also play a role in generating actual use. The actual revenues depend on the actual use and operational costs, including the recuperation of the network investments. The conceptual model encapsulates the major findings and adds a more dynamic perspective to the analysis of broadband roll-out. The choice in favour of building the conceptual

300

model around four core phases appears to be valid based on the analysis of the centrality of the chosen core variables. Although this model already offers a high level of parsimony compared to the black hole presented in Figure 3.3, it has an important drawback, in that it is based solely on literature, most of which is based on theoretical results. Keeping in mind the objective of this model, namely to understand what actually drives operators in the field to roll-out their access networks, a model based on empirical results is necessary to draw conclusions for real market players. For that reason, a next step has been to validate the conceptual model via a qualitative data analysis through in-depth interviews with strategic decision-makers in Dutch telecom and cable companies. After validating the concepts of our conceptual model via concept mapping (Chapter 6), we can conclude that most concepts of the original conceptual model have been validated by the qualitative data analysis, albeit not always in their original position within the model or at the same level of aggregation. On the other hand, new information has become available by these interviews, which has resulted in additional concepts. Most noticeable are several feedback loops in the process, which imply it is an ongoing and repeating process. Secondly, interviews show the importance of the ‘soft side’ of the decision-making process. Market developments, the behaviour of competitors and the strategy of the business create a high sense of urgency to make a decision about rolling out or upgrading the network and are, for that reason, important potential accelerators of the decision-making process. The qualitative data analysis has, however, shown to be insufficient to achieve the desired level of parsimony. Moreover, the complexity of several technological, market-related and financial aspects resulted in an even more complex model. Because the applied methodology of concept mapping did not offer the possibility to test the relationships with regard to their significance, further simplification of the model could only be based on the number of times individual variables were mentioned in interviews. We feel that this method is not robust enough to arrive at a parsimonious model. A next step was to build a model based on quanitative data analysis. The problem that now emerges is that the conceptual model combines process and content into a single model. On the one hand, it is a time-phased model. The four phases in the core of the conceptual model follow each other in time. However, the factors that influence each of these phases are content-related. This makes this model very complex and difficult to validate. It was mainly for this reason that we decided to split content and process and build the model in such a way as to gain insight into the drivers and outcomes of broadband network upgrades. The decision-making process and the various phases that came after (operational roll-out, adoption and revenues) were investigated via qualitative data

301

analysis based on in-depth interviews with strategic decision-makers in the field. These interviews provided a clear and very detailed insight into the way the decision-making process evolves as well as into the further operational phases, the role of adoption and use and revenues (see paragraph 8.3.5). The qualitative analysis could, however, not provide the desired level of simplicity combined with a high enough level of significance with respect to the relationships between the content-related variables. For this reason, we conducted exploratory and explanatory factor analysis and structural equation modelling to build and test a parsimonious model regarding broadband roll-out. The quantitative analysis does not stand on its own, but includes results from our earlier theoretical and qualitative study. The constructs used in our measurement model were developed earlier on in the pre-test phase of this study. The constructs and underlying items were based on concepts from our conceptual model resulting from extensive Meta-analysis on broadband literature and validated in our qualitative analysis via several expert interviews. Structural equation modelling resulted in a robust, explanatory, parsimonious model. Starting from a situation with no clear conceptual relationships and no available theories or models to validate (referring to the ‘black hole’ in Figure 3.3) this can be considered a unique contribution to the broadband research field. Although our model is not perfect (see paragraph 7.5 on limitations) there at least is now a testable model which is open for validation and adjustments by other researchers. This model can function as a starting point for further research projects and provides some stability in this multidisciplinary, unpredictable and highly dynamic research field.

8.3.7 Incremental decision-making is a logic response to uncertain

interrelation between factors from different domains

There is a clear link between the content-related and process-related part of this study. Several technological, financial and market-related factors as well as implications from technology choices with regard to internal business processes, act upon each other like in a garbage can. At certain moments they come together and present a small window of opportunity in which decisions can be made: choice opportunities. The way these factors influence each other is highly unpredictable, creating a dynamic and uncertain environment for strategic decision-makers having to decide how and when to upgrade their access networks. The operators respond to the dynamics and uncertainty involved in an incremental manner. In practice, this can be seen in evolutionary network upgrades that go hand in hand with relatively small investments and technological upgrades that can be earned back within several years, rather than opting in favour of revolutionary developments. To conclude, opting in favour of an evolutionary roll-out strategy is a logical and

302

sensible response to uncertainties in the broadband market environment that are a result of the unpredictable interrelation between factors involving technology, market, internal business and finance. Here, there is a clear link with Real Options Analysis, which also offers a way to assess financial investment options in an environment that is characterised by high levels of uncertainty and where fast changes in the environment have to be incorporated into the decision-making process and organisations have to adapt to these sudden changes in the environment. It is clear that operators incorporate option-based thinking into their decision-making processes, in that they keep open several options for the future, to be able to adjust their decisions based on changes in the environment and the emergence of new developments. Building flexibility into the decision-making process to prepare for possible changing circumstances is typical of the way telecom operators behave in this respect. They prepare their network for possible new technologies, even though these technologies may never be launched, for example by putting empty fiber ducts into the ground. Moreover, it has become clear that strategic decision-making is an incremental process, keeping open the possibility to adapt the network to future changes in the environment. The applicability of the theory of Logical Incrementalism, as well as Real Options Analysis is, moreover, demonstrated by the fact that both cable operators and telecom operators prefer evolutionary paths above a revolutionary strategy when it comes to their physical networks. Networks are upgraded when a critical mass for a certain bandwidth is visible and operators follow an ‘invest-as-you-grow’ strategy. For this reason, it is striking that both the qualitative results and the quantitative results of this study indicate that Real Option Analysis is not applied by network operators in the financial assessment of technological options. We can conclude that, even though the thinking that forms the basis of ROA is applied in decision-making with regard to access network roll-out, the methods that are used to translate this thinking into practical applications to assess investment options are not adopted by decision-makers. There are several possible explanations for this state of affairs. First of all, decision-makers are simply not familiar with this method, even though it is quite well known in academic circles and there have been a number of studies investigating its applicability in uncertain environments. In light of the fact that the method is still being developed and tested to determine in what kind of situations it can be applied, this is hardly striking. Secondly, it is a highly complex and difficult theory. Decision-makers will have to be trained before they can use this method in their decision-making process. Moreover, it requires an entirely different way of thinking and analysing, which is not something decision-makers will learn overnight. To conclude, in this study, we have provided greater insight and a clear

303

explanation for the evolution of the decision-making process of private operators on broadband roll-out in the local loop.

8.4 Limitations of this research and recommendations for

further research

There are some limitations to our research, which are related to the focus of this study, the methodology we applied and the development of our model. Research focus

The fact that we have decided to adopt a supply side point of view can be seen as a clear limitation. We are aware that users are also important actors. Research into the adoption and use of broadband by consumers has, however, been the subject of another, related PhD research project (Vermaas, 2007). Although we are aware of the co-creating role of users in the case of new broadband services, we decided to focus exclusively on the supply side. Further research We believe that carrying out the same type of research from the regulatory/policy perspective is likely to yield interesting results. Combined with this study and that by Vermaas (2007), a multidimensional comparison could be made between the various stakeholders on the broadband market. Research could look at actor behaviour, their mutual dependency and ways the various actors react to each other and to market developments based on their own interests. Game theory could play an interesting role in this kind of research. Research methodology

All of the research methods discussed above have some limitations, which we briefly discuss below. Meta-analysis & conceptual model A first limitation has to do with the fact that we have been quite rigorous in terms of the way we selected the papers for investigation. However, we assume that the 50 papers we examined represent the body of know-how on broadband roll-out. The main problems in terms of interpreting the information had to do with the aggregation. We made a number of decisions regarding how to cluster very specific remarks into more general categories. It is almost impossible to avoid subjectivity in this phase of a research project. A next limitation is related to the focus of this study. It is clear that, had we started from another dependent variable (for instance economic welfare), we would have found a different network. In that sense, a network reflects the choice made by the researcher with regard to the key concepts (see Chapter 3 for a more detailed discussion).

304

Quantitative research Our model is the result of quantitative research with a relative small sample of 87 respondents. Although the sample contains experts, decision-makers and researchers from a broad range of countries, our results would have been more robust with a larger sample. Another limitation of working with a sample consisting of competing market parties is the fact that some information may not (exactly) be provided based on competitive considerations. However, because the questions underlying our model were not sensitive from a competitive perspective, we assume that these considerations are not reflected in our final model (see Chapter 7 for a more detailed discussion). Qualitative research A general risk of doing interviews with decision-makers in companies is that they always have a hidden agenda and will never provide insight into their business strategy, because such information is highly confidential. However, for our research, we needed the variables influencing the various phases of the process and their relative importance, rather than specific numbers. A second limitation is related to the software programme we used to analyse our interviews. A major drawback of the software tool Atlas.ti is that the number of times relationships between variables are mentioned remains unknown. Although the programme does provide insightful indications of the strength of relationships between concepts, further validation of these relationships is necessary. We ensured this second validation via quantitative analysis (see Chapter 6 for a more detailed discussion). Further research We used Atlas.ti to build a general model that reflects the opinion of all market players together. An interesting adjustment would have been to make separate models for cable, copper and alternative network providers as well, to compare their visions and the resulting concept maps. However, we did not have enough time to include these analyses in this thesis, but would recommend doing so for further cross-actor analysis.

Model development

Three limitations can be mentioned with regard to the development of our final model. Our model proves that customer satisfaction is increased by broadband upgrade and roll-out, as well as by the availability and development of broadband services. Our sample did, however, not contain end-users of broadband services. As a result, the causal relationships we identified were validated by end-users themselves, but reflect the operators’ point of view of how end-user satisfaction can be reached. The underlying hypotheses are, however, based on a theoretical background. Secondly, during model development, it became clear that many of our measures contained only 2 or 3 items. Considering the fact that 2 items is the absolute minimum, more items per construct would have increased construct

305

stability. Finally, one could question the added or predictive value of striving for a parsimonious model for a decision-making process that takes place in a complex, highly dynamic and uncertain environment (see Chapter 7 for a more detailed discussion). Further research Due to the limited time available and the somewhat small dataset, some relationships could not be tested, although they do deserve to be investigated in further research. The first relationship is the positive relationship between customer value (containing customer satisfaction and image) and critical mass. The relationships ‘customer value’ leads to ‘critical mass’ and ‘customer value’ leads to ‘end-user demand’ have, moreover, not been tested in our research. These, too, deserve to be investigated in further research. The same goes for the positive relationship between critical mass and social & cultural effects, which is interesting to test because it could give governments an indication as to how they could stimulate critical mass development. A final relationship recommended for testing in further research is the positive relationship between actual and expected demand and service development. The decision-makers in our interviews clearly indicated that the relationship between the availability & development of end-user services and broadband roll-out is a positive one. Testing the positive relationship between end-user demand and development of services is recommended for further research. The relevance of ‘broadband’

We are aware of the fact that broadband is a relative concept and that, as a result, depending on the actual availability of bandwidth at a certain place at a certain time, broadband has different connotations. By extension, the focus on new innovative technologies is both a strength and a weakness of this study. While technologies are described and their pros and cons investigated, new, alternative and better versions emerge at a rapid speed. Despite the fast development of these technologies, we believe it is useful to analyse the concept of evolutionary development paths and outline possible alternative scenarios on the basis of relevant criteria, as we have attempted to do in Chapter 2 of this thesis.

However, there are many regional variations with regard to the phases and rates of network development, adopted services and implemented network technologies. Because of this, what is an evolutionary step in one situation may be a revolutionary step in another. In western countries, the roll-out of fixed networks like copper TP and coax networks is highly advanced, whereas in countries where no such networks are rolled out, the choice is mainly between different wireless technologies. If they are to make a balanced decision, operators need to be aware of the flexible, cheaper and potentially better technological scenario’s that are available, allowing them to respond flexibly to changes in user demand. Each new

306

generation brings improvements in functionality, performance, price and QoS. This is a continuing trend that will keep presenting operators with new opportunities as well as uncertainties.

8.5 Research Contributions

This research has different implications for several stakeholders. This paragraph provides insight into the main implications for scientists, strategic decision-makers and policy-makers.

8.5.1 Implications for scientists

Model development For scientists more insight into the evolution of this decision-making process provides more structure and offers the possibility to do more focused research. Here the need comes up for an integrated multidisciplinary scientific framework or a unifying theory which describes the way technological, economic, market-related and regulatory factors affect migration of broadband networks in the local loop. Although many technological, regulatory and market-related factors are relevant, it is unclear how these factors are interrelated and which factors are dominant within the different phases of the roll-out process. In literature on R&D, innovation and marketing many models have been developed (see for an overview de Reuver et al., 2008). Research into literature on broadband roll-out has not, however, been very helpful in delivering clear concise models to support decision-makers, nor has it managed to isolate the factors that explain successful roll-out. As we demonstrated in our literature analysis, there is as yet no unambiguous theory on the coherence of driving factors, outcomes and theories. Thus far, researchers have been unable to create any model for broadband network roll-out which is validated by experts as well decision-makers in the field, covering multidimensional factors as well as decision-making and adoption theory. Limitations of existing models are that they focus only on technological and/or economic aspects. The models are, moreover, based on specific real or fictitious case studies. None of them take technological, economic, policy and social aspects into account. In addition, none of these researches has included a process-oriented approach. The model we built in this study covers economic, social and technological aspects of broadband development and is integrated with decision-making and adoption theory. These additional aspects make a model for migration towards broadband infrastructures much more realistic. Starting from a situation with no clear conceptual relationships and no available theories or models to validate (referring to the ‘black hole’ in Chapter 3), this can be considered a unique contribution to the area of broadband research. Although our model is not perfect there at least is now a testable model which is open for validation and adjustments by other researchers.

307

This model can function as a starting point for further research projects and provides some stability in this multidisciplinary, unpredictable and highly dynamic research field.

Relationships between financial and decision-making theories Another important contribution is the relationship we proved between financial and decision-making theory. The broadband market is characterised by many uncertainties in the field consisting of changes in technology, economic circumstances and the behaviour of end-users, suppliers and regulators. Literature (Neufville, 2001 & 2003; Oslington, 2004; Miller & Waller, 2003, among others) suggests that the Real Option Analysis, a financial assessment method which takes a high level of environmental uncertainty into account, would be well applicable to decision-making regarding investments in broadband network roll-out. This method enables a flexible strategy in which environmental changes can be incorporated and can lead to adapting earlier chosen paths. Flexibility and incremental decision steps as an answer to uncertainty forms a clear similarity between Real Option Analysis and the decision-making theory of Logical Incrementalism. Based on this clear similarity between these two theories and the fact that it was shown in literature that Real Option Analysis can be applied very well to uncertain markets as the telecom market we assume that the theory of Logical Incrementalism will as well be applicable to the decision-making processes concerning broadband network roll-out as well. Based on the results from qualitative analysis, this assumption has been verified. The examples of evolution of the decision-making process that were provided clearly indicate that operators incorporate options thinking in their decision-making processes. This options thinking reveals itself in the attempt to keep open several options for the future to be able to adjust decisions based on changes in the environment and the emergence of new developments. Building flexibility into the decision-making process to prepare for possible changing circumstances is typical of the behaviour of telecom operators. They invest in anticipating new technologies by preparing their networks, although these technologies may never be launched. An example of this is putting empty fiber ducts in the ground, which require additional investments, but give operators the option to switch to fiber networks as soon as this may be necessary, although they are not sure when they will use this option. These unused investments are ‘options’ that allow operators to change the design or product mix, without requiring them to do so (de Neufville, 2003). For this reason, it is striking that the results of our qualitative and quantitative analysis show that Real Option Analysis is not applied in the financial assessment of technological options by network operators. We can conclude that, even though the thinking that forms the basis for ROA is applied in decision-making regarding access network roll-out, the methods to translate this thinking into practical applications to assess investment options are not adopted by decision-makers.

308

Usability for other infrastructural sectors High initial investments, which come with high risks, are common characteristics of new infrastructures in general. When, for example, electricity networks are privatised, the same problems may occur. Understanding how these complex networks and the decision-making process may speed up (future) decision-making around projects that come with high (financial) risks, major long-term investments and high levels of uncertainty concerning the actors involved. Based on the qualitative and quantitative analysis, we validated the applicability of the theoretical framework on risk and uncertainty based on Meijer et al. (2006), Walker et al. (2003) and Courtney et al. (1997). Although the typology framework and the classification into sources used by Meijer et al. (2006)., is developed for transition within the energy sector, the sources of uncertainty defined in this typology are also clearly visible in the broadband market, which underlines the parallels between decision-making in these infrastructure sectors where network transition is concerned. For this reason, we may assume that this typology framework could also be used to gain more insight into the source, nature and level of risks and uncertainties in other infrastructural sectors, for instance transportation. In Meijer (2008), the typology framework was introduced, but not, however, applied via empirical data. Our research shows that the framework can be used for empirical research and for that reason, applying it to gain more empirical insight into risk and uncertainty during other infrastructure transitions is recommended.

8.5.2 Implications for strategic decision-makers

Although strategic decision-makers broadly know how the process works, they are usually unaware of how the separate factors function within the process and what their correlative influence is. Analysing the content-related side as well as the process-related side of broadband-related decision-making gives market parties greater insight into the influencing factors of the broadband-related decision-making process. Most of the times operators function on a day-to-day basis, influenced by ad hoc market developments and competitive behaviour, without realising the long-term strategy. This can result in ‘me too’ behaviour, which is not always the best long-term strategy. The model we developed in this study provides operators with insight into the most important drivers and outcomes of the roll-out of broadband in their access networks. Moreover, the structural analysis of the market in which they operate can provide much clarity in the unpredictable and uncertain market in which they operate.

309

Recommendations for strategic decision-makers

Realise a good balance between long and short-term strategy Although in qualitative research decision-makers indicate that long-term strategy is the main business-related driver with regard to the decisions they make, it would appear that, in practice, decisions are driven most of all by the behaviour of competitors and sudden changes in end-user demand. These are mainly short-term drivers. Although the market is dynamic and competitive pressure is high, operators cannot afford to lose sight of their long-term strategy. Day–to-day decisions should be in line with the long-term strategy. This can lead to internal difficulties because the daily problems and targets of marketing departments do not always seem to be in line with the long-term strategy objectives of the company. There should be sufficient attention to clear and frequent communication between departments with long and short time horizons to make sure the company is not just chasing after its competitors. Familiarise yourself with Real Options Analysis As concluded before, Real Options Analysis is a financial assessment method that better fits the characteristics of the broadband market than the now mostly applied NPV method. Although the method is not applied in practice on a large scale, CFOs and strategic decision-makers are recommended to familiarise themselves with the method and explore the possibilities of the method for their own financial assessments. Applying the method will demand a learning process and a completely new approach to financial decision-making. It will be difficult to introduce a completely new method within a short time frame. For that reason it is recommended to introduce the method via training and creating awareness within the company. Research institutes or universities can provide valuable assistance in this process.

8.5.3 Implications for policy-makers

Research into the behaviour of broadband suppliers is necessary to understand their reserved attitude concerning fiber roll-out in the local loop. First of all, this is important because broadband roll-out has, as demonstrated in several market studies as well as by the European Commission (see Chapter 1), significant economic and social relevance. As a result, broadband (FttH) roll-out is an important objective of Dutch and European governance. As far as governments are concerned, understanding the complicated nature of the decision-making process with regard to broadband network roll-out may help focus policy on removing the risks involved in these infrastructural investments. Although many stimulation projects have been initiated by European, national and local governments, FttH has not yet been rolled out in the Netherlands by private operators, with the exception of some small pilots. Investments with very long ROI times that were

310

common in the days before the liberalisation of the telecom market are no longer considered possible by private operators. These kinds of investments are possible in a monopolistic market, which is stable and predictable, but in current dynamic, uncertain and unpredictable market they no longer present a viable option. Stimulating the broadband flywheel effect by small pilot projects and R&D projects has, as a result, produced only little effect. The approach policy-makers have adopted until now can, for that reason, be considered ineffective. In case governments want to stimulate private fiber roll-out in an effective way, they have several options. An important problem is related to the costs in relation to geographical scale of a network. The result of insufficient demand in relation to the huge upfront investments is that infrastructural competition is no longer possible at the passive layers of the network with a viable fiber network on a national scale. Governments should, for that reason, keep in mind that fiber roll-out will not take place on a large scale within the coming 5-10 years, unless they either support network roll-out financially or put significantly more effort and money in broadband service development and adoption to push market demand towards drastically higher bandwidths. Financial support includes investing significantly in the roll-out of fiber networks, arranging interesting financial incentives for operators and functioning as significant launching customer. The positive effects of these methods can be seen in Japan and Korea. A problem that comes with this solution is that in Europe governments are legally speaking not allowed to provide major state aid to private companies. Another, rather drastic option brings us back to the initial role of governments in taking care of the roll-out of infrastructure and the introduction of natural monopolies. Here, we leave the model of infrastructural competition and shift towards a model of service competition based on a concession-based network model. The question is how far governments want to go in pushing market demand or investing in bandwidth that will not be used within the near future. This is an important question because the trend of rolling out higher speeds and upgrading networks has become clearly visible in the market in recent years, as a result of competitive pressure originating from infrastructural competition (see also Chapter 2). Regulatory uncertainty causes incremental network roll-out In addition to these rather drastic solutions, government and regulators can also focus on removing uncertainties from the market. As we have seen, evolutionary paths and the avoidance of large, long-term investments are a clear answer to high levels of uncertainty in the broadband market. The two most unpredictable types of uncertainty which can, moreover, not be reduced by gaining more knowledge are uncertainty about end-user demand and regulatory uncertainty. A large majority of our respondents consider changes in regulation the most important variability uncertainty, indicating that gaining more knowledge on this subject does not

311

reduce uncertainty. 10% of the respondents still hold the opinion that, as far as changes and inconsistency in regulation are concerned, the future is completely unpredictable. These results are alarming, considering the fact that operators in their strategy partially depend on regulatory decisions. Because regulation sets the framework for operators in the market and as such determines several boundaries for their operations, regulatory uncertainty is clearly an undesirable phenomenon in the telecom market. Regulatory uncertainty has several causes. First of all, it takes too long before operators know whether or not they obtain regulatory approval in issues like price developments and technological migrations. Secondly, there is uncertainty about future regulatory changes. In addition to the question if certain regulations will be implemented, an even more important question for operators is when they will be implemented. Uncertainty about which way the new market definitions will influence the business and market position of telecom operators as well as the direction of the regulation on emerging markets are two other sources of regulatory uncertainty. A serious issue is that the time horizon for regulatory policy is far shorter then the payback time of large infrastructural investments. Where regulatory policy is made for three to five years, large infrastructural investments will not be earned back before 15 to 20 years. As a result, operators have to take into account that they have to adjust their initially built business cases due to regulatory changes in between. This realistic possibility causes high uncertainty and discourages operators to make investments with long payback times. Regulatory uncertainty is, moreover, caused by uncertainties concerning the attitude of national and local governments with regard to public participation in the private broadband market. The general attitude of private market players concerning public involvement in the market is a relatively negative one. They see governments as new competitors and find it difficult to deal with the oftentimes unclear entanglement of private and public interests. Governments should be aware of the possible negative side effects of their interference in the market. Foreign market parties indicate that they are considering postponing or terminating their investments in the Dutch market due to the instable and distorted market environment. Moreover, market distortion takes place because new players are buying the public networks of bankrupt broadband initiatives. When these networks are no longer viable for local governments, they are sold to a market party for much less money than telecom operators have invested in their fiber networks. Finally, operators consider public fiber initiatives a distortion of the private market and see them as unfair competitors. A result of this is that in areas where operators used to work together with local governments (for example the geography of their

312

networks), they are now reluctant to do so, and sometimes will only collaborate when they are legally obliged to do so. Although it is difficult for regulators to keep pace with the fast developing telecom market, they should be aware of the impact the uncertainty about the outcomes of regulatory processes have on the decision-making processes of telecom operators. A climate of political and regulatory uncertainty inhibits the actions of operators concerning large infrastructural investments.

Recommendations for policy-makers

Avoid regulatory uncertainty Based on our findings we can conclude that both the current behaviour of regulators and current policies as applied by (local) government with regard to public participation in the private market add such a level of uncertainty to the market that they reduce rather than stimulate private investments in FttH. In spite of the fact that public interference in the market has helped create a more competitive attitude among operators and has certainly contributed to an accelerated roll-out of ADSL, the current regulatory framework and public market participation have, as a result of the high level of uncertainty they create, brought about a counter-productive than stimulating effect with regard to the roll-out of FttH networks. Governments and regulators should, for that reason, begin by looking at themselves. Because their own behaviour is also a cause of uncertainty in the market, they are themselves responsible for further evolutionary roll-out themselves and merely increase the uncertainties that keep operators from making long-term investments in revolutionary technologies. Regulation must evolve faster and regulatory uncertainty must be avoided as much as possible to eliminate a possible cause of delay of infrastructural investments in the market. Our final model has proven that customer satisfaction is increased both by broadband roll-out and by the availability and development of end-user services. More focus on service development The importance of service development is highly relevant, because we concluded that the development of broadband end-user services also causes high levels of uncertainty. The fact that innovations in broadband service also play a major role in reconsidering delayed broadband roll-out decisions is an argument in favour of service development and more stimulation of broadband service development by the government. Moreover, in our qualitative analysis we have shown that service development is seen by operators as an important driver for network upgrade and roll-out. Based on these findings it can be argued that the focus of governments should not only be on stimulating the roll-out of infrastructure, but even more on broadband service development. For that reason, we argue that service development should be stimulated and subsidised to create critical mass. Also, actions to increase adoption are strongly recommended.

313

8.6 References

Alleman, J. (2002). A new view of telecommunications economics. Telecommunications Policy, 26 (1) 87-92(6).

Cohen, March & Olsen (1972). A Garbage Can Model of Organisational Choice. Administrative Science Quarterly, 17 (1),1-25.

Courtney, H., J. Kirkland and P. Viguerie (1997). Strategy Under Uncertainty. Harvard Business Review 75(6),67-79.

Das, T.K. and Bing-Sheng Teng (1999). Cognitive biases and strategic decision processes: an integrative perspective. Journal of Management Studies, 36 (6), 757-778.

Baas, H. de (1998). Hectiek en vereenvoudiging van besturingscontext- Afvalbeleid in Nederland naar een nieuwe ronde. Bestuurskunde, 5, 210-218.

De Reuver, Mark, Harry Bouwman and Ian McInnes (2008). Business models dynamics for start-ups and innovating e-businesses. Accepted for publication in The International Journal for Electronic Business.

Fijnvandraat, Marieke and Harry Bouwman (2006). Flexibility and broadband evolution. Telecom Policy 30 (8-9), 424-444

Hayes, Sherman L., and Patricia B. McGee (1998). Garbage can decision-making in a ‘structured anarchy’ for your CWIS. Could you translate that for me please? Campus-Wide information systems, 15 (1), 29-33.

Keswani, A. and Schackleton, M.B. (2004). How real option disinvestment flexibility augments project NPV. European journal of operational research 168 (1), 240-252.

Lin, Hui-Chao (2004). Decision Theory and Analysis. Futurics, 28 (1&2), 27-45.

Lipson, Michael (2007). A Garbage Can Model of UN Peacekeeping. Global

Governance, 13 (1), 79-97.

Meijer (2008). Uncertainty and entrepreneurial action- The role of uncertainty in teh development of emerging energy technologies. (Doctoral dissertation, Delft University of Technology, 2008).

Meijer, Ineke S.M., Marko P. Hekkert, Jan Faber and Ruud E.H.M. Smits (2006). Perceived uncertainties regarding socio-technological transformations: towards a framework. International Journal of Foresight and Innovation Policy, 2 (2), 214-240.

314

Miller, Kent D. and H. Gregory Waller (2003). Scenarios, real options and integrated risk management. Long Range Planning, 36 (1), 93-107.

Neufville, R. (2001). Real options: dealing with uncertainty in systems planning and design, paper prepared for presentation June 29, 2001 at 5th international conference on Technology, Policy and Innovation, Technical University Delft, Delft, The Netherlands. Retrieved: January, 29 2008 from: http://msl1.mit.edu/mib/dsp/curricula.MIT.edu/~dsplan/docs/papers/Delft.htm

Neufville, Richard de (2003). Real options: Dealing with uncertainty in systems planning and design. Integrated Assessment,4 (1), 26-34.

Oslington, Paul (2004). The Impact of Uncertainty and Irreversibility on Investments in Online Learning. Distance Education, 25 (2), 233-242. Retrieved May 13th, 2008 from: http://www.unsw.adfa.edu.au/sbus/OslingtonLinks/TimingInvestEdTech.pdf

Piyatrapoomi, N., A. Kumar & S. Setunge (2004). Framework for investment decision-making under risk and uncertainty for infrastructure asset management. Research in Transportation Economics, 8, 199-214, 2004.

Porter (1980). Competitive strategy: Techniques for analyzing industries and

competitors. New York: Free Press.

Quinn, James Brian (1980) Strategies for change-logical incrementalism, 1st Edition. Homewood: R D Irwin.

Remer, S. & Nieto, A.P. (1995a). A compendium and comparison of 25 project evaluation techniques. Part 1: Net present value and rate of return methods. International Journal of Production Economics, 42 (1), 79-96.

Remer, S. & Nieto, A.P.(1995b). A compendium and comparison of 25 project evaluation techniques. Part 2: Ratio, payback, and accounting methods. International Journal of Production Economics, 42 (2), 101-129.

Rui, Luo, Ti-gang Ning, Tang-jun Li, Li-bo Cai, Feng Qiu, Jian Shui-sheng and Jing-jing Xu (2005). FttH- a promising broadband technology. Proceedings from the 2005 International Conference on Communications, Circuits and Systems, 1, 609-612.

Simon, H.A. (1960). The new science of management decision (3rd print). New York: Harper and Row.

Van de Wijngaert (2001). Fysieke en affectieve toegang, geschiktheid; vraag, aanbod en context. In Harry Bouwman (Ed.), Communicatie in de

informatiesamenleving (51-71).Utrecht: Lemma.

Vermaas, Karianne (2007), Fast diffusion and broadening use (Doctoral dissertation, University of Utrecht, 2007).

315

Walker, W.E., P., Harremoes, J. Rotmans, J.P. van der Sluijs, M.B.A. van Asselt, P. Janssen and M.P. Krayer von Krauss (2003). Defining Uncertainty- A Conceptual Basis for Uncertainty Management in Model-Based Decision Support. Integrated Assessment, 4 (1), 5-17.

316

317

Annexes

318

319

Annex A

Numeric Tables Factor analysis quantitative

results Pre-test

Table A1: Factor Analysis technology items

I II III

2.8 Geographical scalability .89

2.9 Flexibility of technological solution .82

2.7 Technological scalability .80

2.6 Maturity of technology .79

2.3 Compatibility with existing network technologies .53

2.4 Standardisation of technology or use of de facto standards

.88

2.5 Openness of standard .89

2.1 Capacity: max number of Mbps .82

2.2 Reach: maximum number of Mbps transmitted over a certain distance at a reasonable quality

.85

Eigen value 3.111 1.619 1.550

Explained Variance 34% 18% 17%

KMO measure of sampling adequacy .66

Cronbach’s α .83 .73 .63

Table A2: Factor Analysis Market items

I II III IV

4.1 Density (potential customers per square km) .52

4.10 Expected revenues .82

4.11 Expected CAPEX/OPEX .89

4.4 Demand aggregation .63

4.5 Role of lead users, for instance government .82

4.7 Marketing of broadband .56

4.9 Broadband offer of (local) governments .73

4.3 Market demand: consumers .81

4.6 Availability of broadband services .73

4.8 Broadband offering by competitors .62

4.2 Reliability of market demand forecasts .69

4.7 Marketing of broadband .54

Eigen value 2,616 1,880 1,367 1,259

Explained Variance 24% 17% 12% 11%

KMO measure of sampling adequacy .49

Cronbach’s α .61 .66 .60 .30

320

Table A3: Factor analysis organisation-related items

I II III

6.3 (Effect on) internal business processes (billing, customer care)

.71

6.4 (Effect on) internal processes for supplying additional

equipment (set-top boxes) or services

.83

6.7 Operations and maintenance of the network .70

6.11 Alternative/competing projects within the company .62

6.2 Impact of stakeholders on company decisions .74

6.8 Company size .59

6.9 Power of holding/mother company .81

6.5 Marketing of new broadband access service .85

6.6 Marketing of end-user services .76

Eigen value 3,040 2,052 1,537

Explained Variance 28% 19% 14%

KMO measure of sampling adequacy .56

Cronbach’s α .71 .70 .77

Table A4: Factor analysis outcomes of broadband roll-out

I II III IV

13.4 Know-how on how to roll-out/upgrade

broadband networks

.96

13.5 Increased Know-how and learning on how to

roll-out, upgrade broadband networks

.94

13.16 Stimulation for changed network configuration .79

13.3 Next step in broadband roll-out achieved .77

13.6 Stimulation for further investments .70

13.8 Cultural consequences: high quality content .89

13.7 Social consequences: e-learning, remote working

.87

13.15 Innovations in broadband services .71

13.10 INcreased digital divide .58

13.14 Expanded customer base .81

13.13 Image .80

13.12 Customer satisfaction .59

13.2 Strategic advantage over competitors .84

13.1 Clear revenues .64

13.11 Decreased digital divide

Eigen value 3.800 2.627 2.188 1.549

Explained Variance 25% 18% 15% 10%

KMO measure of sampling adequacy .61

Cronbach’s α .89 .81 .70 .51

321

Annex B Numeric tables descriptive results quantitative

analysis main study

Table B1: Relevant market characteristics that play a role in decision whether or not to

upgrade/roll-out

No

influence

at all

Little

influence

Influence Large

influence

Decisive

influence

Mean Std

dev

Density (number of –

potential- customers per

square km)

1.2 17.4 38.4 43.0 4.23 .777

Reliability of market

demand forecasts 1.1 16.1 42.5 33.3 6.9 3.29 .861

Actual market demand:

consumers 3.5 24.4 47.7 24.4 3.93 .794

Expected end-user

demand 2.3 17.2 51.7 28.7 4.07 .744

End-user demand

clustering 1.3 7.8 45.5 36.4 9.1 3.44 .819

Role of lead users, for

instance government 2.3 26.7 41.9 23.3 5.8 3.03 .913

Development of new

broadband services 1.1 10.3 24.1 43.7 20.7 3.72 .949

Sense of urgency 1.2 18.8 49.4 23.5 7.1 3.16 .857

Marketing of broadband

end-user services 1.2 13.1 42.9 35.7 7.1 3.35 .843

Availability of broadband

services 1.1 9.2 31.0 41.4 17.2 3.64 .915

Broadband offer of

commercial competitors 1.1 2.3 19.5 46.0 31.0 4.03 .841

Broadband offer of (local)

governments 5.7 21.8 43.7 20.7 8.0 3.03 .994

322

Table B2: Relevant financial issues that play a role in decision whether or not to roll-out

No

influence

at all

Little

influence

Influence Large

influence

Decisive

influence

Mean Std

dev

Expected revenues 10.3 31.0 58.6 4.48 .680

Expected capital

expenditures

(investments) of network

upgrade/roll-out (CAPEX)

2.3 11.5 39.1 47.1 4.31 .767

Expected operational costs

of network upgrade/roll-

out (OPEX)

2.3 13.8 42.5 41.4 4.23 .773

Cost reduction (cost

rationalisation) 8.0 17.2 46.0 28.7 3.95 .888

Availability of financial

resources 11.5 25.3 35.6 27.6 3.79 .978

Actual revenues (higher or

lower than expected) 1.1 6.9 33.3 42.5 16.1 3.66 .874

Impact of stakeholders 2.4 17.9 45.2 15.5 19.0 3.31 1.053

OPEX reduction due to

new and more efficient

platforms

1.2 10.5 27.9 39.5 20.9 3.69 .961

Table B3: Relevant organisational issues that play a role in decision whether or not to roll-

out

No

influence

at all

Little

influence

Influence Large

influence

Decisive

influence

Mean Std

dev

Long-term strategy 2.3 17.2 52.9 27.6 4.06 .737

Impact of stakeholders on

company decisions 2.4 15.5 33.3 29.8 19.0 3.48 1.047

Entrepreneurship 1.2 14.5 38.6 32.5 13.3 3.42 .939

Vision 1.1 12.6 24.1 42.5 19.5 3.67 .972

(Effect on) internal

business processes

(billing, customer care)

3.5 22.4 47.1 20 7.1 3.05 .925

(Effect on) internal

processes for supplying

additional equipment (set-

top boxes) or services

4.7 25.9 47.1 20.0 2.4 2.89 .859

(Effect on) operations and

maintenance of the

network

1.1 10.3 41.4 34.5 12.6 3.47 .887

Timing of network upgrade 3.5 15.1 36.0 37.2 8.1 3.31 .949

Company size 4.7 31.8 28.2 27.1 8.2 3.02 1.058

323

No

influence

at all

Little

influence

Influence Large

influence

Decisive

influence

Mean Std

dev

Power of holding/mother

company 7.0 25.6 29.1 22.1 16.3 3.15 1.183

Strong financial position

company (’deep pockets’) 1.1 13.8 24.1 37.9 23.0 3.68 1.017

Alternative/competing

projects within the

company

3.5 17.4 43.0 30.2 5.8 3.17 .910

Regulation by regulatory

authorities 2.3 18.4 28.7 23.0 27.6 3.55 1.149

Dependency on

innovations of equipment

suppliers

1.1 28.7 37.9 29.9 2.3 3.03 .855

Participation in the

telecom market by (local)

governments

4.7 38.4 38.4 14.0 4.7 2.76 .920

Availability of human

resources 9.3 24.4 39.5 22.1 4.7 2.88 1.011

Table B4: Factors influencing the viability of broadband roll-out

No

impact

on

network

viability

Small

impact

on

network

viability

Impact

on

network

viability

High

impact

on

network

viability

Decisive

impact

network

viability

Mean Std

dev

Competitive pricing 1.1 9.2 19.5 50.6 19.5 3.78 .908

Broadband offering by

competitors 1.1 4.6 28.7 54.0 11.5 3.70 .779

Marketing campaigns

focused on broadband 1.1 12.6 50.6 28.7 6.9 3.28 .817

Satisfaction of lead users 12.6 39.1 37.9 10.3 3.46 .846

Brand name (image) 2.3 12.6 43.7 27.6 13.8 3.38 .995

Critical mass of available

‘real’ broadband services 1.1 10.3 26.4 41.4 20.7 3.70 .954

Critical mass of consumers

using broadband 1.1 4.6 24.1 48.3 21.8 3.85 .856

Triple play concepts 2.3 16.1 39.1 32.2 10.3 3.32 .946

3rd party end-user service

promotion 2.3 31.0 44.8 20.7 1.1 2.87 .804

The time it takes to

change to higher

bandwidths (conversion

rate)

2.3 25.3 42.5 21.8 8.0 3.08 .943

324

Table B5: Outcomes of broadband roll-out

Not

important

at all

Some

what

important

Important Mostly

important

Very

important Mean

Std

dev

Clear revenues 3.4 20.7 35.6 40.2 4.13 .860

Strategic advantage over

competitors 1.1 3.4 25.3 39.1 31.0 3.95 .901

Next step in broadband

roll-out achieved 4.6 21.8 52.9 17.2 3.4 2.93 .846

Increased know-how and

learning on how to roll-

out, upgrade broadband

networks (learning effects)

4.6 34.5 43.7 14.9 2.3 2.76 .849

Stimulation for further

investments 5.7 21.8 52.9 16.1 3.4 2.90 .863

Acceleration of broadband

roll-out 4.6 23.0 47.1 25.3 2.93 .818

Next step in broadband

roll-out achieved 4.6 35.6 37.9 20.7 1.1 2.78 .868

Social benefits: e-learning,

teleworking 18.4 40.2 21.8 17.2 2.3 2.45 1.054

Cultural benefits: high

quality content 17.2 31.0 33.0 16.1 2.3 2.55 1.032

Increased communication

between people 8.0 35.6 32.2 20.7 3.4 2.76 .988

Increased digital divide 41.4 35.6 14.9 8.0 1.90 .940

Decreased digital divide 26.4 33.3 20.7 19.5 2.33 1.075

Customer satisfaction 2.3 6.9 21.8 42.5 26.4 3.84 .975

Image 3.4 10.3 32.2 39.1 14.9 3.52 .987

Increased customer base 1.1 2.3 16.1 42.5 37.9 4.14 .851

Innovations in broadband

services 1.1 13.8 40.2 39.1 5.7 3.34 .833

Stimulation for changed

network configuration 11.5 37.9 41.1 8.0 1.1 2.49 847

325

Table B6: Nature of risk and uncertainty

R&U are due to

imperfect

knowledge

R&U are due to

characteristics

environment

Network & services

Innovation and development of new broadband technologies 39.1 60.9

The rate of new technological developments 39.1 60.9

Availability of alternative technological solutions 51.7 48.3

Future-proofness of technology 39.1 60.9

Using non-standardised technologies 56.3 43.7

Path dependency of choice of technology 52.9 47.1

Proper functioning of intelligence within the network 69.0 31.0

Development of broadband end-user services 25.3 74.7

Operations

Manageability of QoS of service delivery and network 69.0 31.0

Impact of new broadband networks on operational systems 61.6 38.4

Right timing of availability/delivery of necessary supplies 55.2 44.8

Quality of necessary supplies 66.7 33.3

Finance

Price of necessary supplies 49.4 50.6

Availability of financial resources 58.6 41.4

Unforeseen costs 56.3 43.7

Revenues 39.1 60.9

Development equipment price 33.3 66.7

End-users

Consumer preferences/characteristics 42.5 57.5

Development of demand 32.2 67.8

End-user tariffs 44.8 55.2

Market

Market developments 25.3 74.7

Participation in the broadband market by (local)

governments 36.8 63.2

Actions of potential or actual competitors 37.9 62.1

Other

Availability of necessary human resources 56.3 43.7

Effects of being first mover 39.1 60.9

Uncertainty about adoption of standards 36.8 63.2

Unclear/inconsistent regulation 28.7 71.3

Changes in regulation 23.0 77.0

326

Table B7: Degree of unpredictability for relevant risks and uncertainties

Near future

is predictable

There is a

limited

number of

possible

future

scenarios

There is a

broad range

of possible

future

scenarios

The future is

completely

unpredictable

Network and services

Innovation and development of

new broadband technologies 14.9 54.0 31.0

The rate of new technological

developments 14.9 49.4 31.0 4.6

Availability of alternative

technological solutions 17.2 47.1 35.6

Future-proofness of technology 13.8 49.4 29.9 6.9

Using non-standardised

technologies 13.8 46.0 27.6 12.6

Path dependency of choice of

technology 12.6 62.1 24.1 1.1

Proper functioning of intelligence

within the network 14.9 56.3 25.3 3.4

Development of broadband end-

user services 5.7 37.9 40.2 16.1

Operations

Manageability of QoS of service

delivery and network 34.5 52.9 12.6

Impact of new broadband

networks on operational systems 21.8 57.5 20.7 1.1

Right timing of

availability/delivery of necessary

supplies

20.7 57.5 20.7 1.1

Quality of necessary supplies 18.4 64.4 15.1 1.1

Finance

Price of necessary supplies 25.3 62.1 11.5 1.1

Availability of financial resources 23.0 54.0 19.5 3.4

Unforeseen costs 4.6 50.6 35.6 9.2

Revenues 4.6 49.4 36.8 9.2

Development equipment price 12.6 64.4 23.0

End-users

Consumer

preferences/characteristics 2.3 36.8 41.4 19.5

Development of demand 4.6 35.6 48.3 11.5

End-user tariffs 10.3 57.5 26.4 5.7

Market

Market developments 47.1 46.0 6.9

Participation in the broadband

market by (local) governments 13.8 51.7 29.9 4.6

327

Near future

is predictable

There is a

limited

number of

possible

future

scenarios

There is a

broad range

of possible

future

scenarios

The future is

completely

unpredictable

Actions of potential or actual

competitors 2.3 56.3 34.5 6.9

Other

Availability of necessary human

resources 28.7 55.2 14.9 1.1

Effects of being first mover 6.9 62.1 29.9 1.1

Uncertainty about adoption of

standards 6.9 59.8 26.4 6.9

Unclear/inconsistent regulation 8.0 54.0 27.6 10.3

Changes in regulation 6.9 49.4 33.3 10.3

Table B8: Factors influencing the reconsidering of the decision to delay the roll-out of

broadband.

No

impact

Small

impact

Impact High

impact

Decisive

impact

Mean Std

dev

Technological innovations 23.0 35.6 40.2 1.1 3.20 .805

Innovations in broadband

services requiring high

bandwidth

1.1 9.2 37.9 43.7 8.0 3.48 .819

Increasing customer

demand 1.1 4.6 23.0 55.2 16.1 3.80 .805

Change in the behaviour of

suppliers 4.6 26.4 57.5 6.9 4.6 2.80 .819

Change in the behaviour

of competitors 1.1 10.3 37.9 35.6 14.9 3.53 .913

Potential breakthrough of

a killer broadband service 3.4 10.3 24.1 44.8 17.2 3.62 1.003

Change in the behaviour of

government 8.0 36.8 32.2 19.5 3.4 2.74 .982

Increasing revenues 1.1 10.3 31.0 46.0 11.5 3.56 .872

Internal business strategy 2.3 8.0 37.9 31.0 20.7 3.60 .982

328

329

Annex C Numeric tables confirmatory factor analysis &

structural equation modelling

Table C1: Measures for technology-related factors influencing the decision to further rol -

out broadband

Measure Item

To what extent do technological characteristics influence the decision to roll-out new broadband networks or to upgrade existing ones?

Tech_1 Compatibility with existing network technologies

Tech_2* Standardisation of technology (either formally or de facto)

Tech_3* Openness of standards

Tech_4 Maturity of technology

Tech_5 Technological scalability

Tech_6 Flexibility of technological solution

Tech_7 Path dependency of choice of technology

Technological factors

Tech_8 Possibility to phase out old parts of the existing network

* removed from final measurement model

Table C2: Measures for demand-related factors influencing the decision to further roll out

broadband

Measure Item

To what extent do market-related characteristics influence the decision to roll-out new broadband networks or to upgrade existing ones?

Demand_1 Density (number of – potential - customers per square

km)

Demand_2 Reliability of market demand forecasts

Demand_3* Actual market demand: consumers

Actual &

expected demand

Demand_4 Expected end-user demand

Services_1 Development of new broadband services

Services_2 Marketing of broadband end-user services

Service development

Services_3 Availability of broadband services

* removed from final measurement model

330

Table C3: Measures for organisational related factors influencing the decision to further

roll out broadband

Measure Item

To what extent do organisational characteristics influence the decision to roll-out new broadband

networks or to upgrade existing ones?

Int_proc_1 (Effect on) internal business processes (billing, customer care)

Internal business processes Int_proc_2 (Effect on) internal processes for supplying additional

equipment (set-top boxes) or services

Vision_1* Entrepreneurship Vision&strategy

Vision_2* Vision

* removed from final measurement model

Table C4: Measures for financial related factors influencing the decision to further roll out

broadband

Measure Item To what extent do financial characteristics influence the decision to roll out new broadband networks or to upgrade

existing ones?

Fin_1* Expected revenues

Fin_2 Expected capital expenditures (investments) of network upgrade/roll-out (CAPEX)

Fin_3 Expected operational costs of network upgrade/roll-out (OPEX)

Fin_4* Availability of financial resources

Financial

factors

Fin_5* Actual revenues (higher or lower than expected)

* removed from final measurement model

Table C5: Measures for network viability

Measure No Could you indicate to what degree the following factors determine network viability?

Comp_1* Competitive pricing Offer competitors Comp_2* Broadband offering by competitors

CM_1 Critical mass of available ‘real’ broadband services Critical mass

CM_2 Critical mass of consumers using broadband

M&S_1* Marketing campaigns focused on broadband

M&S_2* Satisfaction of lead users

M&S_3* Brand name (image)

M&S_4* Triple play concepts

M&S_5* 3rd party end-user service promotion

Marketing & services

M&S_6* The time it takes to change to higher bandwidths (conversion rate)

* removed from final measurement model

331

Table C6: Measures for outcomes of broadband roll-out

Measure Item What tangible and intangible benefits and/or (social) effects are important as a result of broadband roll-out/network upgrade for a telecommunication company?

BBupg_1* Next step in broadband roll-out achieved

BBupg_2 Increased know-how and learning on how to roll-out, upgrade broadband networks (learning effects)

BBupg_3 Stimulation for further investments

BBupg_4 Acceleration of broadband roll-out

Broadband upgrade & roll-out

BBupg_5* Next step in broadband roll-out achieved

Soc_cult_1* Social benefits: e-learning, teleworking

Soc_cult_2 Cultural benefits: high quality content

Soc_cult_3 Increased communication between people

Soc_cult_4 Increased digital divide

Social & cultural effects

Soc_cult_5* eEcreased digital divide

Cust_value_1* Strategic advantage over competitors

Cust_value_2 Customer satisfaction

Cust_value_3 Image

Customer value

Cust_value_4* Increased customer base

* removed from final measurement model

332

Table C7: Convergent and discriminant validity of measurement model

Construct Item (No)

Std

factor

loading

Variance

extracted

Composite

reliability

Tech_1 2.3 0,58

Tech_4 2.6 0,67

Tech_5 2.7 0,83

Tech_6 2.10 0,67

Tech_7 2.14 0,56

Technological factors

Tech_8 2.15 0,61

0.43 0,67

Fin_2 5.2 0,76 Financial factors

Fin_3 5.3 0,82 0.63 0,93

Int_proc_1 4.5 0,77 Internal processes

Int_proc _2 4.6 0,87 0.67 0,90

Demand_1 3.1 0,70

Demand_2 3.2 0,76

Actual & expected

demand

Demand_4 3.4 0,53

0.45 0,79

Services_1 3.7 0,66

Services_2 3.9 0,76

Service development

Services_3 3.10 0,73

0.52 0,76

CM_1 14.6 0,92 Critical mass

CM_2 14.7 0,64 0.63 0,86

BBupg_2 20.4 0,71

BBupg_3 20.5 0,82

Broadband upgrade &

roll-out

BBupg_4 20.6 0,76

0.59 0,67

Soc_cult_2 20.9 0,77

Soc_cult_3 20.10 0,79

Social & cultural effects

Soc_cult_4 20.11 0,46

0.48 0,64

Cust_ value _2 20.13 0,80 Customer value

Cust_ value _3 20.14 0,88 0.71 0,90

333

Tech

nolo

gic

al

fact

ors

Org

anis

atio

nal

fact

ors

Pro

gre

ss b

road

ban

d

upg

rad

e an

d ro

ll o

ut

Act

ual &

exp

ecte

d

dem

and

Crit

ical

mas

sA

vaila

bili

ty &

dev

elo

pm

ent

bro

adb

and

ser

vice

s

Cus

tom

erva

lue

So

cial

& c

ultu

ral

effe

cts

Fin

anci

al f

acto

rs

Pos

s_ph

ase_

out_

old_

nwpa

rts

Pat

hdep

_tec

hn_c

hoic

e

Flex

ibili

ty_t

echn

_sol

s

Tech

n_sc

alab

ility

Mat

urity

_tec

hn

Com

patib

ility

_exi

st_t

echn

d1 d2 d3 d4 d5 d6

Eff

ect_

on_i

nt_s

uppl

y_pr

oces

s

Eff

ect_

on_i

nt_b

illin

g_pr

oces

s

d6 d7

Exp

ecte

d_O

PE

X

Exp

ecte

d_C

AP

EX

d11

d12

Exp

ecte

d_en

duse

rdem

and

Rel

iabi

lity_

dem

and_

fore

cast

s

Use

r_de

nsity

d8 d9 d10

Critical_mass_broadband_users

Critical_mass_real_BB_serv

d13

d14

Increased_knowhow_rollout

Stimulation_further_investments

d15

d16

Acceleration_broadband_rollout

d17

Cul

tura

l_be

nefit

s_hi

ghqu

al_c

onte

nt

Incr

ease

d_co

mm

_peo

ple

Dev

_new

_BB

_ser

vice

s

Mar

ketin

g_B

B_s

ervi

ces

Ava

ilabi

lity_

BB

_ser

vice

s

Imag

e

Cus

tom

er_s

atis

fact

ion

d22

d23

d24

d17

d19

d20

d21

Fig

ure

C1:

Fin

al

measu

rem

en

t m

odel

334

335

Annex D Questionnaire

Issues influencing the decision-making process to roll-out/upgrade networks In the decision to roll-out a new broadband network or to upgrade the existing network a number of issues play a role. We would like you to indicate the relative importance of the impact of these issues on the actual decision to roll-out or upgrade the network. Question 1

To what extent do the following emerging technologies influence the decision to roll-out

new broadband networks or upgrade existing ones?

No

influence at

all

Little influence

Influence Large influence

Decisive influence

My knowledge is not sufficient

to answer this

question

New copper-based

technologies (e.g. xDSL,

etherloop, Ethernet first mile

over copper (EFMC))

� � � � � �

New coax-based

technologies (e.g. Narad

networks, Ethernet over

coax)

� � � � � �

New optical fiber

technologies (e.g. APON,

EPON, Gigabit Ethernet)

� � � � � �

New radio-based

technologies(e.g. Wimax, WIBRO, microwave or

millimetre wave distribution

systems)

� � � � � �

New wireless optical-based

technologies (e.g. free space optics)

� � � � � �

New hybrid radio-optical technologies (e.g. Hybrid

Fiber Radio)

� � � � � �

Development of new compression technologies

� � � � � �

336

Question 2

To what extent do technological characteristics influence the decision to roll-out new

broadband networks or to upgrade existing ones?

No

influence at all

Little

influence

Influence Large

influence

Decisive

influence

My knowledge

is not sufficient to

answer this

question

Transmission speed

(expressed as the speed in

Mbps for a typical

subscriber)

� � � � � �

Reach (expressed as a maximum number distance

between user premises and

POP guaranteeing the

desired speed)

� � � � � �

Compatibility with existing network technologies

� � � � � �

Standardisation of

technology (either formal or

de facto)

� � � � � �

Openness of standards � � � � � �

Maturity of technology � � � � � �

Technological scalability � � � � � �

Future-proofness of

technology � � � � � �

Geographical scalability � � � � � �

Flexibility of technological solution

� � � � � �

Bandwidth development on existing copper and coaxial

networks

� � � � � �

Development of new

broadband technologies � � � � � �

Development of broadband

services � � � � � �

Path dependency of choice of technology

� � � � � �

Possibility to phase out old

parts of the existing network � � � � � �

337

Question 3

To what extent do market-related characteristics influence the decision to roll-out new

broadband networks or to upgrade existing ones?

No

influence at all

Little

influence

Influence Large

influence

Decisive

influence

My knowledge

is not sufficient to

answer this

question

Density (number of –

potential – customers per

square km)

� � � � � �

Reliability of market

demand forecasts � � � � � �

Actual market demand: consumers

� � � � � �

Expected end-user demand � � � � � �

End-user demand clustering � � � � � �

Role of lead users, for instance government

� � � � � �

Development of new

broadband services � � � � � �

Sense of urgency � � � � � �

Marketing of broadband end-user services

� � � � � �

Availability of broadband

services � � � � � �

Broadband offer of

commercial competitors � � � � � �

Broadband offer of (local) governments

� � � � � �

338

Question 4

To what extent do organisational characteristics influence the decision to roll-out new

broadband networks or to upgrade existing ones?

No

influence at all

Little

influence

Influence Large

influence

Decisive

influence

My knowledge

is not sufficient to

answer this

question

Long-term strategy � � � � � �

Impact of stakeholders on

company decisions � � � � � �

Entrepreneurship � � � � � �

Vision � � � � � �

(Effect on) internal business

processes (billing, customer

care)

� � � � � �

(Effect on) internal

processes for supplying

additional equipment (set-

top boxes) or services

� � � � � �

(Effect on) Operations and

maintenance of the network � � � � � �

Timing of network upgrade � � � � � �

Company size � � � � � �

Power of holding/mother

company � � � � � �

Strong financial position

company (’deep pockets’) � � � � � �

Alternative/competing projects within the company

� � � � � �

Regulation by regulatory authorities

� � � � � �

Dependency on innovations of equipment suppliers

� � � � � �

Participation in the telecom

market by (local)

governments

� � � � � �

Availability of human

resources � � � � � �

339

Question 5

To what extent do financial characteristics influence the decision to roll-out new

broadband networks or to upgrade existing ones?

No

influence at all

Little

influence

Influence Large

influence

Decisive

influence

My knowledge

is not sufficient to

answer this

question

Expected revenues � � � � � �

Expected capital

expenditures (investments)

of network upgrade/roll-out

(CAPEX)

� � � � � �

Expected operational costs of network upgrade/roll-out

(OPEX)

� � � � � �

Cost reduction (cost

rationalisation) � � � � � �

Availability of financial resources

� � � � � �

Actual revenues (higher or lower than expected)

� � � � � �

Impact of stakeholders � � � � � �

OPEX reduction due to new and more efficient platforms

� � � � � �

Question 6

How suitable are the following investment methods to support the decision to rol- out a

new broadband network or to upgrade the existing network?

No

influence at

all

Little

influence

Influence Large

influence

Decisive

influence

I do not know

this investment

method

Cost-benefit analysis

(assessment of CAPEX

(Capital expenditures) and OPEX (operational

expenses) and revenues)

� � � � � �

Return on Investment

method (ROI) � � � � � �

Net Present Value approach (NPV)

� � � � � �

Real Options Analysis approach (ROA)

� � � � � �

Other method � � � � � �

If other: ______________________________________

340

Categories of risk and uncertainty

An important issue in the decision-making process is the assessment of risks and uncertainties and the question as to how to deal with them. Risks and uncertainties can be caused by a lack of sufficient knowledge (which can be reduced by gaining more knowledge on the subject) or by the inherent characteristics of the environment (which cannot be reduced by gaining more knowledge about the subject). Question 7

Could you to indicate which risks and uncertainties are, in your opinion, caused by

imperfect knowledge and which are inherent in the environment?

Risks and uncertainties are due to

imperfect knowledge

Risks and uncertainties are due to

characteristics of the environment

Network and services

Innovation and development

of new broadband

technologies

� �

The rate of new

technological developments � �

Availability of alternative technological solutions

� �

Future-proofness of

technology � �

Using non-standardised

technologies � �

Path dependency of choice

of technology � �

Proper functioning of

intelligence within the network

� �

Development of broadband

end-user services � �

Operations

Manageability of QoS of

service delivery and network � �

Impact of new broadband

networks on operational systems

� �

Right timing of

availability/delivery of

necessary supplies

� �

Quality of necessary

supplies � �

Finance

Price of necessary supplies

Availability of financial

resources � �

Unforeseen costs � �

Revenues � �

341

Development equipment

price � �

End-users

Consumer

preferences/characteristics � �

Development of demand � �

End-user tariffs � �

Market

Market developments � �

Participation in the

broadband market by (local) governments

� �

Actions of potential or

actual competitors � �

Other � �

Availability of necessary

human resources � �

Effects of being first mover � �

Uncertainty about adoption

of standards � �

Unclear/inconsistent

regulation � �

Changes in regulation � �

342

Level of risk and uncertainty

A second aspect of risk and uncertainty is their level of predictability. Risks and uncertainties can be predictable at several levels, varying from a predictable future to true ambiguity (completely unpredictable) Question 8

As far as broadband projects are concerned, how would you judge the level of

uncertainty or risks for the following topics?

Near future is predictable

There is a limited number of

possible future

scenarios

There is a broad range of possible

future scenarios

The future is completely

unpredictable

Network and services

Innovation and development

of new broadband

technologies

� � � �

The rate of new

technological developments � � � �

Availability of alternative

technological solutions � � � �

Future-proofness of

technology � � � �

Using non-standardised

technologies � � � �

Path dependency of choice

of technology � � � �

Proper functioning of intelligence within the

network

� � � �

Development of broadband

end-user services � � � �

Operations

Manageability of QoS of service delivery and network

� � � �

Impact of new broadband

networks on operational systems

� � � �

Right timing of

availability/delivery of

necessary supplies

� � � �

Quality of necessary

supplies � � � �

Finance

Price of necessary supplies � � � �

Availability of financial

resources � � � �

Unforeseen costs � � � �

Revenues � � � �

343

Development equipment

price � � � �

End-users

Consumer

preferences/characteristics � � � �

Development of demand � � � �

End-user tariffs � � � �

Market

Market developments � � � �

Participation in the

broadband market by (local) governments

� � � �

Actions of potential or

actual competitors � � � �

Other � � � �

Availability of necessary

human resources � � � �

Effects of being first mover � � � �

Uncertainty about adoption

of standards � � � �

Unclear/inconsistent

regulation � � � �

Changes in regulation � � � �

344

Evolutionary versus revolutionary upgrade

With regard to network upgrading and roll-out, a telecom company can choose between an evolutionary or a revolutionary roll-out strategy. An evolutionary strategy is defined as an incremental technological upgrade aimed at increasing the capacity of existing networks (invest-as-you-grow model). A revolutionary strategy is defined as introducing or replacing the existing network by substantially new technologies. Question 9

Considering your own position, would you opt /advise others to opt in favour of an

evolutionary or revolutionary upgrade/roll-out strategy?

� Evolutionary upgrade/roll-out strategy � Revolutionary upgrade/roll-out strategy Question 10

Considering your own position, in which of the following technologies would you prefer

to invest/advise others to invest?

� New copper-based technologies (e.g. xDSL, etherloop, Ethernet first mile over copper (EFMC)) � New coax-based technologies (e.g. Narad networks, Ethernet over coax) � New optical fiber technologies (e.g. EPON, APON, Gigabit Ethernet) � New radio-based technologies, (e.g. Wimax,, WIBRO, microwave or millimetre wave distribution systems) � My knowledge is not sufficient to answer this question � Other: ______________________________________ Question 11

Considering your own position, what percentage of your freely available budget would

you invest/advise others to invest in the technology selected in question 10?

________ (Min. 0 - Max. 100) Question 12

How suitable are the following methods when it comes to reducing risk and uncertainty?

Not suitable at all

Not very suitable

Somewhat suitable

Mostly suitable

Very suitable

Technological analysis � � � � �

Cost-benefit analysis � � � � �

Obtaining information from

suppliers � � � � �

Scenario analysis � � � � �

Demand & revenue

forecasting � � � � �

Benchmarks with competitors

� � � � �

345

Reconsidering a decision to delay a network upgrade

Sometimes there are several circumstances that demand a decision to delay the roll-out or upgrade of a network. Examples of these circumstances are market-related/economic developments, disappointing demand, a lack of financial resources and other priorities within the company. In case the decision to upgrade/roll-out has been delayed, a number of issues may play a roll in reconsidering this decision. Question 13

Could you indicate to what degree the following issues have an impact on the

reconsideration of the initial decision to delay?

No impact on

reconsidering

decision

Small impact

on

reconsidering

decision

Impact on

reconsidering

decision

High impact

on

reconsidering

decision

Decisive

impact on

reconsidering

decision

Technological innovations � � � � �

Innovations in broadband

services requiring high

bandwidth

� � � � �

Increasing customer demand � � � � �

Change in the behaviour of

suppliers � � � � �

Change in the behaviour of competitors

� � � � �

Potential breakthrough of a

killer broadband service � � � � �

Change inthe behaviour of

government � � � � �

Increasing revenues � � � � �

Internal business strategy � � � � �

346

Question 14

Assuming that a network roll-out or upgrade is taking place, the real viability of the

network depends on the number of users and on the commercial performance of the

network. Could you indicate to what degree the following factors determine network

viability?

No impact on network

viability

Small impact on network

viability

Impact on network viability

High impact on network

viability

Decisive impact network viability

Competitive pricing � � � � �

Broadband offering by competitors

� � � � �

Marketing campaigns focused on broadband

� � � � �

Satisfaction of lead users � � � � �

Brand name (image) � � � � �

Critical mass of available ‘real’ broadband services

� � � � �

Critical mass of consumers using broadband

� � � � �

Triple play concepts � � � � �

3rd party end userend-user service promotion

� � � � �

The time it takes to change to higher bandwidths (conversion rate)

� � � � �

Question 15

End-user demand and adoption is an important factor in broadband roll-out decisions.

Can you indicate to what extent end-user demand and adoption influences the following

factors?

No influence at all

Little influence

Influence Large influence

Decisive influence

Payback time of investments

� � � � �

Choosing an evolutionary path strategy (invest-as-you-grow model)

� � � � �

Acceleration of network upgrade/roll-out

� � � � �

Delay of network upgrade/roll-out

� � � � �

Revenues � � � � �

Bandwidth upgrade of the existing network

� � � � �

347

Measuring success of broadband network upgrade projects

There are several criteria to determine whether or not a broadband network upgrade project is successful. In the following questions, we would like you to indicate when you would consider a broadband project successful. Question 16

What return on investment (between 0% and 100%) would you expect after TWO YEARS

for you to consider a broadband network upgrade project successful?

________ (Min. 0 - Max. 100) Question 17

What return on investment (between 0% and 100%) would you expect after FIVE

YEARS for you to consider a broadband network upgrade project successful?

________ (Min. 0 - Max. 100) Question 18

What percentage of the potential target groups should have a connection to the upgraded

broadband network after THE FIRST YEAR for you to consider a broadband upgrade

project successful?

________ (Min. 0 - Max. 100) Question 19

What percentage of the potential target groups should have a connection to the upgraded

broadband network after FIVE YEARS for you to consider a broadband upgrade project

successful?

________ (Min. 0 - Max. 100)

348

Outcomes and effects of network upgrade/roll-out

In the final question on broadband roll-out, we would like you to rate the importance of the outcomes and effects of network upgrade/broadband roll-out to a telecommunication company. Question 20

What tangible and intangible benefits and/or (social) effects are important as a result of

broadband roll-out/network upgrade for a telecommunication company?

Not important at

all

Somewhat important

Important Mostly important

Very important

Clear revenues � � � � �

Strategic advantage over competitors

� � � � �

Next step in broadband roll-out achieved

� � � � �

Increased Know-how and learning on how to roll-out, upgrade broadband networks (learning effects)

� � � � �

Stimulation for further investments

� � � � �

Acceleration of broadband roll-out

� � � � �

Next step in broadband roll-out achieved

� � � � �

Social benefits: e-learning, teleworking

� � � � �

Cultural benefits: high quality content

� � � � �

Increased communication between people

� � � � �

Increased digital divide � � � � �

Decreased digital divide � � � � �

Customer satisfaction � � � � �

Image � � � � �

Increased customer base � � � � �

Innovations in broadband services

� � � � �

Stimulation for changed network configuration

� � � � �

349

Closing questions Finally, we would like to ask you some questions concerning your background, etc. Question 21

Do you work as.. � Decision-maker for a telecom infrastructure provider � Decision-maker for a cable provider � Decision-maker for another type of infrastructure provider � Decision-maker for a DSL provider � Decision-maker for a content provider � Decision-maker for an equipment supplier � Advisor, consultant � Regulator or policy-maker � Researcher � Other: ______________________________________ Question 22

What is your current job title?

Question 23

How long have you worked in the telecommunications sector?

Question 24

Would you be alright with me mentioning your name, job title and company name in my

final PhD thesis as a participant in this survey? � Yes � No If you have no objections, please indicate your own name and company name as you wish them to be included in my thesis (This information will not be linked to your answers): Question 25

Would you like to receive the final results of this research? � Yes � No If you do, please leave your e-mail address here: Question 26

Do you have any suggestions about issues that you felt should have been included in this

questionnaire?

350

351

Annex E Articles and papers included in the Meta-

analysis

Adachi, F. (2002). Evolution towards broadband wireless systems. Proceedings of The 5th International Symposium on Wireless Personal Multimedia Communications, 2002, 1, 19 - 26.

Antunes, C. H., J.F.Craveirinha, and Joao N. Climaco (1997). Planning the evolution to broadband access networks: A multicriteria approach. European

journal of operational research, 109 (2), 530-540.

Auer, E., A.Jugel (1996). DBAS- A Digital Broadband Access System for flexible introduction of broadband services. Conference proceedings of ICCT ‘96: International Conference on Communication Technology, 2, 743-746.

Ball, P., S. Pinto, V. Barker, I. Cooper, M. Spencer, N. Wood, N., S. Unagami, and H. Yamashita (1998). Future access networks for high bandwidth services. Conference proceedings of the 6th IEE Conference on Telecommunications (Conf. Publ. No. 451), 230-234.

Bernstein, L. and C. M. Yuhas (1997). Managing the last mile. IEEE

Communications Magazine 35(10), 72 –76.

Christodoulou, K. and K. Vlahos (2001). Implications of regulation for entry and investment in the local loop. Telecommunications Policy, 25 (10-11), 743–757.

Cook, A., P. Rosher, and J. Stern (1993). Evolution options for broadband residential networks. Proceedings of the IEEE 5th Conference on Optical/Hybrid Access Networks, 6.01/01 - 6.01/06.

Cuchran, J. and R. Roka (2001). The evolution of the access network's infrastructure for the expansion of new broadband services and applications. Conference proceedings of EUROCON2001: International Conference on Trends in Communications, 2, 446-449.

Deventer Oskar van M., Y.M. van Dam, P.J.M. Peters, F. Vermaerke, and A.J. Phillips (1997). Evolution phases to an ultra-broadband access network: results from ACTS-PLANET. IEEE Communications Magazine 35(12), 72-77.

352

Driel, v. C. J., A.M van Grinsven, V. Pronk, and W.A.M. Snijders (1997). The (r)evolution of access networks for the information superhighway. IEEE

Communications Magazine 35(6), 104-112.

Elnegaard, K. S. (2002). Deciding on the right timing of VDSL roll-outs: A real options approach. Conference proceedings of ISSLS2002: International Symposium on Subscriber Loops and Services.

Elnegaard, N K, L A Ims, and K Stordahl (2000). Roll-out strategies for the copper access network- evolution towards a full service access network. Conference proceedings of ISSLS2002: International Symposium on Subscriber Loops and Services.

El-Sayed, M., J. Jaffe (2002). A View of Telecommunications Network Evolution. IEEE Communications magazine 40(12), 74-81.

Ferreira, P., W. Lehr, et al. (2002). Optical networks and the future of broadband services. Technological Forecasting and Social Change 69(7), 741-758.

Foros, O. (2004). Strategic Investments with Spillovers, Vertical Integration and Foreclosure in the Broadband Access Market. International Journal of

Industrial Organisation, 22,1-24.

Gillett, S. E., W. H. Lehr, et al. (2004). Local government broadband initiatives.Telecommunications Policy, 28 (7-8), 537-558.

Glass, V., Talluto, S., Babb, C. (2003). Technological breakthroughs lower the cost of broadband service to isolated customers. Government Information

Quarterly 20 (2), 121-133.

Grabenhorst, R., H. Grusdt, et al. (1993). A FITL system for deployment of optical fibers in Germany (OPAL 93)- a basis for evolution of access networks. Conference proceedings of ISSLS1993: International Symposium on Subscriber Loops and Services.

Haag, T. (2002). Steps from broadband access to broadband services. Conference proceedings of ISSLS2002: International Symposium on Subscriber Loops and Services.

Henriques, B., B. T. Olsen et al. (2002). DSL roll-out strategies based on group segmentation. Conference proceedings of ISSLS2002: International Symposium on Subscriber Loops and Services.

Ims, L. A. (1996). Multiservice access network upgrading in Europe: a techno-economic analysis. IEEE Communications Magazine 34(12), 124-134.

Ims, L. A., D.Myhre, et al. (1997). Economics of residential broadband access network technologies and strategies. IEEE Network 11(1), 51-57.

353

Ims, L. A., D.Myhre, et al. (1998). Costs of upgrading the residential telecommunications infrastructure to broadband. Conference proceedings of GLOBECOM'98: Global telecommunications conference, The Bridge to Global Integration, 6, 3153 – 3158.

Ims, L. A., B. T. Olsen, et al. (1996). Cost benefits and business opportunities of high capacity optical broadband access network upgrade technologies. Conference proceedings of ECOC’96:. ECOC '96. 22nd European Conference on Optical Communication, 2, 281- 284.

Ims, L. a., K. Stordahl, et al. (1997). Risk analysis of residential broadband upgrade in a competitive and changing market. IEEE Communications Magazine

35(6), 96-103.

Javed, A., P. O'Keilly, et al. (1994). Wireless technology evolution and impact on the access network. International Conference on Personal Wireless Communications, 12-16.

Kamada, T., Kawai, S., (1989). An algorithm for drawing general undirected graphs. Information Processing Letters 31(1), 7-15.

Kawata, O. S., I. Okada, K. (1997). Access network evolution scenario and key technological concepts for the broadband network. Conference proceedings of GLOBECOM'97: Global telecommunications conference, 3, 1471 – 1475.

Learned, J. (2002). Hybrid fiber/coax options explode for last mile. Communication systems design, August 2002, 28-31.

Lee, J. H., S.M. Shin, M.K. Choi (1993). Strategic planning and considerations of broadband access networks. Conference proceedings of TENCON'93: IEEE Region 10 Conference on Computer, Communication, Control and Power Engineering, 3, 241-245.

Ng, P. T. A., D. Lu, et al. (2004). A stretgic perspective on future resididential broadband market. Technovation 24(8), 665-669.

Olsen, B. T., et al. (1996). Techno-economic evaluation of narrowband and broadband access network alternatives and the evolution scenario assesment. IEEE Journal on selected area's in communications 14(6), 1184-1203.

354

355

Nederlandse samenvatting

Het zwarte gat belicht

De uitrol van breedbandige toegangsnetwerken door private operators Dit onderzoek bevat zowel een content als een proces kant. De hoofdstukken 2 en 3 richten zich op de content zijde van het theoretisch kader en focussen op de belangrijkste drijfveren en uitkomsten van breedband uitrol in de technologische, markt- en beleidsdimensies. De hoofdstukken 4 en 5 vormen het procesdeel van het theoretisch kader en richten zich op het besluitvormingsproces rondom breedband uitrol in toegangsnetwerken en op de impact van risico’s en onzekerheden op deze besluitvorming. De hoofdstukken 6 en 7 richten zich op de validatie van zowel de content als de proces kant van het theoretische kader, door middel van kwalitatieve en kwantitatieve data analyse.

Introductie

Hoofdstuk 1 introduceert de hoofdonderzoeksvraag en de onderliggende deelvragen. Het hoofdstuk verschaft inzicht in zowel de sociale als de economische relevantie van breedband uitrol en de nationale Europese publieke initiatieven om de uitrol van breedband in de local loop (het toegangsnetwerk) te stimuleren. Ondanks deze publieke aandacht en de daaruit voortgekomen initiatieven, bevindt de uitrol van glasvezel in toegangsnetwerken zich in Nederland pas in zijn beginfase. Een symmetrische dataconnectie van 10 Mb/s, zoals breedband binnen dit onderzoek wordt gedefinieerd, wordt tot op heden nog niet aangeboden in de Nederlandse residentiële markt. De vertraging in uitrol van deze netwerken wordt door zowel de nationale als Europese overheden als een probleem beschouwd, omdat de belangrijke economische en sociale voordelen voor de maatschappij niet behaald worden. Investeringen in netwerken dragen een hoog risico met zich mee ten gevolge van specifieke karakteristieken van de infrastructuur en in het bijzonder de local loop, het zogenaamde ´kip-eiprobleem´ en onzekere en onvoorspelbare eindgebruikersvraag. Additionele complexiteit wordt veroorzaakt door de multi-actor en multidisciplinaire omgeving waarin telecom operators zich bevinden. De huidige beleidsinitiatieven hebben niet het gewenste effect gesorteerd, namelijk de uitrol van breedbandige toegangsnetwerken door private operators. Het verwerven van meer inzicht in de complexe relaties tussen technologische, financiële, markt- en beleidsgerelateerde factoren en de manier waarop zij breedband uitrol door private operators beïnvloeden, is daarom van groot belang. Daarbij is inzicht in het besluitvormingsproces noodzakelijk om te begrijpen hoe de verschillende drijvende

356

factoren uiteindelijk leiden tot eindbeslissingen. Als resultaat hiervan is het doel van dit onderzoek: “Het ontwikkelen van een geïntegreerd multidisciplinair model dat

inzicht biedt in de drijfveren en uitkomsten van breedband uitrol in de local loop door

private operators en om de manier waarop besluitvorming over deze uitrol plaatsvindt te

begrijpen en te verklaren.” De bijbehorende onderzoeksvraag luidt: “Welke content en procesgerelateerde

aspecten bepalen breedband upgrades in de local loop, de uitkomsten van breedband uitrol

alsmede het besluitvormingsproces rondom het upgraden van breedbandige

toegangsnetwerken?”

Domein beschrijving

Hoofdstuk 2 focust op de content zijde van dit onderzoek en verschaft meer inzicht in het breedband domein. Het beschrijft het speelveld en de voornaamste actoren die zich in dit veld bewegen. Met name drie domeinen zijn zichtbaar binnen de omgeving van besluitvormers rondom breedband: technologie, markt en beleid & regulering. Het marktperspectief beschrijft de aanbodzijde (concurrenten en apparatuur leveranciers) en de vraagzijde (eindgebruikers). De Nederlandse markt wordt gekarakteriseerd door een sterke (duopolistische) infrastructuur concurrentie tussen opgewaardeerde koper (ADSL) en kabel (HFC) netwerken. Desalniettemin leveren verscheidene operators toegangsdiensten aan eindgebruikers op zowel het koper als kabelnetwerk. De drie voornaamste oorzaken van de sterke concurrentie tussen koper en kabelnetwerken in Nederland zijn het naast elkaar bestaan van twee vaste netwerken met nationale dekking, toegangsregulering en convergentie van data, spraak en beelden. De technologische dimensie van dit hoofdstuk behandelt mogelijke technologieën voor het upgraden van de huidige netwerken naar de volgende generatie breedband netwerken en deze technologieën worden vergeleken op basis van relevante criteria. Daarnaast worden de mogelijke migratiepaden richting Fiber-to-the-Home (FttH) netwerken behandeld. De meest waarschijnlijke ontwikkelingspaden zijn de separate ontwikkelingen van evolutionaire paden in koper TP (via VDSL) en coax netwerken richting FttC. Breedband Fixed Wireless toegangsnetten, zoals Wimax, worden beschouwd als een aanvullende of alternatieve oplossing voor high-speed kabel en DSL varianten zoals ADSL2+ en VDSL en als een kosteneffectieve oplossing in gebieden die buiten het bereik liggen van kabel en ADSL netwerken. Fixed wireless access (FWA) netwerken als Wimax, LMDS en MMDS kunnen bovendien functioneren als back-up lijn in het geval high-speed vaste netwerken rendabel en uitgerold zijn of op het moment dat de geografische uitrol van high-speed netten de gebieden heeft bereikt waarin FWA lijnen functioneerden als tijdelijke oplossingen. Om de uitrol van FttH te stimuleren, is door (lokale) overheden tussen 2002 en 2005 een groot aantal pilot- en onderzoeksprojecten uitgevoerd. Ondanks deze

357

inspanningen loopt de uitrol van glasvezel infrastructuur in de local loop duidelijk achter ten opzichte van landen als Japan en Korea. De hoge penetratie glasvezel in de toegangsnetwerken in deze landen kan voornamelijk worden toegeschreven worden aan hoge publieke investeringen en financiële stimulansen. Binnen Europa is dergelijke staatssteun echter niet zomaar toegestaan. De beleids- en reguleringsdimensie gaat in op (toegangs-) regulering en wetgeving, breedband stimuleringsprojecten en regulering van publieke glasvezelinitiatieven.

De ontwikkeling van een conceptueel model voor breedband uitrol: een

‘black hole’

Hoofdstuk 3 verschaft inzicht in de analyse van bestaande literatuur over content gerelateerde factoren die de besluitvorming omtrent breedband uitrol beïnvloeden. In dit hoofdstuk is gekeken naar de beschikbare breedband literatuur om te onderzoeken of een kader of theorie voor breedband uitrol beschikbaar is binnen de huidige literatuur. Dit hoofdstuk beschrijft eerst de toegepaste methodologie, bestaande uit Meta-analyse, content analyse en netwerk analyse. Daarna worden de resultaten van deze analyses en de ontwikkeling van het conceptuele model behandeld. Als resultaat van de netwerkanalyse hebben we gevonden dat de meeste concepten zeer sterk en nauw gerelateerd zijn, hetgeen zich uit in een dichte Euclidische ruimte. Deze zogenaamde ‘black hole’ representeert in min of mindere mate het gebrek aan een coherente en gerichte conceptualisatie. Deze Meta-analyse van breedband literatuur heeft ons geleerd dat er geen wetenschappelijk, geïntegreerd multidisciplinair kader of een formele theorie bestaat om te beschrijven hoe technologische, financiële, markt en beleidsfactoren migratie in local loop breedbandnetwerken beïnvloeden. De literatuur analyse in dit onderzoek toont tevens aan dat er geen theorie bestaat die complexe besluitvormingstheorie combineert met ontwikkelingen in breedband technologie en innovatie. Hoofdstuk 3 concludeert dat, hoewel technische, regulerings- en eindgebruikersaspecten een rol spelen, het accent binnen de aannamen en proposities met betrekking tot breedband uitrol ligt op een sterk financieel karakter. De Meta-analyse heeft geresulteerd in een eerste conceptueel model voor breedband uitrol.

Investeringen in infrastructuur: omgaan met onzekerheid en

onomkeerbaarheid

Hoofdstuk 4 richt zich op de proceskant en verschaft inzicht in het financiële karakter van investeringsbeslissingen rondom breedband netwerken en de risico’s en onzekerheden die samenhangen met deze investeringsbeslissingen. Er zijn verscheidene methoden beschikbaar voor operators om de waarde van zakelijke investeringsmogelijkheden met betrekking tot local loop infrastructuren te beoordelen. De traditionele, regulier toegepaste methoden zijn de methoden van

358

discounted cash flow (NPV en RoR) en ratio (ROI en KBA) worden uitgelegd. Tevens wordt meer inzicht verschaft in de methode van Reële Optie Analyse (ROA). De relevantie van ROA voor besluitvorming binnen de telecom markt is aangetoond in de literatuur maar ondanks dat wordt de methode nog niet frequent toegepast in de praktijk. Literatuur laat zien dat in het geval van grote technologie investeringen onder sterke technologische en markt gerelateerde onzekerheid, ROA een geschiktere methode zou kunnen zijn om netwerk investeringen te vergelijken dan de discounted cash flow methode. Als gevolg van haar afwijkende manier om risico’s en onzekerheden te beoordelen kan de ROA methode significante impact hebben op besluitvormingsprocessen rondom breedband upgrades en keuzes tussen technologieën. De risico’s en onzekerheden waar telecom operators mee geconfronteerd worden gedurende het besluitvormingsproces rondom het upgraden en uitrollen van hun breedband netten hebben verschillende oorzaken. Enerzijds zijn zij intrinsiek aan het karakter van grote infrastructurele projecten, anderzijds worden ze veroorzaakt door (onbekend) gedrag van concurrenten, (onbekende) eindgebruikersvraag, snelle technologische ontwikkelingen en verschillende mogelijke technologische ontwikkelingspaden. We hebben aangetoond dat een combinatie van de onzekerheidstypologieën van Meijer et al. (2006) Walker et al. (2003) en Courtney et al. (1997) een bruikbaar kader vormen om risico’s en onzekerheid te categoriseren naar soort, niveau, en bron (nature, level, en source). Tenslotte zijn verschillende strategieën uitgelegd die operators kunnen toepassen om risico’s en onzekerheden in hun besluitvormingsproces te verminderen.

Strategische besluitvorming in een complexe, dynamische en onzekere

omgeving

Hoofdstuk 5 verschaft meer inzicht in de typen besluiten en in het besluitvormingsproces rondom breedband uitrol. We concludeerden in hoofdstuk 5 dat, gebaseerd op de theorieën van Simon (1960) en de Baas (1998) besluiten over breedband uitrol zowel strategisch (non pogrrammed) als complex zijn. Deze complexiteit komt naar voren in vele, onderling afhankelijke, actoren met verschillende, conflicterende belangen en in de vele technologische, organisatorische, regulerings- en marktgerelateerde factoren. Er zijn verscheidene besluitvormingstheorieën voorhanden om strategische besluitvormingsprocessen te beschrijven. Na vergelijking van de karakteristieken van de breedbandmarkt met die van verschillende besluitvormingstheorieën, hebben we twee theorieën geselecteerd die geschikt leken als kader voor ons onderzoeksmodel. De eerste is het Garbage Can model van Cohen, March en Olsen (1972). Op het eerste gezicht is er een goede fit met de gedefinieerde criteria van georganiseerde anarchie, waaraan het Garbage Can model geacht wordt te voldoen. Op basis van de overeenkomsten tussen de theorie van het Logisch Incrementalisme (Quinn, 1980,

359

Das & Teng, 1999) en Reële Optie Analyse, hebben we aangenomen dat tevens de theorie van het Logisch Incrementalisme toegepast kan worden om het besluitvormingsproces rondom breedband uitrol te onderzoeken.

Het ontdekken van de ratio achter breedband uitrol - kwalitatieve

analyse

Hoofdstuk 6 richt zich op het kwalitatieve deel van de validatie. Het kwalitatieve onderzoek omvatte interviews met 12 afgevaardigden van grote Nederlandse telecom- en kabeloperators. Het hoofdstuk omvat de validatie van zowel content als proces gerelateerde aspecten. Proces Binnen de procesvalidatie zijn de toepassingsmogelijkheden van verschillende theorieën, zoals geïntroduceerd in de hoofdstukken 4 en 5, getest. We hebben geconcludeerd dat zowel het verschil tussen niet geprogrammeerde (non programmed) en geprogrammeerde (programmed) besluiten, zoals geïntroduceerd door Simon (1960), als een verdere opsplitsing tussen korte termijn operationele besluiten en periodieke controle besluiten (Lin, 2004) goed toepasbaar zijn op de verschillende besluiten die onderdeel uitmaken van het proces van breedband upgrade en uitrol. Financiële theorie is gevalideerd door de toepassing van financiële beoordelingsmethoden te onderzoeken. We concludeerden dat binnen alle geconsulteerde bedrijven investeringen in verschillende upgrade alternatieven worden beoordeeld aan de hand van de NPV methode, al dan niet in combinatie met andere methoden. De methode van Reële Optie Analyse (ROA) wordt door geen van de geconsulteerde bedrijven toegepast. Het eerste deel van ons theoretisch kader betreffende risico en onzekerheid is gevalideerd door middel van het meten van de toepasbaarheid van de zes bronnen (sources) van risico en onzekerheid, zoals geformuleerd door Meijer et al. (2006). Interviewresultaten bewijzen dat deze zes bronnen van risico en onzekerheid duidelijk ervaren worden door besluitvormers rondom breedband. Om risico’s en onzekerheden te verminderen worden technologische analyses, forecasting en pilots als methoden het meest toegepast. De toepasbaarheid van twee besluitvormingstheorieën, i.e. het Garbage Can model van Cohen, March & Olsen (1972) en de theorie van het Logisch Incrementalisme (Quinn, 1980; Das & Teng, 1999) zijn gevalideerd. Onze analyse heeft aangetoond dat telecombedrijven gekwalificeerd kunnen worden als georganiseerde anarchieën zoals bedoeld in de theorie van Cohen et al. uit 1972. Onze resultaten hebben bovendien aangetoond dat besluitvormingsprocessen rondom de uitrol van breedband dezelfde karakteristieken vertonen als beschreven in de Garbage Can theorie van Cohen et al. (1972). Gedurende de interviews zijn verscheidene voorbeelden gegeven die de drie karakteristieken van het Logisch Incrementalisme, zoals geformuleerd door Das & Teng (1999), onderbouwen. Het

360

is tevens duidelijk geworden dat strategische besluitvorming een stapsgewijs proces is.

Content Met betrekking tot content is de impact van technologische, organisatorische, financiële, en marktgerelateerde factoren op het besluitvormingsproces gemeten. Binnen de technologische dimensie zagen we dat IP, Ethernet en compressietechnieken beschouwd worden als de belangrijkste technologische ontwikkelingen. Keuzes tussen technologieën worden gemaakt op basis van vele karakteristieken van deze alternatieven. Binnen het marktdomein zagen we dat concurrentie een uiterst belangrijke factor is in het besluiten om netwerken te upgraden. Een sterke onderlinge afhankelijkheid tussen operators resulteert in een nauwe, niet altijd vrijwillige, samenwerking tussen deze partijen. Eindgebruikersvraag is een belangrijke drijfveer voor upgrades, alhoewel het imago, gerelateerd aan het aanbieden van hoge bandbreedte, nog belangrijker blijkt te zijn. Regulering beïnvloedt duidelijk het besluitvormingsproces. Het is echter niet leidend in de strategie van operators maar wordt beschouwd als een gegeven kader waarbinnen operators moeten functioneren. Als laatste onderdeel van de validatie van contentgerelateerde factoren is een validatie van de variabelen (concepten) van het conceptueel model, zoals ontwikkeld in hoofdstuk 3, gepresenteerd. We hebben aangetoond dat, op basis van kwalitatief onderzoek alleen, een ‘parsimonious’ model voor de drijfveren en uitkomsten van breedband uitrol niet ontwikkeld kan worden.

Naar een verklarend, ‘parsimonious’ model46

- kwantitatieve analyse

Hoofdstuk 7 behandelt het kwantitatieve gedeelte van de validatie van het theoretisch kader. De procesgerelateerde aspecten zijn gevalideerd door middel van descriptieve resultaten. Aanvangspunt van de kwantitatieve validatie van het content gedeelte is het conceptueel model dat resulteerde uit de Meta-analyse (hoofdstuk 3). De kwantitatieve validatie bestaat uit twee online vragenlijsten: een voorstudie en een hoofdstudie. De hoofdstudie omvat ten eerste een confirmatory factor analyse om een meetmodel te creëren. Dit meetmodel vormt vervolgens de basis voor een structural equation model.

Proces In dit hoofdstuk zijn verschillende theorieën getoetst. Financiële theorie is gevalideerd door middel van het onderzoeken van de toepasbaarheid van verschillende financiële beoordelingsmethoden. Kosten-batenanalyse, Return on Investment (ROI) en de Net Present Value (NPV) methoden worden beschouwd als zijnde merendeels of zeer goed toepasbaar. Aan kosten-batenanalyse wordt het meest de voorkeur gegeven. De methode van reële optie analyse (ROA) is een

46 Binnen deze context kent de term ‘parsimonious’ geen letterlijke vertaling in het Nederlands. De term kan het best omschreven worden als een zo optimaal mogelijke balans tussen eenvoud en verklaarbaarheid van een model.

361

minder bekende methode en de meningen over de toepasbaarheid van deze methode zijn sterk verdeeld. Het overgebleven deel van ons theoretisch kader voor risico en onzekerheid (hoofdstuk 4) werd gevalideerd door van verschillende typen risico’s en onzekerheden soort (nature) en niveau (level) te meten. Onze resultaten geven aan dat risico’s en onzekerheden die gerelateerd zijn aan technische en operationele netwerk aspecten de duidelijkste ‘epistemic’ of kennisonzekerheden zijn (reduceerbaar door middel van het vergaren van meer kennis). Gedrag van reguleringsinstanties en eindgebruikers zijn duidelijke variabiliteitsonzekerheden (niet reduceerbaar door het vergaren van meer kennis). Reguleringsinstanties zijn echter beter voorspelbaar dan eindgebruikers in hun doen en laten. Aspecten gerelateerd aan eindgebruikersgedrag zoals preferenties en karakteristieken, ontwikkeling van eindgebruikersdiensten en -vraag, worden beschouwd als zeer onvoorspelbaar. Content Binnen de content kant van de validatie is de impact van technologische, organisatorische, financiële en marktgerelateerde factoren op het besluitvormingsproces gemeten. Descriptieve resultaten laten zien dat van alle opkomende technologieën, nieuwe koper en glasvezeltechnologieën de grootste impact op het besluitvormingsproces hebben. Nieuwe draadloze technologieën als FSO hebben de kleinste impact. Transmissiesnelheid is het belangrijkste kenmerk voor het kiezen voor een bepaalde netwerktechnologie. Met het oog op marktgerelateerde factoren heeft dichtheid, in de zin van het aantal - potentiële- klanten per vierkante kilometer, doorslaggevende impact op het besluitvormingsproces. De organisatorische factor die het meeste invloed heeft op het besluitvormingsproces is lange termijn strategie. Er is verschil van mening tussen respondenten zichtbaar over de impact van regulering en het breedbandaanbod van lokale overheden. Verwachte opbrengsten en verwachte CAPEX zijn duidelijk de belangrijkste financiële variabelen binnen besluitvorming. Ten slotte richt dit hoofdstuk zich op het ontwikkelen van een ‘parsimonious’ model voor breedband uitrol. De constructen en onderliggende items zijn gebaseerd op de concepten van ons conceptueel model (hoofdstuk 3) en gevalideerd in onze kwalitatieve analyse van meerdere expert interviews. Het model combineert de belangrijkste, complex gerelateerde drijfveren, de belangrijkste uitkomsten van breedband uitrol en adoptietheorie. Structural equation modelling heeft geresulteerd in een robuust, verklarend, ‘parsimonious’ model voor de drijvende en resulterende factoren van breedband uitrol.

Conclusie - Het zwarte gat belicht

Hoofdstuk 8 bevat een overzicht van onze resultaten en verschaft overkoepelende concluderende opmerkingen. Gebruik makend van de resultaten van onze Meta-analyse, expert interviews en online vragenlijsten zijn in dit hoofdstuk de

362

onderzoeksvragen beantwoord die gepresenteerd zijn in hoofdstuk 1. Verschillende beperkingen van dit onderzoek met betrekking tot methodologie, afbakening en modelvorming zijn inzichtelijk gemaakt. We zien dat er een duidelijke link bestaat tussen het content en procesdeel van dit onderzoek. We kunnen concluderen dat het volgen van een evolutionaire uitrolstrategie een logisch en duidelijk verklaarbaar antwoord is op risico’s en onzekerheden in de breedbandmarkt. Deze risico’s en onzekerheden zijn het gevolg van de onvoorspelbare interactie tussen factoren op het gebied van technologie, markt, interne bedrijfsvoering en financiën. Ons onderzoek heeft uiteindelijk geleid tot meer inzicht in en een duidelijke verklaring voor het verloop van het besluitvormingsproces van private telecomoperators rondom de uitrol van breedband in de local loop. Daarnaast heeft dit onderzoek geresulteerd in een verklarend model voor de drijvende en resulterende factoren voor breedband uitrol. Uitgaande van een situatie zonder heldere conceptuele relaties en zonder beschikbare theorieën om te valideren (refererend aan het ‘zwarte gat’ in hoofdstuk 3), kan dit beschouwd worden als een unieke bijdrage aan het onderzoeksveld rondom breedband. Het onderzoek heeft enkele belangrijke implicaties voor onderzoekers, besluitvormers en reguleringsinstanties.

363

Curriculum Vitae

Marieke Fijnvandraat was born on 26 March 1979 in Leeuwarden, the Netherlands. From 1997-2003, she studied Systems Engineering, Policy Analysis and Management at Delft University of Technology and obtained her Master’s degree in 2003. Her studies included an internship at the Dutch consultancy firm Berenschot, while her Master’s thesis was carried out at Stratix Consulting Group at Schiphol, the Netherlands. From 2003-2006 she worked at Dialogic Innovatie & Interactie, a Dutch consultancy firm, focusing on the technological, financial and policy-related aspects of fiber optics roll-out by public institutes, provinces and municipalities. In the meantime, Marieke began working on her PhD project in 2004 for 2 days a week at Delft University of Technology. In addition, between 2006 and 2008 she worked as a telecom consultant at Capgemini, Telecom, Media and Entertainment in the Netherlands. In July 2008, she began working as a strategist & business developer at KPN Business Strategy and Innovation in The Hague, the Netherlands.

364

365

Publications

Journal papers

- Fijnvandraat, Marieke and Harry Bouwman (2006). Flexibility and Broadband evolution. Telecommunication Policy, 30 (8,9), 424-444.

- Bouwman, H., M. Fijnvandraat en L. van de Wijngaert (2006). White spots and black holes: Developing a conceptual model for Broadband roll-out. INFO, 8 (1), 72-90.

- Fijnvandraat, Marieke en Harry Bouwman (2003). Zes gebruiksdrempels en de local loop. I&I, 4, 24-31.

- Fijnvandraat, Marieke, Gedrang op het telefoonnet: een multifocusbenadering van internetuitkoppeling (2002). Computerrecht, 3, 131-140.

- Bouwman, Harry, Heleen de Vlaam & Marieke Fijnvandraat (2002). Internet and Interconnection Regulation. Case Studies from the USA, the UK and the Netherlands. Communications & Strategies, 47(3), 147-165.

Conference papers

- Fijnvandraat, M.L. How to deal with risk and uncertainty in the decision-making process on broadband roll-out? (2008). Proceedings of the IEEE International conference on Infrastructure systems 2008: Building networks for a brighter future, November 10-12, 2008, Rotterdam, the Netherlands.

– Fijnvandraat, M. and H. Bouwman (2008). Towards a parsimonious explanatory model for broadband roll-out. Proceedings of ITS World 2008: the 17th Biennial Conference of the International Telecom Society, Montreal, Canada.

– Fijnvandraat, M. and H. Bouwman (2007). Analyzing the complexity of decision-making in broadband network upgrading & roll-out. Proceedings of ITS Europe 2007: The 18th Regional Conference of the International Telecom Society, Istanbul, Turkey.

– Bouwman, H. and M. Fijnvandraat (2007). Exploring the rationale behind broadband roll-out - a quantitative analysis. Proceedings of ITS Europe 2007: The 18th Regional Conference of the International Telecom Society, Istanbul, Turkey.

– Fijnvandraat, Marieke L., and Harry Bouwman (2006). Risk and uncertainty in broadband roll-out. Proceedings of ITS World 2006: the 16th Biennial Conference of the International Telecom Society, Beijing, China.

366

– Fijnvandraat. Marieke, Harry Bouwman and Lidwien van de Wijngaert (2005). Broadband in perspective- The impact of technology, economy and policymaking on decision-making about broadband in the local loop. Proceedings of ITS Europe 2007: The 16th Regional Conference of the International Telecom Society, Porto, Portugal.

- Bouwman H., M. Fijnvandraat, H.I.M. de Vlaam; Internet and interconnection: Case Studies from The Netherlands, US, UK and Finland (2002). Conference proceedings of ENCIP 2002: The 17th European Communications Policy Research Conference (Barcelona, Spain, 24,25,26 March), Barcelona, Spain, p. 292-323.

Conference presentations

– How to deal with risk and uncertainty in the decision-making process on broadband roll-out? (2008). Presentation at the International conference on Infrastructure systems 2008: Building networks for a brighter future, November 10-12, 2008, Rotterdam, the Netherlands.

– Towards a parsimonious explanatory model for broadband roll-out. Presentation at ITS World 2008: the 17th Biennial Conference of the International Telecom Society, Montreal, Canada.

– Exploring the complexity of decision-making in broadband network upgrading & roll-out with quantitative & qualitative data analysis. Presentation at ITS Europe 2007: the 18th European Regional ITS Conference, September 3-5, 2007, Istanbul, Turkey.

– All IP networks: No pain, no gain. Presentation at the Capgemini TME annual global event, June 5-10, 2007, Malaga, Spain.

– Risk and uncertainty in broadband roll-out. Presentation at ITS World 2006: the 16th Biennial Its Conference, June 12-16, 2006, Beijing, China.

– Broadband in perspective, the impact of technology, economy and policymaking on decision-making about broadband in the local loop. Presentation at ITS Europe 2005: the16th European Regional ITS Conference, September 4-6, 2005 in Porto, Portugal.

– Public fiber network initiatives- questions for municipalities, provinces and housing companies. Presentation at the Euroforum congress, January 25, 2005, Amsterdam.

367

Non-scientific publications

- Marieke L. Fijnvandraat (2005). Brabant Breedband, Inventarisatie van breedbandinitiatieven in Noord-Brabantse gemeenten. Utrecht: Dialogic.

- Rudi Bekkers, Marieke Fijnvandraat, Ton van Mil (TwoMinds) en Jouko Huismans (TwoMinds) (2005). Schakelen we om? Verwachte ontwikkeling voor alternatieve kanalen voor televisieverspreiding in 2006-2008, quick scan. Utrecht: Dialogic.

- Marieke Fijnvandraat, Menno Smidts, Pim den Hertog, Rudi Bekkers en Sven Maltha (2005). Breedbandvraagstukken, beschrijvingen uit de praktijk in: Goed op weg met breedband, Handreiking voor gemeenten, provincies en woningbouwcorporaties- gezamenlijke handreiking van EZ, BZK, VROM, IPO en VNG, 49-70.

- Bert-Jaap Koops, Rudi Bekkers, Frank Bongers en Marieke Fijnvandraat (2005). Aftapbaarheid van telecommunicatie, Een evaluatie van hoofdstuk 13 Telecommunicatiewet. Tilburg, Utrecht: TILT - Centrum voor Recht, Technologie en Samenleving/Dialogic.

- Den Hertog, Pim, Christiaan Holland, Rob Bilderbeek en Marieke Fijnvandraat (2005). Quickscan economische betekenis van bredere toepassing van ICT in maatschappelijke sectoren. Utrecht: Dialogic.

- Maltha, S.R., M.L. Fijnvandraat, F.J. Bongers & R.N.A. Bekkers (2005). Innovatief frequentiemanagement, inventarisatie en analyse van kennisbehoefte en ontwikkeling. Utrecht: Dialogic.

- Maltha, S.R., M.L. Fijnvandraat, R.H. Bilderbeek (2005). Effecten breed uitgemeten. Internationale vergelijking van kosten-batenanalyses breedband. Utrecht: Dialogic.

- Drs. Christiaan Holland, Drs. Pim den Hertog, Drs. Menno Smidts, Ir. Marieke L. Fijnvandraat en Prof. Dr. Wouter Keller (Argitek) (2004).Slim kopiëren en opschalen. Verkenning naar opschaalbaarheid en kopieerbaarheid van innovatieve ICT-diensten in zorg, onderwijs en veiligheid. Utrecht: Dialogic.

- Menno Smidts, Rudi Bekkers, Ton van Mil (TwoMinds), Marieke Fijnvandraat (2004). Businessmodel breedbandambitie gemeente Amersfoort. Utrecht: Dialogic.

- Smidts, Menno, Rudi Bekkers, Marieke Fijnvandraat en Ton van Mil (TwoMinds) (2004). Businessmodel glasvezelnet Havenbedrijf Rotterdam. (Vertrouwelijk).Utrecht: Dialogic.

- Bekkers, Rudi, Marieke Fijnvandraat, Sven Maltha, Krijn Schuurman, Menno Smidts, Rens Vandeberg, Karianne Vermaas (2004). Analyse breedbandplannen publieke en private partijen- Vooruitgang zonder marktverstoring? BIJLAGE met casussen. (Vertrouwelijk). Utrecht/Amsterdam: Dialogic /SEO.

368

- Smidts, M, Marieke Fijnvandraat, Rens Vandeberg, Quickscan glasvezel G30 (2004). Resultaten van een korte inventarisatie van de activiteiten gericht op het met glasvezel ontsluiten van (semi-) publieke instellingen door de dertig GSB gemeenten (G30). Utrecht: Dialogic.

- Bekkers, R, Marieke Fijnvandraat, Sven Maltha (2004). De business case van Carrier PreSelect (CPS) aanbieders in Nederland (Vertrouwelijk). Utrecht: Dialogic.

− Bongers, F, K. Schuurman, R. Vandeberg & M. Fijnvandraat (2003). Broadband in the Netherlands. A quick scan of R&D performance. Utrecht: Dialogic.

− Fijnvandraat, Marieke L. (2003). Breedband en de local loop- Innovatie-Acceleratie-Domesticatie (Master Thesis). Delft/Schiphol: TU Delft/Stratix..

− F. van den Berg, P.J. Plug MA, R.C. van der Mark, F. van Abbema, B. van der Wal, R.M. Timmerman, M.L. Fijnvandraat (2002). Evaluatie Mededingingswet. Doelmatigheid handhaving M-Wet. Utrecht: Berenschot.

369

NGInfra PhD thesis Series on Infrastructures

1. Strategic behavior and regulatory styles in the Netherlands energy industry Martijn Kuit, 2002, Delft University of Technology, The Netherlands.

2. Securing the public interest in electricity generation markets, The myths of the invisible hand and the copper plate Laurens de Vries, 2004, Delft University of Technology, The Netherlands.

3. Quality of Service Routing in the Internet: Theory, Complexity and Algorithms Fernando Kuipers, 2004, Delft University of Technology, The Netherlands.

4. The role of power exchanges for the creation of a single European electricity market: market design and market regulation François Boisseleau, 2004, Delft University of Technology, The Netherlands, and University of Paris IX Dauphine, France.

5. The ecology of metals Ewoud Verhoef, 2004, Delft University of Technology, The Netherlands.

6. MEDUSA, Survivable information security in critical infrastructures Semir Daskapan, 2005,Delft University of Technology, The Netherlands.

7. Transport infrastructure slot allocation Kaspar Koolstra, 2005, Delft University of Technology, The Netherlands.

8. Understanding open source communities: an organizational perspective Ruben van Wendel de Joode, 2005, Delft University of Technology, The Netherlands.

9. Regulating beyond price, integrated price-quality regulation for electricity distribution networks Viren Ajodhia, 2006, Delft University of Technology, The Netherlands.

10. Networked Reliability, Institutional fragmentation and the reliability of service provision in critical infrastructures Mark de Bruijne, 2006, Delft University of Technology, The Netherlands.

11. Regional regulation as a new form of telecom sector governance: the interactions with technological socio-economic systems and market performance Andrew Barendse, 2006, Delft University of Technology, The Netherlands.

12. The Internet bubble - the impact on the development path of the telecommunications sector Wolter Lemstra, 2006, Delft University of Technology, The Netherlands.

13. Multi-Agent Model Predictive Control with Applications to Power Networks Rudy Negenborn, 2007, Delft University of Technology, The Netherlands.

14. Dynamic bi-level optimal toll design approach for dynamic traffic networks Dusica Joksimovic, 2007, Delft University of Technology, The Netherlands.

370

15. Intertwining uncertainty analysis and decision-making about drinking water infrastructure Machtelt Meijer, 2007, Delft University of Technology, The Netherlands.

16. The New EU Approach to Sector Regulation in the Network Infrastructure Industries Richard Cawley, 2007, Delft University of Technology, The Netherlands.

17. A functional legal design for reliable electricity supply, How technology affects law Hamilcar Knops, 2008, Delft University of Technology, The Netherlands and Leiden University, The Netherlands.

18. Improving Real-Time Train Dispatching: Models, Algorithms and Applications Andrea D’Ariano, 2008, Delft University of Technology, The Netherlands.

19. Exploratory modeling and analysis: A promising method to deal with deep uncertainty Datu Buyung Agusdinata, 2008, Delft University of Technology, The Netherlands.

20. Characterization of Complex Networks: Application to Robustness Analysis Almerima Jamaković, 2008, Delft University of Technology, Delft, The Netherlands.

21. Shedding light on the black hole, The roll-out of broadband access networks by private operators Marieke Fijnvandraat, 2008, Delft University of Technology, Delft, The Netherlands.

Order information: info@nginfra.nl