in the aircraft engine industry

0 downloads 0 Views 1MB Size Report
productive services available to a firm from its own resources, particularly the ...... actors involved in the industry, engine makers need to span their capabilities over a .... Industries from Japan and Samsung Aerospace from Korea (Nakamoto, 1997). ... industry has received government support via research on commercial ...
DIVIDE AND RULE: FIRM BOUNDARIES IN THE AIRCRAFT ENGINE INDUSTRY

Andrea Prencipe

Ph.D. Thesis

SPRU: Science and Technology Policy Research University of Sussex

April 2000

1

TABLE OF CONTENT LIST OF ABBREVIATIONS .................................................................................................................... 6 LIST OF TABLES ..................................................................................................................................... 7 LIST OF FIGURES ................................................................................................................................... 9 CHAPTER 1: INTRODUCTION ........................................................................................................... 10 1. 2. 3.

BACKGROUND AND THE RESEARCH QUESTIONS ............................................................................ 10 MAIN FINDINGS ............................................................................................................................ 12 THESIS CONTENT .......................................................................................................................... 13

CHAPTER 2: THE THEORETICAL FRAMEWORK ....................................................................... 15 1. 2.

INTRODUCTION ............................................................................................................................. 15 RESOURCES, COMPETENCIES, AND CAPABILITIES: HISTORY, MEANINGS, AND RELATIONSHIPS ...... 15 2.1. The resource-based perspective: a historical background...................................................... 16 Edith Penrose and the theory of the growth of the firm ................................................................................16 Business policy and firms’ resources ............................................................................................................17 Richardson: firm’s capabilities and the firm/market dichotomy ...................................................................17

2.2. The terms of contention: knowledge, resources, routines, and capabilities ............................ 18 The relationships between resources and competencies/capabilities ............................................................18 Distinctive, core, strategic and dynamic capabilities ....................................................................................21 Capabilities, endowments, and asymmetric information ..............................................................................21 Process as the mainstay of the resource-based approach ..............................................................................22

3.

THE RESOURCE-BASED VIEW: BETWEEN THEORETICAL INTERPRETATION AND NORMATIVE IMPLICATIONS ........................................................................................................................................ 23 3.1. The resource-based view within the tradition of strategic management ................................. 23 Porter and the industry attractiveness models ...............................................................................................23

3.2. The resource-based theory and other theories of the firm: knowledge vs. information .......... 25 Resource-based view vs. contractual view and transaction cost economics .................................................26

4.

THE ANALYSIS OF THE BOUNDARIES OF THE FIRM: TRANSACTION COST ECONOMICS, RESOURCEBASED APPROACH, AND PORTER’S APPROACH ........................................................................................ 27 4.1. Some concluding remarks on the resource-based approach ................................................... 29 THE DIVISION OF INNOVATIVE OF LABOUR: RELATIONSHIPS AMONG PRODUCT, ORGANISATION, 5. AND TECHNOLOGY ................................................................................................................................. 31 5.1. Modularity in products ............................................................................................................ 31 Some criteria for product decomposition ......................................................................................................31 Product architecture and the designer role ....................................................................................................32 Concluding remarks on product modularity..................................................................................................33

5.2. The relationships between product, organisational, and technological modularity ............... 34 Henderson and Clark on the distinction between architectural and component innovation ..........................34 Sanchez and Mahoney on organisational modularity ....................................................................................34 Knowledge and organisational modularity ...................................................................................................34

5.3. Technology, products and organisations: toward modular dynamics? .................................. 35 5.4. Technological and organisational co-ordination: what the management literature suggests 36 Systemic vs. autonomous innovation and the number of external sources....................................................36 Technology phase-shift: from modular to integral and vice versa ................................................................37 Technology vs. product: technological and organisational co-ordination .....................................................38 Components’ hierarchy .................................................................................................................................38

6.

CONCLUSIONS .............................................................................................................................. 40

CHAPTER 3: RESEARCH METHODOLOGY ................................................................................... 43 1. 2. 3. 4.

INTRODUCTION ............................................................................................................................. 43 THE RESEARCH PROCESS ............................................................................................................... 43 THE COMPANIES ANALYSED.......................................................................................................... 44 PRODUCT AND COLLABORATIVE AGREEMENT DATA ..................................................................... 45 4.1. Product data............................................................................................................................ 45 4.2. Collaborative agreement data ................................................................................................. 46 PATENT DATA ............................................................................................................................... 47 5. 5.1. The company dimension .......................................................................................................... 47

2

5.2. The technology dimension ....................................................................................................... 48 Information gathering ...................................................................................................................................49 Regrouping ...................................................................................................................................................49 Validity checking ..........................................................................................................................................50

5.3. The sector-specific maps ......................................................................................................... 50 Product-related patents .................................................................................................................................51 Manufacturing-related patents ......................................................................................................................51 Material-related patents ................................................................................................................................51 Testing ..........................................................................................................................................................51 New architectures .........................................................................................................................................52 Control system-related patents ......................................................................................................................52

5.4. Limitations of the patent analysis............................................................................................ 52 QUALITATIVE DATA AND CASE STUDY .......................................................................................... 53 6.1. Data sources ........................................................................................................................... 53 6.2. Collecting information through interviews ............................................................................. 54 6.3. Analysing case study evidence ................................................................................................ 56 6.4. Writing the case studies .......................................................................................................... 56 6.5. Drawbacks of interview data and case study method ............................................................. 56 CONCLUSIONS AND SUMMARY ...................................................................................................... 57 7. 6.

CHAPTER 4: AN OVERVIEW OF THE AIRCRAFT ENGINE INDUSTRY AND TECHNOLOGY....................................................................................................................................... 60 1. 2.

INTRODUCTION ............................................................................................................................. 60 THE STRUCTURAL CONTEXT OF THE INNOVATION PROCESS IN THE AIRCRAFT ENGINE INDUSTRY . 61 2.1. Mass-manufactured products vs. complex product systems industries ................................... 61 2.2. The aircraft engine industry as a CoPS industry .................................................................... 62 2.3. The structure of the innovation process in the aircraft engine industry.................................. 62 2.4. Engine makers ......................................................................................................................... 64 2.5. The innovation infrastructure ................................................................................................. 64 Risk and revenue sharing partners and specialised suppliers ........................................................................64 Government-funded laboratories: the role of national governments .............................................................65 The role of Universities ................................................................................................................................66

2.6. The innovation superstructure ................................................................................................ 67 The regulatory network imposed by national governments ..........................................................................67 The certification agencies and the professional bodies .................................................................................68 The airlines ...................................................................................................................................................69 The airframers ...............................................................................................................................................70 Airframers’ involvement during new engine programme .............................................................................71

2.7. Concluding remarks ................................................................................................................ 72 INTRODUCTION TO THE AIRCRAFT ENGINE .................................................................................... 72 3.1. A brief history of the gas turbine............................................................................................. 72 3.2. Some hints on the basic principles of the gas turbine ............................................................. 73 3.3. Engine performance and efficiency ......................................................................................... 73 3.4. Turbojet and turboprop: technical characteristics and ‘selected’ applications ..................... 74 3.5. The turbofan engine ................................................................................................................ 76 3.6. Engine efficiency and related design parameters ................................................................... 77 3.7. Factors influencing engine design .......................................................................................... 77 PRODUCT CHARACTERISTICS AND TECHNOLOGICAL REQUIREMENTS IN THE AIRCRAFT ENGINE 4. INDUSTRY ............................................................................................................................................... 79 4.1. Combustion technology ........................................................................................................... 79 4.2. Control technologies ............................................................................................................... 80 4.3. Material technologies.............................................................................................................. 80 4.4. Manufacturing techniques ....................................................................................................... 81 4.5. Computation fluid dynamics ................................................................................................... 82 4.6. Testing technologies ................................................................................................................ 82 CONCLUSIONS .............................................................................................................................. 83 5. 3.

CHAPTER 5: INTRAMURAL CAPABILITIES AND EXTERNAL LINKAGES IN THE AIRCRAFT ENGINE INDUSTRY ........................................................................................................ 86 1. 2.

INTRODUCTION ............................................................................................................................. 86 THE DRIVING FORCES.................................................................................................................... 87 2.1. The enablers ............................................................................................................................ 87

3

3. 4.

2.2. The pushers ............................................................................................................................. 90 THE CHANGING PATTERN OF DIVISION OF LABOUR ....................................................................... 93 THE PATENT ANALYSIS ................................................................................................................. 95 4.1. The Big Three .......................................................................................................................... 95 A macro overview.........................................................................................................................................95 A more detailed picture.................................................................................................................................98 A note on some relevant differences ...........................................................................................................102

4.2. The RRSP companies ............................................................................................................ 103 A macro overview.......................................................................................................................................103 A more detailed picture...............................................................................................................................105

5.

STABILITY, PERSISTENCY, AND VARIATION OF AIRCRAFT ENGINE MAKERS’ TECHNOLOGICAL PROFILES .............................................................................................................................................. 107 5.1. The relationships across companies’ technological profiles: what correlation analysis suggests........................................................................................................................................... 108 5.2. The breadth of technological capabilities ............................................................................. 109 The breadth of companies’ technological capabilities: The Herfindal Index ..............................................111

5.3. Stability and dynamics of change in the long term: some statistical evidence ...................... 112 The methodology proposed by Cantwell ....................................................................................................113 The empirical findings ................................................................................................................................116

6.

CONCLUSIONS: AN INITIAL INTERPRETATION OF THE BOUNDARIES OF ENGINE MAKERS ............. 117

CHAPTER 6: ALL-ENGINE CAPABILITIES AND CRITICAL COMPONENTS: DEFINING THE BOUNDARIES OF ENGINE MAKERS’ TECHNOLOGICAL CAPABILITIES ................ 120 1. 2.

INTRODUCTION ........................................................................................................................... 120 THE DEVELOPMENT PROCESS FOR NEW ENGINE .......................................................................... 121 2.1. The multidisciplinary character of the project teams............................................................ 123 2.2. The engine testing programme: some hints ........................................................................... 123 THE CASE OF THE INTEGRATION GROUPS .................................................................................... 124 3. 3.1. Organisation by engine subsystems and by processes .......................................................... 124 3.2. The all-engine activities ........................................................................................................ 125 3.3. The integration groups .......................................................................................................... 127 The performance group and the engine control system...............................................................................127 The Fluid System Group .............................................................................................................................128 The Air System Group ................................................................................................................................129 Concluding remarks on the integration groups ...........................................................................................131

4.

THE CASE OF THE FAN ................................................................................................................. 131 4.1. The first generation wide-chord fan blade ............................................................................ 132 4.2. The second generation wide-chord fan blade ....................................................................... 134 4.3. The Fan Key System .............................................................................................................. 135 The Electronic Product Definition: the context ...........................................................................................135 Digital Pre-Assembly: the antecedent .........................................................................................................135 The Fan Key System: an overview .............................................................................................................136 The Fan Key System: integration between design, analysis, and manufacturing ........................................136

5.

4.4. Concluding remarks on the fan subsystem ............................................................................ 137 CONCLUSIONS ............................................................................................................................ 138

CHAPTER 7: RATE OF TECHNOLOGICAL CHANGE AND THE BREADTH AND DEPTH OF TECHNOLOGICAL CAPABILITIES OF ENGINE MAKERS: THE CASE OF THE ENGINE CONTROL SYSTEM ............................................................................................................................ 140 INTRODUCTION ........................................................................................................................... 140 ENGINE CONTROL SYSTEM’S TECHNOLOGIES AND FUNCTIONS ................................................... 140 2.1. Components and functions .................................................................................................... 140 THE TECHNOLOGICAL AND FUNCTIONAL EVOLUTION OF THE ENGINE CONTROL SYSTEM............ 141 3. 3.1. The hydromechanical control system .................................................................................... 141 3.2. Presumptive anomalies: hydromechanical vs. analogue vs. digital control systems ........... 142 3.3. Supervisory systems as halfway technological and functional evolution .............................. 144 3.4. The digital engine control system.......................................................................................... 145 3.5. The anatomy of the FADEC and its main characteristics ..................................................... 146 3.6. Future control systems: distributed electronics .................................................................... 147 3.7. Concluding remarks on the technological evolution of the engine control system ............... 147 THE AIRFRAMERS, AIRLINES AND FADECS ................................................................................ 148 4. 4.1. The airframe/avionics/FADEC interface and the role of the airframer ................................ 148 1. 2.

4

4.2. FADEC’s functions and the economics of airline operation ................................................. 148 TECHNOLOGICAL SHIFT AND FIRMS’ TECHNOLOGICAL CAPABILITIES ......................................... 150 5.1. Changing importance of engine control system components ................................................ 150 5.2. The relationships between engine makers and specialised suppliers: an overview .............. 151 5.3. The specialised suppliers ...................................................................................................... 152 5.4. The engine makers ................................................................................................................ 152 5.5. Technological shift, systemic change, and fuzzy interfaces................................................... 153 ENGINE CONTROL SYSTEM’S DEVELOPMENT PROCESS AND THE DEPTH OF THE ENGINE MAKERS 6. TECHNOLOGICAL CAPABILITIES ............................................................................................................ 155 7. ARCHITECTURE AND COMPONENT KNOWLEDGE AND THE EXTENT OF THE DIVISION OF LABOUR BETWEEN ENGINE MAKERS AND SPECIALISED SUPPLIERS ..................................................................... 156 8. CONCLUSIONS: THE CHANGING BOUNDARIES OF ENGINE MANUFACTURERS’ TECHNOLOGICAL CAPABILITIES ....................................................................................................................................... 159 5.

CHAPTER 8: DIMENSIONS AND DYNAMICS OF SYSTEMS INTEGRATION ....................... 162 1. 2.

INTRODUCTION ........................................................................................................................... 162 THE SKILLS UNDERLYING SYSTEMS INTEGRATION ...................................................................... 163 2.1. Systems integration: knowledge integration vs. component assembly .................................. 163 2.2. The relevant dimensions of knowledge.................................................................................. 164 A FRAMEWORK TO ANALYSE THE BOUNDARIES OF THE FIRM TECHNOLOGICAL CAPABILITIES .... 165 3. 3.1. The synchronic dimension ..................................................................................................... 167 3.2. The diachronic dimension ..................................................................................................... 168 3.3. The relationships between synchronic and diachronic systems integration.......................... 169 3.4. The boundaries of engine manufacturers’ technological capabilities .................................. 173 CONCLUSIONS: COMPETING ON SYNCHRONIC AND DIACHRONIC SYSTEMS INTEGRATION ........... 174 4. CHAPTER 9: CONCLUSIONS ............................................................................................................ 177 1. 2.

INTRODUCTION ........................................................................................................................... 177 THE LITERATURE ADDRESSED AND THE MAIN FINDINGS ............................................................. 177 2.1. The theoretical argument ...................................................................................................... 177 2.2. The empirical evidence: implications for the strategic management of technology ............. 180 Technology outsourcing differs from product outsourcing .........................................................................180 Systems integration: synchronic vs. diachronic capability ..........................................................................181

3.

2.3. The methodological argument ............................................................................................... 183 LIMITATIONS AND FUTURE RESEARCH ........................................................................................ 183

BIBLIOGRAPHY .................................................................................................................................. 186 APPENDIX 1 .......................................................................................................................................... 192 APPENDIX 2 .......................................................................................................................................... 194 APPENDIX 3 .......................................................................................................................................... 195 APPENDIX 4 .......................................................................................................................................... 196 APPENDIX 5 .......................................................................................................................................... 197 APPENDIX 6 .......................................................................................................................................... 198

5

LIST OF ABBREVIATIONS BPR CAD CFD CMC CoPS DARPA EDP FADEC FAA FEA GE IATA ICT IHPTET JAA M&A MMC MTU NASA R&D RPS RRSP RTA SDC SiC US WWT

By-Pass Ratio Computer Aided Design Computational Fluid Dynamics Ceramic Matrix Composite Complex Product System Defense Advanced Research Project Agency Electronic Definition Process Full Authority Engine Control System Federal Aviation Administration Finite Element Analysis General Electric International Air Transport Association Information and Communication Technology Integrated High Performance Turbine Engine Technology Joint Aviation Authority Mergers and Acquisitions Metal Matrix Composite Motoren und Turbinen Union National Aeronautics and Space Administration Research and Development Relative Patent Share Risk and Revenue Sharing Partner Revealed Technological Advantage Securities Data Corporation Silicon Carbide United States Working Together Team

6

LIST OF TABLES Table 2.1 Table 3.1 Table 3.2 Table 3.3 Table 4.1 Table 4.2 Table 5.1 Table 5.2 Table 5.3 Table 5.4 Table 5.5 Table 5.6 Table 5.7 Table 5.8 Table 5.9 Table 5.10 Table 5.11 Table 5.12 Table 5.13 Table 5.14 Table 5.15 Table 5.16 Table 5.17 Table 5.18 Table 5.19 Table 5.20 Table 5.21 Table 5.22 Table 5.23 Table 5.24 Table 5.25

Components’ hierarchy……………………………………….. Collaborative agreement two-tier classification………………. Relationships among the three taxonomies…………………… Definition of the patent groups………………………………... RRSP research projects in collaboration with Universities…... Comparison of some salient aircraft parameters……………… The driving forces affecting engine makers’ technological capabilities…………………………………………………….. Engine manufacturers: number of versions per engine family 1977-1996……………………………………………………... Total number of collaborative agreements……………………. Company A’s number of collaborative agreements…………... Company B’s number of collaborative agreements…………... Company C’s number of collaborative agreements…………... RRSP A’s number of collaborative agreements.……………… RRSP B’s number of collaborative agreements………………. Financial breakdown of new engine programmes…………….. Shares of in-house design and manufacturing of aircraft engine companies……………………………………………... Big Three’s technological profiles on the 7-technological fields: total patent number 1977-1996………………………... Big Three’s technological profiles on the 7-technological fields: relative patent share 1977-1996……………………….. Company A’s technological profile on 23-technological fields: relative patent shares 1977-1996………………………. Company B’s technological profile on 23-technological fields: relative patent shares 1977-1996……………………………… Company C’s technological profile on 23-technological fields: relative patent shares 1977-1996……………………………… RRSP A’s technological profiles on the 7-technological fields: total patent number 1977-1996……………………………….. RRSP B’s technological profiles on the 7-technological fields: relative patent share 1977-1996………………………………. RRSP A’s technological profile on 23-technological fields: relative patent shares 1977-1996……………………………… RRSP B’s technological profile on 23-technological fields: relative patent shares 1977-1996……………………………… Correlation coefficients derived from companies’ technological profile in terms of patent shares in 1977-86…… Correlation coefficients derived from companies’ technological profile in terms of patent shares in 1987-96…… The distribution of engine makers and RRSPs technological capabilities according to the 7-technological field map………. Herfindal index distribution…………………………………... Results of company regression analysis of the RTA index in 1987-96 on the index in 1977-86……………………………... Results of company regression analysis of the RTA index in 1987-96 on the index in 1977-86……………………………...

51 62 77 79 90 105 118 120 122 122 123 123 124 124 125 126 130 131 132 134 135 139 139 141 142 145 145 147 149 155 156 7

Table 6.1 Table 7.1 Table 7.2 Table 7.3 Table 7.4 Table 7.5 Table 7.6 Table 7.7 Table 7.8 Table 8.1

Component efficiency loss and research activity…………….. Computing function split between electronics and hydromechanical……………………………………………… Coexistence of technologies for computing functions………... Ranking of the most critical components of the engine control system………………………………………………………… Cumulative patent shares of control system-related technologies in the Big Three…………………………………. Ideal division of labour between engine makers and suppliers.. Comparison between engine maker, first-tier supplier, and second-tier supplier: total patent number……………………... Comparison between engine maker, first-tier supplier, and second-tier supplier: patent shares……………………………. Applied research capability related to the integration of the engine control system…………………………………………. Systems integration: underlying skills………………………...

178 195 196 204 207 212 213 214 215 222

8

LIST OF FIGURES Figure 2.1 Figure 4.1 Figure 4.2 Figure 4.3 Figure 4.4 Figure 6.1 Figure 6.2 Figure 6.3 Figure 7.1 Figure 8.1

The relationships among resources, capabilities, strategy and competitive advantage………………………………………… The aircraft engine industry meso-system……………………. Broad-brush comparative propulsive efficiency……………… The Rolls-Royce Trent 800 turbofan engine…………………. The drivers of technological and organisational capabilities of engine manufacturers…………………………………………. The propulsion system definition lifecycle…………………… Organisation of the Fluid System Group and major interfaces.. Air System Group organisation and major interfaces………… Activities and actors in requirement definition, design, and manufacturing of engine control system……………………… The synchronic and diachronic dimensions of systems integration……………………………………………………...

23 85 102 103 114 162 173 175 211 240

9

CHAPTER 1: INTRODUCTION 1. Background and the research questions The issue of firm boundaries has attracted much attention in industrial practice and academic research. Practitioners are interested in understanding to what extent they can rely on firm’s external sources for the design, production, and marketing of their products. Academic researchers have tried to identify those factors that explain where the line between in-house and external activities can be most effectively drawn. Two research traditions have analysed the boundaries of the firm. The first tradition is the theoretical and empirical literature on the theory of firm. The second tradition is the composite stream of literature on the division of innovative labour. As regards the first stream, the competing theoretical paradigms on the theory of the firm provide different and diverging interpretations of firms’ boundaries. In the last decade there has been a surge of interest in the internal resources of the firm. Drawing on the seminal work of Penrose (1959), the resource-based view of the firm has emerged as an alternative paradigm for theoretical interpretations of the business organisation. In contrast to the contractual view that maintains that a firm can be conceived as a nexus of contracts, the resource-based view conceives a firm as an organised bundle of resources. This offers a completely new perspective on the firms’ boundaries. This renewed interest in the internal workings of the firm has inspired theoretical and empirical works in the normative field of strategic management (Prahalad and Hamel, 1990; Teece et al., 1992, 1999). Theoretical studies have attempted to better understand the nature of the source of firms’ competitive advantage. Terms such as competencies and capabilities have been used to describe such sources. Accordingly, firms’ heterogeneity both within and across sectors can be ascribed to differences in competencies and capabilities (Dosi et al., 1999; Patel and Pavitt, 1997). The empirical exercises have attempted to measure and ground these theoretically rich constructs. Some empirical studies have provided invaluable insights on firms’ capabilities. However, most of these studies rely on aggregate measures of capabilities that do not capture all their relevant dimensions as proposed in the theoretical literature. Furthermore, these studies have mainly focused on mass-manufactured goods leaving unexplored a large number of industries that Miller et al. (1994) and Hobday (1998) have labelled complex products systems (CoPS). The second tradition of research that deals with the boundaries of the firm is the composite stream of literature on the division of innovative labour. Recently, much of the debate has revolved around the concept of modularity. Industrial economists, management scholars, and organisation theorists have discussed the implications of this concept for inter-firm division of labour (Arora et al., 1998; Sanchez and Mahoney, 1996). It is argued that modularity is a characteristic of the product, its underlying technological knowledge, and firm’s organisation design. The conclusion drawn is that modularity in product, technology, and organisation is positively correlated and, therefore, increasing modularization entails a greater division of labour between firms at each level. However, this conflation of the product, organisation, and technology dimensions raises a number of problems in the analysis of firms developing complex, multitechnology products.

10

Building upon and departing from these two streams of literature, the aim of this thesis is to investigate the dynamics of the boundaries of firms operating in the aircraft engine industry. Firm’s boundaries can be measured and assessed along multiple dimensions. The thesis aims to analyse the extent to which aircraft engine manufacturers rely on internal or external suppliers for the design, development and manufacture of components, subsystems, materials and production equipment of the aircraft engine. The civil aircraft engine industry shows some peculiar features that render it interesting in relation to the issue of boundaries of the firm. The aircraft engine is a multicomponent, multitechnology product. This poses significant managerial challenges for aircraft engines manufacturers in terms of the technological capabilities to be developed and make-buy decisions. These challenges are exacerbated by the existence of various driving forces at work in the industry that ‘enable’ and ‘push’ engine manufacturers to externalise greater parts of their activities in relation to the research, design, development, and manufacture of new engines. Therefore, from the perspective of opportunities for and constraints to make-buy decisions this industry offers a very interesting case for analysis. It is worth underlining that the aircraft engine industry is a large industry worldwide. Due to growth in world air transportation, the value of the world civil aircraft engine market is projected to be 420 billion US dollars in the period between 1996 and 2015 1. Both developed and developing countries regard the civil aircraft engine industry as strategic for the development and the diffusion of technological and managerial capabilities. The thesis addresses the following two sets of questions: 1. In the aircraft engine industry, how do technology and product requirements affect the dynamics of the boundaries of engine manufacturers? 1.1. What is the scope for technology outsourcing for engine manufacturers? 2. How does the trend towards greater externalisation impact on the technology bases of the engine manufacturers? 2.1. Does modularity enable inter-firm co-ordination to be achieved via arms’ length relationships? Or is active co-ordination exerted by engine manufacturers still required? 2.2. Are product and technology modularity positively correlated? In order to address these research questions, the thesis uses a combination of two methods, namely comparative quantitative analysis and case study. The combination of these two methods enables analysis of the phenomenon using multiple lenses from different perspectives. The use of two methods also enables methodological triangulation (Yin, 1994). The study is based on two distinct types of data, qualitative and quantitative. The two types of data have been used both as supplements for mutual verification (Glaser and Strauss, 1967) and as complements to deepen and corroborate specific findings. As suggested by Eisenhardt (1989) and Yin (1994), data triangulation is made possible by multiple data collection methods, which can provide a valid substantiation of constructs and hypotheses.

1

This figure refers to the whole period and includes installed and spare engines for corporate, passenger, and cargo aircraft as well as spare parts (Rolls-Royce, 1997).

11

2. Main findings The contribution of the thesis is primarily, but not exclusively, empirical. The purpose of the study is to investigate a fundamental issue in the theory of the firm and relate the findings to current theoretical views. In doing so, the thesis relies on the theory as a ‘framework of appreciation’ (Nelson and Winter, 1982). As Nelson and Winter clearly state, in appreciative theorising a theory represents a tool of inquiry that is complemented by other tools and employed flexibly in the empirical investigation in order to fit the problem under examination. Appreciative theorising differs from formal theorising, since in the latter the empirical work is undertaken to check, extend, or corroborate the theory (Nelson and Winter, 1982). The thesis develops a theoretical framework based on the resource-based approach to the theory of the firm. Relying on this framework, the thesis aim is to explore the changing boundaries of the firms and to identify the relevant dimensions within which to understand them. Firms are characterised by a technological knowledge base and an organisational knowledge base. As Pisano (1997) argues, the former includes scientific theories, principles, algorithms, conceptual models, empirical data both embedded in equipment and embodied in people. The latter encompasses project management, resource allocation mechanisms, and incentive structures. The evolution of these two knowledge bases though distinguishable is tightly intertwined in practice. Firms remember new technological knowledge via changes in their new product development process (Nelson and Winter, 1982). This thesis focuses on the technological knowledge bases of engine manufactures. The thesis distinguishes between products and their underlying technologies (Pavitt, 1998). It understands technology as technological knowledge, and a firm’s technology base as a stock quantity (Dierickx and Cool, 1989) of technological capabilities. A firm’s technological capability is understood as its ability to undertake productive activity (Grant, 1998). In order to achieve a more detailed understanding of the boundaries of the firms, the thesis analyses firms’ technology bases in terms of two dimensions, namely the breadth and depth of technological capabilities. The breadth of technological capabilities is understood as the number of technological fields maintained in-house by engine manufacturers. The depth is understood along two further dimensions. First, the stages (e.g. concept design, detailed design, manufacturing, integration) of the development process performed by engine manufacturers. Second, the different types of knowledge related to the combination or sub-combination of engine components (architectural knowledge or sub-architectural knowledge, respectively) and knowledge concerning each component (or component knowledge). On the basis of this theoretical framework, the empirical analysis unveils the following findings. First, although engine makers are increasingly outsourcing design and manufacturing tasks to external suppliers, they maintain in-house broad and deep technological capabilities to conceive, decompose, and eventually integrate the engine system. Engine manufacturers divide up engine development tasks across a number of external suppliers, but this task-partitioning ability hinges on their own multitechnology bases. Therefore, the scope for technology outsourcing for aircraft engine manufacturers is limited due to the compelling technological requirements for the integration of the engine. Outsourcing the production of components does not necessarily entail outsourcing their underlying technologies. 12

Second, engine manufacturers enter collaborative agreements for both development and research purposes. In order to take full advantage of collaborative relations, they need to be equipped with an adequate and independent set of technological capabilities. Furthermore, collaborative agreements are decentralised modes of acquiring new knowledge. Engine manufacturers perceive collaborative agreements as learning mechanisms to maintain their systems integrator status. The more learning contexts engine manufacturers are involved in, the more likely they can maintain and nurture their systems integration capabilities over time. Third, product modularity and technology modularity follow different dynamics. The aircraft engine industry is characterised by increasing product modularization, which entails a greater division of labour across firms at the product level. However, interfirm co-ordination at the technological level cannot be achieved through arms’ length relationships. It requires active co-ordination efforts by all-round knowledgeable firms (i.e. systems integrators). Increasing product modularization, therefore, brings about two competing but related effects: (a) increasing specialisation leading to a greater division of labour across firms at the product level and (b) active co-ordination efforts by systems integrators at the technological level. These two effects are not mutually exclusive, but they underline that there is not one-to-one mapping between product and technology modularity. Furthermore, this tension between specialisation of activities on one side, and the need to actively manage technological interfaces on the other, reveals the fundamental issue addressed in this thesis. The technological boundaries of the firm differ fundamentally from the boundaries of the firm as defined by make-buy decisions. Therefore, the boundaries of the firm are not and should not be defined by (and confused with) the physical boundaries of the products a firm produces. The boundaries of the firm are complex and multifaceted, that is they are determined by the interplay of the dynamics of the products, their underlying technological knowledge, and the organisational set-up of the firm. Fourth, based on these findings, the thesis proposes an empirically grounded framework to analyse the boundaries of firms developing complex, multitechnology products. Focusing on their role of systems integrators, the thesis identifies two key dimensions of systems integration capability: synchronic and diachronic. Synchronic systems integration refers to engine manufacturers’ capabilities to set the concept design, decompose it, co-ordinate from a technological and organisational viewpoint the work of suppliers, and re-compose the engine within a given product architecture. The synchronic dimension of systems integration also relates to the capabilities required to refine and improve an existing engine architecture through the introduction of new component technologies. Diachronic systems integration refers to engine manufacturers’ capabilities to envisage different and alternative paths of product architectures to meet evolving customer and regulatory requirements in an effective and efficient way. 3. Thesis content The thesis is organised as follows. Chapter 2 presents a review of the theoretical and empirical literature the thesis addresses. Two streams of literature are discussed, namely the literature around the resource-based view of the firm and the literature on 13

the division of innovative labour. These two streams of literature are brought to bear on the issue of the boundaries of the firms. The aim of this chapter is twofold: to assess the explanatory power of the studies belonging to these two streams of literature and to provide the conceptual and analytical framework that guides the current study. Chapter 3 describes in detail the methodology employed in the thesis to address the research questions. Chapter 4 presents an overview of the aircraft engine industry and its technology. It brings to light the managerial challenges that engine manufacturers face. In doing so, the chapter provides a background for the discussion on the boundaries of engine makers’ technological capabilities carried out in the following chapters. Chapters 5, 6, 7, and 8 present a detailed analysis of the dynamics of the boundaries of engine manufactures technological capabilities. Using qualitative and quantitative data, Chapter 5 investigates the breadth of engine manufacturers’ technological capabilities in the light of the product characteristics and technological requirements and the driving forces at work in the industry impinging on their technology bases. To corroborate and strengthen the results of Chapter 5 in relation to the breadth and depth of engine makers technological capabilities, Chapter 6 presents two detailed case studies. The first case study is on the all-engine capabilities maintained in-house by engine manufacturers. This case study provides further empirical evidence in relation to the breadth of engine manufactures’ technological capabilities. The second case study focuses on the development by Rolls-Royce of the fan subsystem. This case study supplies empirical evidence on the depth of firms’ technological capabilities. Chapter 7 further explores the issue of breadth and depth of engine manufacturers’ technological capabilities by documenting a case study of the aircraft engine control system. It examines the impact of changes in the technologies underlying a component on firms’ technological boundaries. Building on the findings of the previous chapters, Chapter 8 offers comments on the concept of systems integration capability and identifies two dimensions of systems integration, namely synchronic and diachronic. Based on these dimensions, it proposes a framework within which to analyse the boundaries of firms developing complex, multitechnology products. Chapter 9 summarises the main conclusions of the thesis, underlines the implications of the findings of the thesis for theory and the strategic management of technology, highlights the limitations of the thesis, and discusses some questions for future research.

14

CHAPTER 2: THE THEORETICAL FRAMEWORK 1. Introduction The purpose of this chapter is to provide the theoretical framework for the thesis. The thesis builds upon literature on the resource-based view of the firm and on studies of the division of innovative labour. The chapter reviews these two streams of literature and highlights the theoretical and empirical concepts that the thesis aims to address. As regards the first stream of literature, the chapter presents a review of current research, building on Penrose’s (1959) study of firms’ growth processes. This research is brought to bear on the issues of the boundaries of the firms. Although the renewed interest in firms’ internal workings has produced a wealth of theoretical and empirical research to explain firm behaviour, most of the empirical studies rely on aggregate measures of firms’ capabilities that do not capture all their relevant dimensions as proposed in the theoretical literature. Furthermore, these studies have mainly focused on massmanufactured goods leaving unexplored a large number of industries that Miller et al. (1994) and Hobday (1998) have labelled complex products systems (CoPS). High cost capital goods or CoPS industries represent a group of industries that may offer invaluable insights into firms’ internal workings, given the characteristics and the nature of their products and the organisational and technological capabilities required to produce them. Chapter 2 also reviews the theoretical and empirical literature on the division of innovative labour. Recently, much debate has revolved around the concept of modularity. Industrial economists, management scholars, design and organisation theorists have discussed the far-reaching implications of such a concept at the level of product, organisation, and technology. This chapter attempts to provide an integrated account of these heterogeneous contributions. It also raises the issues left unanswered by this composite stream of literature. In particular, it highlights the problems caused by the conflation of modularity at the levels of product, organisation, and technology. The chapter is composed of six sections. Sections 2, 3, and 4 review literature on the resource-based theory of the firm. Section 2 provides an overview of the history, meanings, and relationships of such concepts as knowledge, resources, capabilities, and competencies. Section 3 compares the resource-based theory of the firm with other theories of the firm and highlights the main differences between the resource-based approach and the ‘competitive force’ approach initiated by Porter (1980, 1998). Section 4 discusses the explanatory power of the resource-based approach, transaction cost economics, and Porter’s approach in relation to the boundaries of the firm. Section 5 reviews the literature on the division of innovative labour and underlines its main deficiencies. It also presents an account of some interpretative frameworks concerned with the relationships between products, organisations, and technologies as proposed in the management literature. Section 6 draws conclusions by surveying the shortcomings of the literature reviewed that this thesis addresses. 2. Resources, competencies, and capabilities: history, meanings, and relationships In the last decade there has been a surge of interest in the internal resources of the firm. Drawing on the seminal work of Edith Penrose (1959), both economists (especially evolutionary) and management scholars have attempted to better understand the relevant 15

dimensions of firm’s internal workings. Terms such as competencies and capabilities have been coined to describe the basis of firm’s competitiveness (Dosi et al., 1992; Prahalad and Hamel, 1990; Teece et al., 1992, 1999). These concepts have become central in the field of strategic management as well as in the theoretical treatment of the business organisation. In the context of this thesis, the resource-based view of the firm represents a powerful tool to analyse firms’ boundaries. 2.1. The resource-based perspective 2: a historical background Edith Penrose and the theory of the growth of the firm The Theory of the Growth of the Firm published by Edith Penrose in 1959 is widely recognised as the pioneering work on the resource-based approach. Penrose’s book represents a path-breaking analysis of the governing principles of the growth of the firm. As discussed below, much of the concepts that were proposed later to understand firm’s behaviour and competitiveness appear in her original study. Penrose argued that the traditional (neo-classical) economic analysis was inadequate to explain the phenomenon of the growth of the firm. According to Penrose (1959: p. 1, original emphasis), such an analysis focused on “the advantages and disadvantages of being a particular size and explains movements from one size to another in terms of the net advantages of different sizes.” As a result, this analysis totally neglected the processes required to bring about such increase in size. “Growth becomes merely an adjustment to the size appropriate to given conditions; there is no notion of an internal process of development leading to cumulative movements in any other direction” (p. 1, original emphasis). Penrose also criticised the notion of ‘most profitable size’ or ‘optimum size’ of firm as espoused by the traditional economic analysis. She maintained “size is but a by-product of the process of growth” (p. 2). The emphasis of Penrose’s work is “on the internal resources of a firm – on the productive services available to a firm from its own resources, particularly the productive services available from management with experience within the firm” (1959: p. 5). Penrose understood firms as collections of productive resources3. Firm’s productive resources and knowledge represent the springboard for the firm’s growth and diversification. As later argued by Penrose (1995: xiii) in the foreword of a new edition of her book “‘history matters’: growth is essentially an evolutionary process based on the cumulative growth of collective knowledge, in the context of a purposive firm.” It is worth noting that Penrose clearly argued that the inputs for production are not the resources themselves, but the services that the resources can render 4. “The services yielded by resources are a function of the way they are used…. The important distinction between resources and services … lies in 2

The resource-based approach is also defined in the literature as competence-based approach. This change of ‘label’ should reflect the evolution of this approach and acknowledge the change of emphasis from firm resources to capabilities (see Section 2.2). However, resource- and competence-based are used in the literature interchangeably. For consistency reason, this thesis uses the term resource-based approach throughout. 3 Penrose (1995: xi) defined the firm as “a collection of resources bound together in an administrative framework, the boundaries of which are determined by the ‘area of administrative co-ordination’ and ‘authoritative communication’”. 4 Penrose (1959: 25) in a footnote clearly stated that she did not use “the term ‘factor of production’ precisely because it makes no distinction between resources and services”.

16

the fact that resources consist of a bundle of potential services and can, for the most part, be defined independently of their use, while services cannot be so defined, the very word ‘service’ implying a function, an activity… it is largely in this distinction that we find the source of uniqueness of each individual firm” (p. 25, emphasis added). The distinction that Penrose made between resources and services is fundamental. It reveals the importance of the different ways, procedures, and modes resources are used to produce different services. These ways of using resources are at the basis of a firm’s uniqueness. Furthermore, these ways of using resources hint at such firm-specific concepts as routines, competencies, and capabilities later developed by firm theorists to explain a firm’s distinctive behaviour. Business policy and firms’ resources The concepts of resources and capabilities can be also found in a leading textbook of the 1960s on business policy. Learned et al. (1969) argued that the opportunities offered by the firm’s environment have to be matched with what the firm is able to do. Their proposition was that “[T]he capability of an organisation is its demonstrated and potential ability to accomplish, against the opposition of circumstances or competition, whatever it sets out to do.” (p. 179). Learned et al. recognised that firms have potential strengths and weaknesses and acknowledged the need to determine and distinguish them to formulate a firm’s strategy. They also acknowledge the role of internal co-ordination to leverage individual experience and that current resources form the basis for a firm’s growth and diversification. “The strengths of a company which constitute a resource for growth and diversification accrue primarily through experience in making and marketing a product line. They reside in (1) the developing strengths and weaknesses of the individuals comprising the organisation, (2) the degree to which individual capability is effectively applied to the common task, and (3) the quality of co-ordination of individual and group effort.” (p. 179). Richardson: firm’s capabilities and the firm/market dichotomy Richardson (1972) can also be considered a precursor of the resource-based approach. Based on the work of Penrose (1959), Richardson (1972) maintained that existing theories of the firm do not provide any explanation of the principle of the division of the labour between firm and market. He argued that economic activity could be coordinated by direction within firms (conscious planning), co-operation across firms, and market transaction through the price mechanism. In particular, he challenged the neoclassical theory of the firm that conceived the firm as a production function, because it did not take into account the role of knowledge, skills, experience, and organisation. “The point is not that production is thus dependent on the state of the art but that it has to be undertaken (as Mrs Penrose has so very well explained) by human organisations embodying specifically appropriate experience and skills. (…) [I]t seems that we cannot hope to construct an adequate theory of industrial organisation and in particular to answer our question about the division of labour between firms and market, unless the elements of organisation, knowledge, experience and skills are brought back to the foreground of our vision” (p. 888, emphasis added). 17

Richardson (1972) argued that industries carry out a great number of activities (e.g. research, development, and design). These activities are performed by “organisations with the appropriate capabilities, or, in other words, knowledge, experience, and skills” (p. 888, emphasis added). According to Richardson, the capabilities of an organisation may be related to a technological skill (“command of some particular material technology”, p. 888), or related to particular marketing skill (“knowledge of and reputation in a particular market”, p. 888). Richardson maintained that “organisations will tend to specialise in activities for which their capabilities offer some comparative advantage” (p. 888). Richardson also introduced the concepts of similar and complementary activities. Similar activities are those that require the same capabilities for their undertaking. As a result by pursuing similar activities firms may enter “a variety of markets and a variety of product markets” (p. 888). Complementary activities are phases of a process of production (e.g. research and development). The concepts of similar and complementary activities proposed by Richardson (1972) represent an attempt to single out the different dimensions of firm capabilities. The concept of complementary activities also anticipates that of complementary assets put forward by Teece (1986). 2.2. The terms of contention: knowledge, resources, routines, and capabilities Drawing on Penrose (1959) the resource-based view conceives the firm as a bundle of organised resources. These resources are idiosyncratic and relatively immobile. They represent the source of the firm’s uniqueness and competitive advantage. The wealth of theoretical and empirical studies on the firm’s internal workings has attempted to make sense of the relevant dimensions of the firm’s resources and their relationships with the firm’s competitive advantage. These studies have put forward a number of different concepts, such as competencies and capabilities, to identify the explanatory variables of the firm’s competitive advantage. The large and eclectic collection of contributions edited by Dosi et al. (1999) is an interesting effort at understanding the microfoundations of firms’ capabilities using multiple lenses. However, this flourishing number of concepts and labels has created some terminological confusion. As Dosi et al. (1999: 5-6) argue “The term ‘capabilities’ floats in the literature like an iceberg in a foggy arctic sea, one iceberg among many, not easily recognised as different from several icebergs nearby”. In fact, in the management literature resources, competencies and capabilities are sometimes used interchangeably. Drawing on Grant (1998), this section provides a survey of the terms used in the literature, to explain their meanings, and how they relate to each other 5. The relationships between resources and competencies/capabilities Definitions and meanings of the term resource such as those proposed by Wernerfelt (1984) and Barney (1991) are relatively broad. For Wernerfelt (1984: 172) 5

There have been some attempts to provide a systematisation of the terms used in the literature. Teece et al. (1992, 1999), for instance, provide a taxonomy that distinguishes: factors of production, resources, organisational routines/competencies, core competence, dynamic capabilities, and products. Though the taxonomy proposed is very interesting and helps to “facilitate theory development and intellectual dialogue” (p. 5), it is not used in their work. In their paper the title of the section following such taxonomy is, in fact, ‘Markets and strategic capabilities’. Strategic capabilities are not actually mentioned in their taxonomy.

18

“a firm’s resources at a given time could be defined as those (tangible and intangible) assets which are tied semipermanently to the firm. (…) Examples of resources are: brand names, in-house knowledge of technology, employment of skilled personnel, trade contacts, machinery, efficient procedures, capital, etc.” Barney (1991: 101, emphasis removed) argues that “firm resources include all assets, capabilities, organisational processes, firm attributes, information, knowledge, etc. controlled by a firm (…).” He then distinguishes three categories of resources: physical (e.g. physical technology), human (e.g. training, experience, and judgement), and organisational (e.g. formal reporting structures, controlling and co-ordinating systems). These comprehensive definitions are interesting as they indicate the different elements and dimensions that contribute to firm’s resources. However, because of their sweeping character they do not identify the essential relationships among the variety of dimensions of a firm’s resources. In addition, they do not do justice to the seminal work of Penrose (1959). As discussed earlier, Penrose made a clear distinction between resources and the use of the resources (their productive services), and argued that it is in the use of these resources that the source of a firm’s uniqueness and the basis of a firm’s growth lies. This distinction can have some disputable consequences. Considering the use of resources as the source of a firm uniqueness implies that resources can be easily acquired from the market and are no longer idiosyncratic. On the other hand, by distinguishing resources from their use Penrose has laid the basis for the differentiation between resources and capabilities. Grant (1998) argues that firm’s resources can be analysed on two levels. The first relates to the individual resources of the firm such as the skills of the employees and items of capital equipment. The second relates to the way these resources work together to achieve competitive advantage. This second level is the capability level. Figure 2.1 shows the relationships among resources, capabilities, and competitive advantage. The first level of analysis encompasses three types of resources: (1) tangible resources, such as the physical assets identified in the firm’s balance sheet; (2) intangible resources, such as technology, reputation, and culture; (3) human resources, such as specialised skills and knowledge embodied in employees. As Grant argues, resources as such are not productive. It is when resources are organised, co-ordinated, and work together (i.e. their productive use in Penrosian terms) that capabilities emerge. According to Grant “capabilities are formed from teams of resources working together” (p. 122). Yet, capabilities can be organised in a hierarchical form in the sense that some capabilities relate to specific and narrowly defined tasks, whereas others involve the integration of capabilities. Therefore, as suggested by Grant, capabilities can refer to broad functional areas such as manufacturing or R&D, or to the integration of several functional areas. For instance, new product development capabilities require the integration of capabilities concerning R&D, marketing, manufacturing, finance and strategic planning. Those capabilities that require the integration of several capabilities are higher-order capabilities.

19

Figure 2.1. The relationships among resources, capabilities, strategy and competitive advantage (source: adapted from Grant, 1998)

Competitive advantage

Strategy

Industry key success factors

Organisational capabilities

Resources

Tangible • Financial • Physical

Intangible • Technology • Reputation • Culture

Human • Specialised skills and knowledge • Communication and interactive abilities • Motivation

Two points emerge from the line of reasoning discussed above. First, organisational capability, or as Grant defines it “a firm’s capacity for undertaking a particular productive activity” (p. 118), can be of different types. This view is also supported by Dosi et al. (1999) who argue that capabilities are not necessarily related only to R&D. Second, capabilities are formed from resource integration. Grant (1998) maintains that the integration of the knowledge and skills of firms’ employees with physical equipment and other firms’ resources takes place through particular organisational mechanisms that Nelson and Winter (1982) have defined as organisational routines. According to Nelson and Winter, routines are “regular and predictable behavioural patterns of firms” (p. 14). Routines act as a co-ordinating mechanism for individual knowledge, skills, and activities. Nelson and Winter consider routines as reflecting the skills of the organisation as they embed organisational knowledge and, as a consequence, are at the basis of organisational capabilities.

20

Distinctive, core, strategic and dynamic capabilities Since not all capabilities are potential sources of competitive advantage, several adjectives have been used to define those capabilities that provide a competitive advantage 6. Prahalad and Hamel (1990) coined the term core capabilities 7 to identify those capabilities key to a firm’s performance. Accordingly, core capabilities should provide potential access to a wide variety of markets, make significant contributions to the perceived customer benefits of the end product, and be difficult to imitate. Drawing on Prahalad and Hamel’s concept, Leonard-Barton (1992: 111, original emphasis) argues that “capabilities are considered core if they differentiate a company strategically”. Capabilities that provide a competitive advantage have been defined also as distinctive (Barney, 1991; Patel and Pavitt, 1997) or strategic (Teece et al., 1992, 1999). Drawing on the resource-based approach, Teece et al. (1992, 1999) and Teece and Pisano (1994) have put forward the dynamic capability approach. They argue that the dynamic capability approach extends the resource-based approach since it adds the dynamic perspective that was partly neglected. Teece et al. (1992, 1999) contend that firms not only accumulate capabilities but also need to continuously reconfigure and integrate internal and external resources to compete effectively in an ever-changing environment. Thus, a dynamic capability is the firm’s ability to reconfigure, redirect, transform, and integrate internal capabilities with external factors of production and resources to meet environmental challenges (Teece et al.). Defined in this way, a dynamic capability represents a super-order capability that pertains to strategic top management. Capabilities, endowments, and asymmetric information One of the main tenets of the resource-based approach is that capabilities are ‘sticky’, that is they are relatively immobile, firm-specific, and show some degree of inertia. As a consequence, firms are ‘stuck’ with their capabilities, i.e.: with what they are able to do. These features of organisational capabilities may render them similar to endowments (Dosi and Marengo, 1993). However, Dosi and Marengo (1993: 160) argue that capabilities differ fundamentally from endowments as they “are subject to learning and change through their very application to actual problem solving”. Dosi and Marengo use as a metaphor the good football player to explain the difference between endowments and capabilities. Though there is a strong endowment component, not all talented football players succeed. Their success depends on “training, the sequence of teams the individual has played for, his or her coaches, and, last but not least, on chance” (p. 160). Dosi and Marengo also argue that organisational learning processes cannot be reduced to mere information gathering and processing. Unlike the Bayesan learning processes, where new information is employed to update the probability distribution within a fixed and unchanging frame of reference, in the learning processes Dosi and Marengo refer to, the frame of reference is continuously updated, constructed, evaluated, and 6

In relation to a firm’s resources, Barney (1991) argues that in order to have potential sustained competitive advantage, resources must have four attributes. They must be valuable, rare, imperfectly imitable, and non-substitutable. 7 Prahalad and Hamel (1990) talk about core competencies not capabilities. However, there is no clear semantic distinction between the two terms inasmuch as they are used interchangeably in the management literature. The thesis uses the term capabilities, throughout.

21

eventually modified. As a consequence, problem solving activity other than being at the basis of the organisational learning process is also problem framing. As Dosi and Marengo (1993: 160, original emphasis) conclude “There are fundamental elements of learning and innovation that concern much more the representation of the environment in which individuals operate and problem solving rather than simple information gathering and processing”. Framed in this way, asymmetric access to information does not and cannot explain the persistent differences in firm performance. Process as the mainstay of the resource-based approach From the line of reasoning discussed above, it emerges that learning processes are another fundamental tenet of the resource-based approach. It is also worth underlining that according to this approach, firms’ historical processes play a fundamental role in explaining firms’ persistent differences and eventually their different performance. Williamson (1998: 25, emphasis added) argues “The common theme that runs through these [different contributions within the resource-based approach] is the importance of process”. Dierickx and Cool (1989) in their critique of Barney’s concept of ‘strategic factor market’ put emphasis on the building process that affects the accumulation of strategic assets, that is those assets that are non-tradable, non-imitable, and non-substitutable 8. They argue that the common feature of strategic assets, such as dealer loyalty or R&D capability, is “the cumulative result of adhering to a set of consistent policies over a period of time. Put differently, strategic asset stocks are accumulated by choosing appropriate time paths of flows over a period of time” (p. 1506, original emphasis). They then go on to say “while flows can be adjusted instantly, stocks cannot. It takes a consistent pattern of resource flows to accumulate a desired change in strategic asset stocks” (p. 1506, original emphasis). This distinction between stocks and flows proposed by Dierickx and Cool is fundamental as it underlines that what it is strategic for a firm must be painstakingly accumulated over time. Dierickx and Cool argue that the process of accumulation of asset stocks is characterised by the interplay of the following properties: (a) time compression diseconomies, (‘crash’ R&D programmes are often characterised by low effectiveness); (b) asset mass efficiencies (‘success breeds success’); (c) interconnectedness of assets stocks (assets stock accumulation is influenced by the stock of other assets); (d) asset erosion (all asset stocks decay and need to be maintained); and (e) causal ambiguity (the process is not deterministic but it is characterised by stochastic elements). The work of Dierickx and Cool draws on the evolutionary theory proposed by Nelson and Winter (1982) and contains concepts that Teece et al. (1992, 1999) and Teece and Pisano (1994) have further elaborated in their dynamic capability approach. Teece et al. (1992) have enriched the concepts of stocks and flows of Dierickx and Cool (1989). They contend that a firm can be understood in terms of its managerial and 8

It is worth noting that strategic assets á la Dierickx and Cool (1989) are very similar to distinctive or strategic capabilities discussed above.

22

organisational processes, its position, and the paths available to it. A firm’s competitive advantage stems from its processes, positions, and paths. In their words (1999: p. 8, emphasis added), “By managerial and organisational processes we refer to the way things are done the firm, or what it may be referred to as routines, or patterns of current practice and learning. By position we refer to its current endowment of technology and intellectual property, its complementary assets, its customer base and its external relations with suppliers and complementors. By paths we refer to the strategic alternatives available to the firm, and the presence or absence of increasing returns and attendant path dependencies”. 3. The resource-based view: between theoretical interpretation and normative implications As mentioned in the foregoing section, the resource-based approach goes back to the seminal work of Penrose (1959). However, although some scattered works refer to the concepts that Penrose first developed, such as Learned et al. (1969) and Richardson (1972) discussed above, the resource-based perspective has developed little since Penrose’s work. This holds true for both theoretical interpretations of firms and for prescriptive tools developed by management scholars. Section 3.1 compares the resource-based approach tradition with the industry attractiveness approach as put forward by Porter (1980, 1998). Following Fransman (1994), Section 3.2 briefly compares the resource-based approach with some more traditional theoretical accounts of the firm, such as those proposed by Jensen and Meckling (1976) and Williamson (1975). 3.1. The resource-based view within the tradition of strategic management In the last decade the strategic management literature has been characterised by a surge of studies on firm-specific factors (Cool and Schendel, 1988; Dosi et al., 1992; Prahalad and Hamel, 1990; Rumelt, 1991; Teece et al., 1992, 1999). Management researchers shifted back their microscope to the ‘internal workings’ of firms, after the influential work of Porter (1980, 1998) had focused it on industry structure as the primary determinant of competitive advantage. It is worth noting, however, that as Wernerfelt (1984) argued, the traditional concept of strategy is based on the firm’s resources in terms of its strength and weaknesses (Andrews, 1971), but the economic tools that have been developed revolve around firm’s products. According to Wernerfelt (1984: 171), the idea of looking at firms as an organised collection of resources received little formal attention, because of “the unpleasant problems (for modelling purposes) of some key examples of resources, such as technological skills”. Porter and the industry attractiveness models The work of Porter (1980, 1998) puts emphasis on the firm’s external environment and in particular on the industry competitive forces. According to Porter, five basic competitive forces drive the industry competition and determine the profit potential in an industry or the industry attractiveness 9. Industries are, therefore, characterised by different degrees of attractiveness according to the intensity of the five basic 9

The five competitive forces identified by Porter (1980) are: rivalry among existing firms, bargaining power of suppliers, bargaining power of customers, threat of substitute products or services, and threat of new entrants.

23

competitive forces. Porter (1980: 4) argues that “The goal of competitive strategy for a business unit in an industry is to find a position in the industry where the company can best defend itself against these competitive forces or can influence them in its favour”. According to this approach, an entry decision would more or less follow certain steps: (1) choice of an industry characterised by high attractiveness; (2) choice of a strategy based on competitors’ strategies; (3) acquisition of the resources required to compete in that industry. Thus, the resource acquisition process seems easy and feasible 10. The resource-based approach is at variance with such an approach. Following Penrose (1959), in the resource-based view firms are heterogeneous because of their organised bundle of resources. Capabilities that emerge from the co-ordination of resources are relatively immobile (sticky) and are the source of the firm’s uniqueness (Grant, 1998). The resource acquisition process is long and painstaking and firms are not able to develop resources quickly. Therefore, as Teece et al. (1999: 3) put it, the entry decision process from a resource-based perspective is roughly as follows “(1) identify your firm’s unique resources; (2) decide in which markets those resources can earn the highest rents; (3) decide whether the rents from those assets are most effectively utilised by (a) integrating into related markets, (b) selling the relevant intermediate output to related firms, or (c) selling the assets themselves to a firm in related business.” Though finding it useful and promising, Porter (1991, 1998) criticised more than once the resource-based approach for being inward-looking. In particular, Porter (1991: 108) argues that resource and capabilities are “meaningful in the context of performing certain activities to achieve certain competitive position” but not for themselves. He also argues that resources can be the source of competitive advantage only if they are optimally integrated within the firm. Furthermore, Porter contends that the presence of resources reflects prior managerial choices. A closer look at the main arguments of the resource-based approach reveals that Porter’s criticisms are superficial. As regards the relationship between resources and managerial choice, the resource-based approach recognises that resources and capabilities are cumulated over time. Of course, this cumulative process involves choices by top managers and it is not therefore incompatible with them. What is important to note is that the resource-based view maintains that history matters (Dierickx and Cool, 1989; Nelson and Winter, 1982; Pavitt, 1990; Penrose, 1959) and therefore, that this cumulative process is path-dependent. Furthermore, it is recognised that resources alone cannot be a source of competitive advantage. They become a source of competitive when they are integrated and co-ordinated and when they work together, in other words, when capabilities emerge (Grant, 1998). Along these lines, scholars belonging to the resource-based tradition acknowledge the fact that resources

10

As Barney (1991: 100) has underlined “Implicitly, this work has adopted two simplifying assumptions. First, these environmental models of competitive advantage have assumed that firms within an industry (or firms within a strategic group) are identical in terms of the strategically relevant resources they control and the strategies they pursue.… Second, these models assume that should resource heterogeneity develop in an industry or group (perhaps through new entry), that this heterogeneity will be very short lived because the resources that firms use to implement their strategies are highly mobile (i.e.: they can be bought and sold in factor markets)”.

24

represent an intermediate stage for achieving competitive position and not a competitive advantage per se. 3.2. The resource-based theory and other theories of the firm: knowledge vs. information The aim of this section is to present a brief comparison of the current theories of the firm in relation to the questions posed in the thesis. However, a comprehensive review of these major approaches is beyond the scope of the section. Fransman (1994) argues that traditional approaches to the firm in economics, such as those proposed by Jensen and Meckling (1976) and Coase (1937), and the resourcebased theory of the firm as put forward by Penrose (1959) and Nelson and Winter (1982) belong to two different paradigms. The differentiating element lies in fact that the former fails to distinguish information from knowledge. Failing to recognise the distinction between knowledge and information has strong implications for the reasoning behind the two paradigms and their understanding of firms. In fact, traditional approaches see firms as responses to information-related problems. The corollary of this is that these approaches understand firms to be information-processing systems. According to Fransman, the approaches proposed by Alchian and Demsetz (1972), Coase (1937), Jensen and Meckling (1976), and Williamson (1975, 1987) regard firms as devices contrived to respond to the problem of ‘asymmetric information’. Fransman (1994) argues that those approaches that understand firms as mere information-processing systems have limited explanatory power of real phenomena 11. On the other hand, the resource-based approach conceives firms as knowledge-based systems. Firms do not merely process information, but elaborate and interpret it according to their beliefs developed and accumulated over time. Firms are information processing systems and repositories of knowledge and manipulators and creators of knowledge. New knowledge is acquired via firm-specific processes, which filter and constrain a firm’s behaviour. Fransman (1994) argues that information and knowledge are two different concepts. He proposes that information may be defined as “...data relating to states of the world and the state-contingent consequences that follow from events in the world...” (p. 755). Knowledge is instead defined as a variety of beliefs, and while it “...is influenced by information processed by the believer, belief is not necessarily wholly determined by processed information. In the formation of the belief, accordingly, there is room for insight, creativity and misconception…” (p.755). 11

Fransman puts forward the story of the US-based firm, IBM, to give enlightenment to this point. At the beginning of the 1990s IBM had gathered and processed information concerning the enlarging market of ever-increasing performance of personal computers at the expense of mainframe computers. Nevertheless, IBM believed that mainframe computers could sustain their profitability and suffered heavy consequence as a result. Fransman argues that if we read this phenomenon through the lenses of the traditional approaches that see firms as information-processing systems then we are confronted with what he calls the ‘IBM paradox’. In fact, although the information possessed by IBM indicated that smaller computers and mini-computers were eroding the market of mainframe computers, IBM clung to the belief that mainframes were able to sustain their profitability. This was due to a disjunction between the information processed and possessed by IBM and its beliefs and its knowledge. Such a disjunction could not be explained if we understand firms only as information-processing systems. If we make a distinction between information and knowledge and we understand knowledge to be a set of beliefs through which information can be interpreted and misinterpreted, firms, as knowledge-based systems, are likely to make mistakes even with the same information, as the IBM case shows.

25

Along the same lines, Dosi et al. (1996) consider information as “well stated and codified propositions about (i) states-of-the-world (e.g. ‘it is raining’); (ii) properties of nature (e.g. ‘A causes B’); (iii) identities of other agents (e.g. ‘I know Mr X and he is a crook’) and (iv) explicit algorithms on how to do things” (p. 22). Knowledge is instead understood as including “(i) cognitive categories; (ii) codes of interpretations of the information itself; (iii) tacit skills and (iv) search and problem-solving heuristics irreducible to well defined algorithms” (p.22). Resource-based view vs. contractual view and transaction cost economics It is worth noting two other points in relation to the resource-based approach. First, the resource-based view of the firm sets itself apart from the contractual view of the firm as proposed by Jensen and Meckling (1976). In fact, according to the scholars from the resource-based tradition, considering firms as ‘nexus of contracts’ or ‘quasi markets’ (Arrow, 1974) fails to recognise the co-operative activity and learning that takes place within the firm’s internal organisation and whose effectiveness cannot be replicated by the market. Second, whether the resource-based approach and the transaction cost economics approach pioneered by Coase (1937) and developed by Williamson (1975) are contrasting or complementary views of the firm is a hotly debated topic. The transaction costs approach begins with Coase (1937) who argued that firms are devices to economise on market transactions. In his words, “…the operation of a market costs something and by forming an organisation and allowing some authority (an ‘entrepreneur’) to direct resources, certain marketing costs are saved” (p. 392). “A firm, therefore, consists of the system of relationships which comes into existence when the direction of resources is dependent on an entrepreneur” (p. 393). Coase’s argument implies that co-ordination can be achieved either via the market through the price mechanism or by direction within firms, the only discriminating criterion being the cost. Coase refers to two types of costs, namely the cost of organising production and the cost of negotiating and contracting. Williamson (1975, 1987) extended Coase’s argument. For Williamson the basic unit of analysis is the market transaction. Williamson distinguishes between transaction costs and production cost. “The modern corporation is mainly to be understood as a series of organisational innovations that have had the purpose and effect of economising transaction costs” (1987: p. 273). He bases his argument on two behavioural assumptions that characterise human nature, namely opportunism and bounded rationality, and two environmental factors, that is to say uncertainty and small numbers. These factors explain the circumstance under which firms or markets emerge as the governance mode of economic activity. Williamson contends that firms emerge “to organise transactions so as to economise on bounded rationality while simultaneously safeguarding them against the hazards of opportunism” (p. 32). According to this brief review of Williamson’s argument, contractual hazards play a prominent role in defining the boundaries between firms and markets. Williamson (1998) also clearly states that transaction cost economics has a limited bearing on the issues of a firm’s growth and learning. Therefore, transaction cost economics and the resource-based approach should be considered as two contrasting paradigms. As discussed earlier, learning and growth play dominant roles in the resource-based 26

approach. Furthermore, one of the mainstays of the resource-based view is that capabilities are non-tradable. In other words, there is no factor market for them, since they need to be developed over time. Nonetheless, scholars within the resource-based tradition do not seem to disagree with transaction cost economics. Dosi et al. (1992) argue that transaction costs are an integral part of their theory on corporate coherence that, in fact, belongs to the resource-based tradition. Similarly, Teece et al. (1999) acknowledge the problem of market contracting and the ensuing contractual hazards, but they contend that transaction costs are not the only reason why firms exist. This interesting debate is still open. A comprehensive discussion of the terms of contention of these two paradigms goes well beyond the scope of this study, however 12. Because the aim of the thesis is to examine the boundaries of the firm, the next section focuses the discussion on a comparison of the explanatory power of the resource-based perspective, transaction cost economics, and Porter’s approach in this connection. 4. The analysis of the boundaries of the firm: transaction cost economics, resource-based approach, and Porter’s approach Drawing on Williamson (1975, 1987), Stuckey and White (1993) hold that the most important reason for vertically integration is vertical market failure. According to them, a vertical market ‘fails’ when there are a small number of buyers and sellers; high asset specificity, durability and intensity; and frequent transactions. With regards to the second aspect, they put forward the case of automobile assemblers and component suppliers who can become locked together when a component is specific to a particular car model. This often results in high switching costs and, therefore, leads to decisions to vertically integrate. In addition, they suggest that vertical integration strategies should be changed as market structures change, that is to say when the number of buyers and sellers and the importance of specialised assets change. According to this approach, therefore, firms can change their vertical integration strategy in an almost instant response to market signals. Similarly, Monteverde and Teece (1982) in their study on the US automotive sector, conclude that the vertical integration structure of GM and Ford is based on efficiency considerations. First, they argue, it reduces the exposure of the automotive manufacturers to the opportunism of suppliers, since the production process generates specialised know-how leading to high switching costs. Second, the vertical integration structure allows for better co-ordination of the production process. The resource-based approach introduces a different perspective on vertical integration. Firms are conceived as repositories of knowledge, resources, and capabilities. Their boundaries are defined by the knowledge bases of the firms. As Kogut and Zander (1992: 396) argue “In contrast to a perspective based on the failure to align incentives in a market as an explanation for the firm, … firms are a repository of capabilities, as determined by the social knowledge embedded in enduring individual relationships structured by organising principles.” They heavily criticise some empirical studies within the transaction cost tradition, namely the work by Walker and Weber (1984) that 12

See, for instance, a set of contributions published in Organization Science (1996) containing articles by Barney, Foss, Kogut and Zander, and Conner and Prahalad and the work of Foss (1993). For an empirically based attempt to integrate transaction cost economics and the resource-based approaches see Argyres (1996).

27

stresses the role of contractual hazards in informing decisions to source components internally, and that by Monteverde and Teece (1982) discussed above. As regards the work of Walker and Weber (1984), Kogut and Zander (1992: 394) argue “… the most important variable is the indicator of differential firm capabilities, that is whether the firm or the supplier has the lower production costs. Transaction costs considerations matter but are subsidiary to whether a firm of other suppliers are more efficient in the production of the component”. Similarly, for the analysis of Monteverde and Teece (1982), Kogut and Zander (1992: 394) contend “…the most significant variable is the dummy for the firm. In other words, despite controls, the heterogeneous and unobserved firm effects were the dominant influence on the make-buy decision.” Kogut and Zander propose that the make-buy decision is influenced not only by transaction costs but also by the set of capabilities that currently characterise the firm. They argue “Many investment decisions inside a firm do not include a make-buy calculation, for the presumption is that the new assets are the extensions, or combinations, of the existing knowledge base” (p. 395). In fact, Kogut and Zander argue that firms have combinative capabilities that is the ability to generate new knowledge from internal existing knowledge and outside knowledge 13. Along the same lines, Winter (1988: 177) argues that the transaction as unit of analysis is troublesome since “Firms perform their function as repositories of knowledge largely by virtue of the extensions in time of the association of inputs, especially human service inputs, with the firm. At any particular time, the costs and benefits are substantially influenced by the network of transacting patterns already in place”. Accordingly, the ability of transacting is a capability per se. The analysis of vertical integration within the resource-based approach departs also from traditional strategic analysis of vertical integration that focuses mainly on the economic issues that the ‘make or buy’ decision entails. Porter (1980) states “[T]he decision must go beyond an analysis of costs and investments requirements to consider the broader strategic issues of integration versus use of market transaction, as well as some perplexing administrative problems in managing a vertically integrated entity that can affect the success of the integrated firm. ...To find the strategically appropriate extent of vertical integration for the firms requires balancing the economic and administrative benefits of vertical integration with the economic and administrative costs”. 13

Lazonick and O’Sullivan (2000) contend that asset specificity, one of the givens of the transaction cost approach, is not a given but it is developed within the firm. “An innovative enterprise … relies on asset specificity as a developmental source of competitive advantage” (p. 42).

28

According to Porter, benefits are to be found in economies of integration (combined operations, internal control and co-ordination, information, avoiding the market, stable relationships), offsetting bargaining power, enhancing the ability to differentiate and raise barriers to entry, entering a higher return business and defending against foreclosure. To tap into technology is considered a benefit as well, since firms that vertically integrate can gain a better understanding of the upstream technology or of its use. On the other hand, Porter argues that integration in order to tap into technology is tapered or partial because full integration implies some technological risks. He argues that firms that integrate may cut themselves off from the flow of technology from suppliers or customers to the extent that “integration usually means that a company must accept responsibility for developing its own technological capability rather than piggybacking on others” (p. 312). According to Porter’s approach, therefore, the development of technological capabilities is viewed as a responsibility for the firms that decide to vertically integrate. To sum up, by emphasising the role of contractual hazards entailed by market transactions, the transaction cost approach argues that the boundaries between firm and market activities stem from firms’ strategies of economising on modes of governance. In this way, it suggests nearly ‘instantaneous’ vertical integration strategies “firms should change their vertical integration strategy when market structure and asset specificity change” (Stuckey and White, 1993: 79) 14. From Porter’s perspective the development of technological capabilities is viewed as a burden that firms should avoid otherwise they forgo the opportunity to tap into the distinctive competencies of suppliers. By contrast, the resource-based approach suggests exactly the opposite, that is to say that capabilities are the sine qua non to attaining and sustaining competitive advantage. The resource-based approach reassesses critically the role of strategic outsourcing for firms’ competitiveness and some metaphors employed to depict responsive organisational structures such as ‘virtual corporations’ (Dosi et al., 1992; Teece, 1988). Accordingly, managerial decisions to ‘hollow out’ the corporation by hiving off key component production, relying only on economic factors and without taking into account economies of learning, may entail a plunge into dependence, loss of important skills and experience, and a source of change generating capacity. In particular, because of the ‘context-dependency’ of knowledge, contracting out means losing ‘contexts of learning’. There is no learning without contexts, or in other words, contexts act as causa prima for learning. To put it in another way, spinning off production means spinning off cognitive activities. 4.1. Some concluding remarks on the resource-based approach The previous sections have highlighted that renewed interest in the internal workings of the firm has inspired both theoretical and empirical works. Theoretical studies have attempted to explain the nature of the source of firms’ competitive advantage. The 14

Arora and Gambardella (1994) have made an insightful point in relation to the merit of the transaction cost approach in explaining the factors governing the division of innovative labour (an issue discussed in Section 5). They argue that “there is a ‘technical’ constraint upon the division of innovative labour which is logically distinct from the constraint posed by opportunism” (p. 528). This constraint derives from two characteristics of knowledge relevant for innovation: context-dependency and firm-specificity. These characteristics render the costs of transferring knowledge between firms much higher than those for intrafirm transfer.

29

empirical exercises have attempted to measure and ground theoretically rich constructs, such as resources and capabilities. Though research on firms’ resources and capabilities is still at a preliminary stage, some empirical studies have provided invaluable insights on firms’ capabilities (Dosi et al., 1999). However, the resource-based approach also has some limitations. First, identifying capabilities is a difficult task. The flourishing of terms and labels and the different meanings attributed to them can exacerbate this difficulty. As Williamson (1998: 26) puts it “Identifying the unit of analysis out of which the competence perspective works will be facilitated by defining what a competence is.… Frankly, however, I have no idea of what observations I should ask my research assistant to take.… Awaiting efforts to operationalise this concept [of capability], the vagueness to which Richardson referred in 1972 continues to day.” Second, related to the issue of identification is that of the degree of aggregation that researchers or practitioners use to operationalise a firm’s capabilities. Most of the empirical studies within the resource-based tradition rely on aggregate measures of firms’ capabilities that do not capture their relevant dimensions as proposed in the theoretical literature. For instance, based on the success of such firms as Honda and NEC, Prahalad and Hamel (1990) suggest that firms should focus on a few core capabilities and outsource peripheral ones. However, identifying what are core capabilities depends very much on the degree of aggregation chosen by the practitioners or researchers. The identification of core capabilities is even more troublesome in the case of firms involved in the production of multitechnology, multicomponent products such as aircraft engines studied in this thesis. In addition, some studies have underlined that a focus on a few core capabilities may have detrimental effects on the firm’s long term survival (Bruck, 1995; Quinn and Hilmer, 1995). Third, studies on firms’ capabilities have mainly focused on mass-manufactured goods, such as automobiles and computers, leaving unexplored a large number of industries that Miller et al. (1994) and Hobday (1998) have labelled complex products systems (CoPS). High cost capital goods or CoPS industries represent a group of industries that may offer insights into the firms’ internal workings of other types of firms. The multicomponent, multitechnology nature of the products that characterise this group of industries poses extreme managerial challenges for firms in terms of the organisational and technological capabilities to be developed and the ensuing make-buy decision process. The thesis builds upon the resource-based approach and represents an attempt to fill some of these gaps. The focus of the thesis is on the technological capabilities of the world’s largest aircraft engine makers. As will be detailed in Chapter 4, the aircraft engine industry shares some of the characteristics of CoPS industries as identified by Miller et al. (1993) and Hobday (1998). The technological fields of US patenting are used as a proxy measure of technological capabilities. In order to provide a detailed account of engine makers’ technological capabilities, the study distinguishes two dimensions of them, namely breadth and depth. The breadth of technological capabilities is understood as the number of technological fields maintained in-house by 30

engine manufacturers. The depth is understood along two further dimensions. First, the stages (e.g. concept or detailed design) of the development process performed by engine makers. Second, the different types of knowledge related to the combination or subcombination of components (architectural knowledge or sub-architectural knowledge, respectively) and knowledge concerning each component (or component knowledge). 5. The division of innovative of labour: relationships among product, organisation, and technology Theoretical and empirical works on the division of innovative labour form the second stream of literature addressed in this thesis. Recently, much of the debate on the appropriate forms of organisational and technological co-ordination for innovation has revolved around the concept of modularity. Drawing on (but not always giving credit to) the concept of decomposability put forward by Herbert Simon (1962, 1976), industrial economists, management scholars, and organisational theorists have elected modularity as the principle that should inform product and organisation design and firms’ cognitive activity. Modularity was first proposed as a product design strategy aimed at decomposing products through the definition of stable interfaces among its components (modules). Accordingly, the design of each module can be improved within a predefined range of variation and a predefined period without impacting on the design of other modules (Ulrich, 1995). Recently, management scholars and industrial economists have used the concept of modularity to analyse the dynamics of products’ underlying technological knowledge, organisational design, and inter-firm relationships (Arora et al., 1998; Sanchez and Mahoney, 1996). The position that emerges from research on modularity is as follows. By adopting a modular design strategy firms fully specify and de-couple both components and the different stages involved in the product development process. They focus either on the development of the product architecture or of the different modules. In this way, firms reflect the modular development process and embed the modular knowledge underlying each component. A corollary of this line of reasoning is that modularity in products, organisations, and technologies is positively correlated. As a consequence, a greater division of labour at each level (product, organisation, and technology) characterises the different industrial settings. The aim of this section is to provide an integrated account of these contributions to the modularity debate coming from industrial economics, management, organisation, and design theory. In doing so, the section highlights the issues that current research has left unanswered and shows how the thesis will address them. 5.1. Modularity in products Some criteria for product decomposition Christopher Alexander (1970) proposed that one of the criteria for decomposing a set of design variables should be to try to maintain as few interactions as possible across the main subsets. In this way, the variables in different subsets of the partition impose as little conflict as possible on one another. Building on this decomposition criterion, software development scholars have put forward various design techniques, such as structured design (Stevens et al., 1974) and modular design (Parnas, 1972). They provide guidelines on how to decompose the software product and its underlying development process into independent ‘pieces’ that can be considered and executed separately. They emphasise the importance of both reducing the coupling of 31

components by minimising relationships among them and maximising relationships within the same component 15. It is worth noting that Parnas et al. (1985) underline that the application of modular design is not easy. Modular design requires the designer’s estimate of the likelihood of change, knowledge about the application area as well as knowledge about hardware and software technology. Therefore, modular design, by definition, calls for more design knowledge. The simplistic conclusion that modular design would call for ‘less design knowledge’ is therefore incorrect. It was for this reason that Parnas et al. developed a hierarchically structured document (the module guide) to allow both designers and maintainers to identify and understand the software product designed according to modular principles, without going into the details of each module. Such a guide acts as an aide-memoir for addition of further modules and for maintenance. Product architecture and the designer role Building on the notion of modular design put forward by Parnas (1972), Ulrich (1995) defines the architecture of a product as “the scheme by which the functions of a product are allocated to physical components”. The concept of product architecture encompasses: (a) the arrangement of functional elements, that is to say the structure of the individual operations that contribute to the overall performance of the product; (b) the mapping from the functional elements to physical parts, that is identifying which component implements which function; and (c) the specification of the interfaces amongst the different components, in terms of contact (e.g. geometric) or non-contact (e.g. infrared) connections and interactions. Ulrich (1995) argues that a product architecture can be modular or integral according to two properties: (a) the mapping from functional requirements to physical objects composing the product and (b) the degree of ‘decoupling’ of the interactions amongst components. In a modular architecture, components implement one or a few functional elements (one-to-one mapping) and component interactions are well specified. In an integral architecture there is a complex mapping between functional elements and components (i.e. components implement many functional elements), and component interactions are ill-defined and coupled. The characteristics of product architectures can have different implications for the firm’s product strategy. In fact, a modular architecture seems more appropriate when firms want to emphasise product variety, change, and standardisation. Indeed, a product with a modular architecture allows firms to change the product by upgrading or adding modules in isolation. An integral architecture, however, may be more appropriate when product performance is the main concern in the firm’s product strategy. Therefore, according to Ulrich, firms have a degree of latitude in their choice of product architecture. These decisions are subject to and linked to the firm’s strategy and, in particular, to product performance, product change, product variety, component standardisation, and manufacturability. In other words, products may lend themselves 15

Modular design techniques require the use of the criterion of ‘information hiding’. According to this criterion, “system details that are likely to change independently should be the secrets of separate modules; the only assumptions that should appear in the interfaces are those that are considered unlikely to change” (Parnas et al., 1985).

32

to either modular or integral architecture according to the firms’ product specific strategy. Designers therefore have some degree of freedom in choosing the appropriate architecture to meet firms’ goals. Concluding remarks on product modularity The preceding sections have highlighted some issues in relation to product modularity. Modularity is a design strategy that if adopted can have profound implications for the firm’s product strategy. A few points are worth stressing here in relation to product modularity. First, the argument that modular product architecture demands less design knowledge is incorrect. As Baldwin and Clark (1997) argue, modular systems are more difficult to design than integral ones as designers need a deep understanding of the ‘inner workings’ of the product, in order to partition and decouple design tasks. As discussed earlier in the case of software development, a modular design calls for more design knowledge (Parnas, 1972). Chapter 5 analyses the adoption impact of modular design strategies on engine manufacturers’ technology bases. Second, the definitions of modular and integral architectures proposed by Ulrich (1995) are to be understood as ideal types. Most products do not fully satisfy the requirements of either and, therefore, lie somewhere along a continuum that goes from fully modular to fully integral architectures. Modularity is a matter of degree. Correspondingly, the degree of modularity depends on the level of analysis. Products can be decomposed at different levels: sub-systems (e.g. control system), sub-sub-systems (e.g. fuel metering unit), components (e.g. valve), and sub-components (e.g. spring). Therefore, modularity can be a characteristic of all or only some of these levels. Third, the product’s technological trajectory may constrain the adoption of a modular design strategy. That is, the interplay between technological and economic/social forces that shapes the technological path may limit the degree of freedom for designers in the choice of the appropriate design strategy. In particular, during the early stage of the development of a technology (i.e. when the degree of novelty is high) the degree of latitude may well be limited (Von Hippel, 1994, 1998). On the other hand, since the economic advantages of modular design are substantial, firms are increasingly designing their products in a modular fashion. In other words, there is an increasing trend towards modular products in different industrial sectors. Chapter 5 shows that a modular design strategy has been also adopted in the aircraft engine industry. Fourth, modular design allows for the ‘de-coupling’ of component interactions at the design level which is different from ‘tight or de-couple’ in an actual product. Take, for example, the personal computer. Its design may have loosely coupled components interactions in that different components (e.g. microprocessor) may be substituted within a certain range dictated by the architecture into the computer design without requiring a redesign of the other components. However, the components in the physical computer must be tightly coupled in the sense that all components must function in unison in order for the computer to function as a system 16 (Sanchez and Mahoney, 16

Suh (1990: 50, 52) made a similar point “Functional coupling should not be confused with physical coupling…[functional] de-coupling does not necessarily imply that a part has to be broken into two or more separate physical parts, or that a new element has to be added to the existing design. Functional

33

1996). Therefore, it is possible to distinguish two different but related sets of interactions: (a) at the level of the design and (b) at the level of the parts (physical structure). Tong and Sriram (1992) argue that the first level informs the second, since design interactions can make non-neighbouring parts interact. Design interactions also inform another sets of interactions, i.e.: those related to the design of interacting parts (Tong and Sriram, 1992). This further set of interactions concerns the organisation of the design activities and leads to the discussion in the following section. 5.2. The relationships between product, organisational, and technological modularity Henderson and Clark on the distinction between architectural and component innovation Building on previous studies on innovation dynamics and product architecture (Abernathy and Utterback, 1975; Clark, 1985), Henderson and Clark (1990) define an architectural innovation as a change in the relationships between a product’s components that leaves untouched the core design concepts it embodies. According to Henderson and Clark, organisations are built around stable product architectures. The product architecture defines key functional relationships, information processing capabilities, communication channels, and information filters. Once a dominant design has emerged, it is encoded and thus becomes embedded in the organisational set-up. This is why architectural innovation represents such a subtle and dangerous challenge for incumbents. While leaving unchanged the relevant technological capabilities, an architectural innovation alters key, yet tacit, interfaces. Sanchez and Mahoney on organisational modularity Building on Henderson and Clark (1990) and Ulrich (1995), Sanchez and Mahoney (1996) argue that the concept of modularity can also be applied to the firm’s organisational processes and, in particular, to its product development processes. Traditional design approaches call for intensive managerial co-ordination throughout the development process, since product designs are composed of tightly coupled components. By contrast, modular product architectures by fully specifying component interfaces which are not permitted to change during a certain period of time, allow the different stages of the development process to become independent, since all the developing components conform to standardised interface specifications. The coordination of the development process, therefore, is ameliorated and simplified. The continual exercise of managerial authority is reduced to a minimum due to an ex ante precise and specified partitioning of the components’ interfaces. As a consequence, the related design tasks are in turn carried out by concurrent and autonomous teams. In the words of Sanchez and Mahoney (1996: 66), “The specifications for standardised component interfaces provide, in effect, an information structure (Radner, 1992) that co-ordinates the loosely coupled activities of component developers”. Knowledge and organisational modularity Arora et al. (1998) contend that not only can the production of new products be conceived in terms of production and combination of modules, but also the knowledge underlying such artefacts is a matter of ‘mixing and matching’ of modules. In considering knowledge as a public good, these modules are readily available to firms, decoupling may be achieved without physical separation, although in some cases such physical decomposition may be the best way of solving the problem”.

34

which can concentrate either on the production of new modules or on the combination of them according to their geographical location. This argument is based on the fact that “relevant information for innovation, whatever its source, can now be cast in frameworks and categories that are more universal” (Arora and Gambardella, 1994: 524). Advances in the theoretical understanding of problems, instrumentation, and computational capability, have given impetus to the use of general and abstract knowledge. Abstract knowledge is understood as the ability to represent phenomena in terms of a limited number of essential elements, whereas general knowledge is described as knowledge that relates the outcome of a particular experiment to the outcome of other, more ‘distant’ experiments. This evolution of an industry knowledge base would pave the way to an increasing division of labour across companies. Some companies would focus on the different components or activities within a given architecture. Other companies would specialise in the ‘mixing and matching’ of these components. 5.3. Technology, products and organisations: toward modular dynamics? The foregoing discussion has highlighted the managerial implications of modularity in product, organisation, and technology. To sum up, the evolution of an industry technological knowledge base into more general and abstract categories allows for a better understanding of the inner workings of products and processes. Such an understanding enables the adoption of more complex design strategies (e.g. modularity). Firms de-couple the development of the product architecture from the development of the modules, and then specialise in either the architecture or a few modules. Furthermore, since the product architecture acts as a sort of ‘glue’, the organisational processes underlying architecture and module development are also separated and carried out independently. The combined effect of the different dimensions of modularity paves the way to a greater division of labour both within and across firms at the level of product, organisation, and technology. Therefore, the literature on modularity offers a number of very strong predictions. There exists a positive correlation between product, organisational and technological modularity. Modular product architectures enable increasing specialisation, within and across companies. Modular product architectures allow for co-ordination to be achieved with minimum managerial effort. Therefore, modularity represents a very powerful concept with cogent managerial implications. Companies designing engineered products use modularity as a design strategy in order to exploit a number of advantages. A modular architecture enables companies to upgrade products throughout their life cycles. This is particularly important in those industrial sectors where the product life cycle is much longer than the components’ life cycle. Modularity also reduces production costs and time. A modular product is easier to maintain. As a result of these points, firms able to offer modular products also use this modularity as a marketing tool. However, research on modularity has left unanswered a number of issues, which the thesis intends to address. In particular, the literature on modularity in arguing that modularity in product, organisation, and technology is positively correlated, does not recognise the different, though intertwined, dynamics underpinning product, technology, and organisation. There is as yet no empirical evidence to support the view 35

that modularity in products, organisations, and technologies is correlated. Therefore, products may show modular architectures, but the mechanisms of technological and organisational co-ordination within and between firms are not achieved through market mechanisms. The following section discusses the works of Chesbrough and Teece (1996), Chesbrough and Kusunoki (1999), and Granstrand et al. (1997) on the issues of organisational and technological co-ordination. 5.4. Technological and organisational co-ordination: what the management literature suggests As discussed in the previous section, modularity advocates argue that due to the positive correlation between product, organisation, and technology modularity, industrial sectors are characterised by a greater inter-firm division of labour at each level. Accordingly, co-ordination across firms is achieved through the market mechanism, since modularity characterises and informs products, the organisations designing and producing them, and the technological knowledge underlying the products. Though management scholars acknowledge the far-reaching effects of the modularity paradigm, they argue that in order to understand the relationships between product, organisation, and technology analysis of further dimensions is required. In particular, the number of external sources, technological change, and the different dynamics underlying technologies and products are factors that should be taken into account. The aim of this section is to critically discuss some studies within the strategic management tradition on the issues of technological and organisational co-ordination (Chesbrough and Teece, 1996; Chesbrough and Kusunoki, 1999; Granstrand et al., 1997) and to highlight their shortcomings. Systemic vs. autonomous innovation and the number of external sources Chesbrough and Teece (1996) deal with the issue of technological and organisational co-ordination within the discussion of the appropriate organisational form for technological innovation. They maintain that the appropriateness of a firm’s organisational form is defined by the interplay of two sets of factors, namely the number of external sources of knowledge and the types of technological innovation. With regard to the former, Chesbrough and Teece argue that when the number of external sources (i.e. suppliers) is low, firms should adopt integrated forms to counteract the likely suppliers’ opportunistic behaviour and vice versa. With regard to the latter, they identify two main types of innovation, autonomous and systemic. Autonomous innovations can be pursued independently of other innovations because they are characterised by standardised interfaces with existing component technologies that are often codified in industry standards. Systemic innovations can be realised only in combination with complementary innovations. The management of this system of innovations requires the continuous exchange and sharing of information and knowledge across units of production since systemic innovations redefine the interfaces “throughout an entire product system” (p. 68). These different features of innovation call for distinct organisational designs. Autonomous innovations can be easily co-ordinated via market mechanisms, since their integration with existing technologies rests on well-defined interfaces. In this case firms can adopt virtual structures that are more responsive than integrated ones, since they can use market-based incentives to access quickly and economically the technical 36

resources they need. Likewise, they can easily form alliances with other companies deemed relevant for the commercialisation of the autonomous innovation. Large integrated companies are unable to duplicate such a responsive mode. Systemic innovations instead are best managed within the same company. When innovations are interdependent, virtual corporations become vulnerable due to their inability to coordinate the required exchanges of knowledge and information through the market mechanism, and to settle conflicts between companies over integration. According to Chesbrough and Teece, integrated companies can therefore take advantage of their scale and their complementary assets when systemic innovations are introduced. Similarly, market leadership is required when new standards have to be set. Although the framework proposed by Chesbrough and Teece helps to clarify the underlying strengths and weaknesses associated with different organisational forms for co-ordinating innovation, it has some limitations. First, Chesbrough and Teece consider product decomposition as an exogenous fact: the autonomous or systemic degree of innovation is given. Although this may sometimes be the case, there is no explicit reference to the firm’s ability to assess and change the decomposability of the product. Second, the role of the number of external sources in Chesbrough and Teece’s scheme points to the interpretative framework proposed by transaction cost economics in relation to the definition of the boundaries of the firms (Williamson, 1975, 1987). As highlighted earlier in Section 4, according to such a framework, firms should vertically integrate in the presence of high asset specificity and a small number of component suppliers, which may lead to opportunistic behaviour by suppliers. This approach fails to recognise that firms may well not carry out any make-or-buy calculation in relation to certain investments, especially when these are perceived to be extensions of their existing capabilities and are therefore carried out irrespective of the number of external suppliers (Kogut and Zander, 1992). Third, according to Chesbrough and Teece (1996) as regards technological capabilities, firms can only be fully integrated (in-house R&D, design, and production) or fully disintegrated (buying externally designed and manufactured items). In other words, the two options proposed are limited to the ‘pure make’ or the ‘pure buy’. Technology phase-shift: from modular to integral and vice versa Building on the framework proposed by Chesbrough and Teece (1996), Chesbrough and Kusunoki (1999) introduced a dynamic element into the analysis of the relationships between product and organisation. Using empirical evidence from the hard disk drive industry, they argue that technologies follow a dynamic cycle that goes from integral to modular. In the integral phase, usually the first stage of technological development, the interactions between systems elements are fast changing and poorly understood. The integral phase calls for tight co-ordination mechanisms typical of a centralised organisation. Such interactions eventually become articulated and codified. In the modular phase, when sub-systems and components and their interactions are better understood and modularised, co-ordination is better achieved through decentralised or virtual forms of organisation. Therefore, firms should align their organisational forms according to the phase of the technology. Firms that do not adapt their organisational forms fall into what Chesbrough and Kusunoki define as ‘traps’. A ‘modularity trap’ emerges when the 37

technology shifts from the modular to the integral phase but firms maintain a decentralised organisational form. Virtual organisational arrangements lack the systems knowledge required to co-ordinate integral technologies. On the other hand, an ‘integrality trap’ emerges when an integral technology becomes modular and firms retain a centralised organisation. In other words, when component interactions are well articulated and codified and innovation activities can be organised via arms’ length market relationships, centralised organisations become cumbersome forms of coordination. The scheme proposed by Chesbrough and Kusunoki thus adds a dynamic element to that of Chesbrough and Teece (1996). However, the framework remains unclear about the distinctions and relationships between products and technologies. It assumes that modular (or integral) products are characterised by modular (or integral) technologies and therefore, there are two main organisational arrangements (i.e.: virtual or integrated) to co-ordinate innovation in the two phases. Technology vs. product: technological and organisational co-ordination Building on Chesbrough and Teece (1996), Granstrand et al. (1997) draw some conclusions on firms’ technology outsourcing decisions. They distinguish two sets of factors that affect corporate outsourcing decisions: (a) the degree to which the innovation is autonomous or systemic and (b) the number of independent suppliers outside the firm. On these grounds, they propose a two-by-two matrix that identifies four cells. Each cell is associated with a different case calling for a particular degree of internal technological capabilities. In particular, Granstrand et al. identify four intermediate corporate positions between full integration and full disintegration, namely exploratory research capability, applied research capability, systems integration capability, and full design capability. For instance, when the number of external sources is low and the innovation is systemic, then companies should maintain a wide range of in-house capabilities, from exploratory and applied research down to production engineering. Granstrand et al. have enriched the framework proposed by Chesbrough and Teece (1996). In identifying the above-mentioned intermediate stages between full integration and full disintegration, they contend that decisions related to products are distinct from those concerning their underlying technological capabilities. They argue that “whatever the type of innovation and the number of external sources, companies should maintain capabilities in exploratory and applied research in order to have the competence to monitor and integrate external knowledge and production inputs” (p.20). This identifies a critical deficiency in the research that does not distinguish between technologies and products (e.g. Singh, 1997). However, the framework proposed by Granstrand et al. also has its limitations. Since it is based on that proposed by Chesbrough and Teece (1996), it inherits its weaknesses. First, the degree of product or system decomposability is assumed to be primarily exogenous. Second, it assumes that transaction costs rather than strategic judgements about corporate capabilities determine the boundaries of the firm. Components’ hierarchy Another interesting concept used by management scholars to understand corporate boundaries is that of component hierarchy. Component hierarchies are based on the assumption that the role of each component is different within a product. According to 38

their technical features, performances, and costs, their importance varies greatly. Organisations, therefore, conceive hierarchies of subsystems and components in order to identify which are the peripheral and the key components and consequently adjust their technological capabilities. An interesting approach to analysing the hierarchical role of components in a system has been put forward by Maïsseu (1995). Considering three variables, namely the impact of the component cost on the cost of the overall system, its influence on the quality of the system, and the technological maturity of the component, he proposes a matrix which identifies four categories (Table 2.1). According to this approach, it is possible to determine the relative weight of each component. Thus, components with low impact in terms of quality and cost of the end product and whose underlying technologies are mature are to be considered to be trivial. Then, there are basic components whose cost is relatively high, while their technologies have reached the maturity stage. Key components are those whose characteristics heavily influence the quality of the end product and whose technologies are at the initial stage, but which do not affect the cost of the system to a great extent. Finally, there are the critical components. Their influence in terms of cost and quality is relatively high and their underlying technologies are at the initial stage. It is worth stressing that this approach heavily underlines the issue that components may evolve across the hierarchy over time. Technological change occurring at different levels of the systems may shift the relative hierarchy of components and system-level critical problems. This may well inform firms’ outsourcing decisions. Pavitt (1998) argued that a critical issue that companies take (or should take) into account when outsourcing components is the rate of change of the underlying technologies and the ensuing technological imbalances (Rosenberg, 1976). When technologies advance at different rates then companies should be able to keep pace with them and incorporate changes in their product and their components’ hierarchies. Table 2.1. Components’ hierarchy (adapted from Maïsseu, 1995) Critical Key Basic component component component strong weak strong Cost impact

Trivial component Weak

Quality impact

strong

from weak to strong

from strong to weak

weak

Technological maturity

first stage

first stage

maturity

maturity

Similarly, it is worth noting that changes in the underlying technologies can modify markedly the relative hierarchy of components. As hierarchies are usually constituted according to a series of ‘rules’ valid at a given point in time, they can provide predictions as long as the ‘rules’ remain unchanged (Allen, 1994). Therefore, a hierarchical taxonomy concerning products made up of many components may be undermined by changes in the underlying component technologies. Evolution may be endogenous in that changes can occur within the system itself, stemming from existing as well as exogenous technological trajectories, that is to say existing technologies can be replaced by new ones or new technologies can be added (Allen, 1994). 39

6. Conclusions This chapter has discussed the two streams of literature the thesis addresses, namely the resource-based view of the firm and the division of innovative labour. As regards the former, the chapter argued that the resource-based perspective represents a more useful and comprehensive approach to understanding firm’s behaviour compared to other approaches such as the contractual view of the firm. The resource-based perspective puts emphasis on the cumulative development of firm-specific capabilities as a source of firm’s competitive advantage. The chapter also highlighted that research on firms’ capabilities has mostly relied on aggregate measures and focused on the analysis of a limited number of sectors, such as consumer electronics and automotives. The thesis relies on the resource-based approach as a ‘framework of appreciation’ (Nelson and Winter, 1982) to analyse the changing boundaries of the firm. It focuses on the technological knowledge bases of firms operating in the aircraft engine industry. The thesis understands a firm’s technology base as a stock of technological capabilities (Dierickx and Cool, 1989). In order to investigate the evolution of the boundaries of engine manufacturers’ technology bases, the thesis identifies two dimensions of technological capabilities, namely breadth and depth. The aircraft engine industry was specifically chosen because of the significant managerial implications posed by the multitechnology, multicomponent nature of the aircraft engine. The multicomponent nature of the aircraft engine requires engine manufacturers to maintain a differentiated set of technological capabilities in terms of component knowledge, knowledge about the ways in which the components are linked together and interact as a system, and knowledge about the system as a whole. The multitechnology nature of the aircraft engine demands that engine manufacturers develop capabilities in multiple technological fields. However, due to financial and managerial constraints, engine manufacturers usually focus on those capabilities regarded as crucial and contract out peripheral ones. This leads to critical outsourcing decisions and puts severe demands on the types of R&D activities that are carried out in-house. As regards the composite stream of literature on the division of innovative labour, the chapter argued that researchers belonging to heterogeneous disciplines such as management theory, industrial economics, and design theory have concentrated their discussion on the concept of modularity. The chapter highlighted that this literature is characterised by a number of shortcomings. Some interpretative frameworks on the relationships between products, organisations, and technologies proposed by some management scholars were critically reviewed and discussed. Below is an integrated account of the main shortcomings of the literature reviewed along with the terms of the contention of the thesis. First, with the exception of Granstrand et al. (1997), the research discussed does not recognise that technologies and products follow different dynamics. Two main positions emerge that reach the same conclusions. The first does not explicitly distinguish the technology and product dimensions. The corollary of this position is that these two dimensions are collapsed together, so that modular or integral products are characterised respectively by modular or integral technologies (Chesbrough and Kusunoki, 1999; Chesbrough and Teece, 1996). The second position explicitly distinguishes between technologies and products, but it contends that they are informed by the same principles (e.g. modularity) (Arora et al., 1998). The thesis argues that the 40

dynamics of technologies and products follow different, though intertwined paths (Chapters 5, 6, 7, and 8). Second, in many cases only two alternative forms of organisation are considered: virtual (or decentralised or modular) and integrated (or centralised). For example, Chesbrough and Kusunoki (1999) give important hints about the influence of technological change on firms’ organisational arrangements, by proposing that technologies develop in cycles (from integral to modular and vice versa). However, they do not recognise fully the analytical importance of the finding that the success of Fujitsu (one of the firms they studied) during a shift from a modular to an integral phase was due to its systems knowledge. Fujitsu was the company that proved to be the most successful in the shift from the modular to the integral phase in the hard disk drive industry. Unlike the other companies that fell in the ‘modularity trap’, its systems knowledge enabled the mastery of the new integral technology (i.e. the magneto-resistive head) 17. Chesbrough and Kusunoki also argue that in a possible shift back from the integral to the modular phase, Fujitsu may have taken the appropriate strategic steps in relation to technological investments and organisational set-ups with other companies, in order to align its organisational form with the technological shift. In spite of these insights, their prescriptions about organisational arrangements (embedding modular or integral technological knowledge) are confined to the two classical ones, namely virtual or integrated. A corollary of this position is that the definition of constant ‘modular’ product interfaces informs also the technological boundaries of the firm and as a consequence its external relationships. In other words, it is argued that component level interfaces define the pattern of division of labour across firms and co-ordinate their relationship thus minimising the need to exert explicit managerial effort. The thesis shows that although the product aircraft engine is becoming more modular and engine manufacturers outsource the production of larger engine parts, they still maintain a broad and deep set of in-house technological capabilities to co-ordinate from a technological viewpoint the network of actors involved in the industry (Chapters 5, 6, and 7). Third, all the frameworks reviewed pay inadequate attention to the rates of improvement in performance, enabled by changes in the technologies underlying the products. The thesis contends that technologies are characterised by uneven rates of development. Uneven rates of technological change can create performance imbalances amongst components and subsystems that may require appropriate technological capabilities to co-ordinate their design and development. These uneven rates of change, therefore, do and should inform companies’ outsourcing decisions, especially in those industrial sectors characterised by multicomponent, multitechnology products. The thesis argues that in the case of components characterised by fast moving technologies, such as digital electronics, engine manufacturers develop and maintain in-house technological capabilities in order to be able to integrate the components back into the engine (Chapter 7).

17

“This was due to its (Fujitsu's) continued investments in systems knowledge and materials and component technology in its R & D labs.” (Chesbrough and Kusunoki, 1999: p.13)

41

Fourth, the research reviewed above neglects firm-specific factors such as firms’ capabilities to assess and modify product decomposability. In the frameworks analysed above, the systemic (or integral) and autonomous (or modular) nature of products is exogenous. Firms may in fact assess product decomposability differently. In particular, a firm’s outsourcing decisions are a function of the interplay of its understanding of the nature of products (systemic or autonomous) and its component hierarchy. The thesis argues that firms may well maintain in-house independent technologies because they are deemed critical for final product performance. In other words, the thesis takes into account the issues of component hierarchy recognising that components and subsystems are characterised by different degrees of criticality and, accordingly firms adjust their technological capabilities (Chapter 6). The thesis also highlights that this concept of component hierarchy is firm-specific. That is, the way firms perceive and envisage the role of each subsystem is a function of their past R&D investments, the scientific and technological background of firms’ scientists and engineers and the technological trajectory underlying the product. In short, it is the firm’s historical process that shapes the hierarchy of the in-house of technological capabilities that underlie components (Chapter 7).

42

CHAPTER 3: RESEARCH METHODOLOGY 1. Introduction The previous chapter reviewed the relevant literature for the research undertaken in this thesis. The aim was to define the theoretical and analytical framework that guides the thesis by determining what is known about firms’ capabilities and the inter-firm division of labour and by highlighting the shortcomings of the literature reviewed. The purpose of this chapter is to outline the research methodology employed to address the thesis research questions. It traces the process that shaped the research design and documents data sources, data types, and database compilation methods used in the thesis. The literature reviewed in the previous chapter made a substantial contribution to the analysis of firms’ capabilities and inter-firm division of labour. However, a number of issues were left unclear in relation to the boundaries of the firm. This research intends to address these issues. The thesis analyses the boundaries of firms operating in the aircraft engine industry by taking into account that products and their underlying technological knowledge follow different dynamics (Granstrand et al., 1997; Pavitt, 1998). In order to have a more detailed understanding of the boundaries of the firm, the thesis identifies two dimensions of firms’ technological capabilities, namely breadth and depth. The analysis of breadth and depth of firm’s technological capabilities is based on comparative quantitative analysis (based on patent statistics) and case study methods. The combination of these two methods enables an analysis of the phenomenon from different perspectives. Using two methods allows also methodological triangulation (Yin, 1994). The study is based on two distinct types of data, namely quantitative and qualitative. These two forms of data have been used both as supplements for mutual verification (Glaser and Strauss, 1967) and as complements to deepen and corroborate specific findings. The data were collected from multiple sources. Using multiple data sources allows data triangulation. Qualitative data were collected through a systematic review of the technical literature, specialised engineering journals, trade publications, and interviews. Interviewees included company personnel, industry experts, and academics. US patent statistics, collaborative agreements, and product data were the main sources of quantitative data. Some company-specific quantitative data were also gathered via interviews with companies’ engineers using a structured questionnaire. This chapter is organised as follows. Section 2 documents the research process. Section 3 describes the companies included in the study. Section 4 documents sources, methods of collection, and database compilation methods in relation to product and collaborative agreement data, whereas Section 5 analyses the use of patent data. Section 6 summarises methods of collection and sources of qualitative data. Section 7 presents the conclusions to this chapter. 2. The research process The thesis research method aims to integrate qualitative and quantitative data to exploit their synergism. The idea is to provide quantitative support for the phenomenon under study corroborated by qualitative information. Qualitative and quantitative data are used in a mutually reinforcing way. The quantitative analysis is a useful testing ground for qualitative insights. On the other hand, qualitative information provided by the 43

interviewees is an extremely useful source of insights for the interpretation of the results of the quantitative analysis. It also helps pose questions to be explored further in the quantitative research. The use of multiple sources of evidence helps establish solid construct validity 18 (Yin, 1994). Rather than being set and fixed from the beginning, the research design has unfolded during the research process. The original idea was to analyse the changing boundaries of engine makers’ technology bases by performing a comparative quantitative analysis using US patent and collaborative agreements data. The plan was to carry out on-site interviews to validate the accuracy of the quantitative analysis. The quantitative patent analysis had two outcomes, however. First, it provided interesting findings that addressed the research questions (i.e. a thorough account of the breadth of engine makers’ technological capabilities as shown in Chapter 5). Second, it helped identify a number of cases that warranted further and finer-grade scrutiny in relation to both breadth and depth of engine makers’ technological capabilities. Based on the results of the patent analysis and, in particular, on its limitations in giving an all-round account of engine makers’ capabilities, a multiple case study design was adopted to investigate some of the cases that the analysis of patent data identified (Yin, 1994). The cases analysed are related to (a) engine makers’ all-engine capabilities that are embedded in specific organisational groups (Chapter 6); (b) a critical engine subsystem, namely the fan (Chapter 6); and (c) the engine control system (Chapter 7). These cases turned out to be very revealing with respect to the thesis research questions as they enabled a deeper analysis of the engine makers’ technological capabilities. In addition, the cases share some the characteristics of what Eisenhardt (1989) labelled as ‘emergent themes’ or ‘special opportunities’ that require adjustments to data collection instruments. In fact, in order to investigate these cases in detail new ad-hoc questionnaires were developed (Appendices 2 and 3). The research process was both deductive and inductive, running from the more macro quantitative analysis to the more micro case studies and vice versa. For instance, as regards the case study of engine control system, based on qualitative insights gathered during interviews, the patents were re-arranged in order to provide a more comprehensive understanding of the depth of engine makers’ technological capabilities as compared to suppliers’ in relation to that specific subsystem (Chapter 7). In this way, the qualitative insights were employed to exploit further the patent data. 3. The companies analysed This study focuses on the analysis of changing boundaries of the world’s three largest aircraft engine makers, namely General Electric Aircraft Engines, Pratt & Whitney, and Rolls-Royce. In the trade literature, these companies are usually called the Big Three or the Primes. They dominate the large turbofan market. They are also present in the small turbofan market both directly and via joint ventures. The study also includes a comparative analysis of the world’s two largest risk and revenue sharing partners

18

Construct validity is one of the four criteria employed to assess the quality of social science research. The other criteria are internal validity (see footnote 28), external validity (see footnote 30), and reliability. Construct validity refers to the measurement of the variables under examination, “how we measure what we want to measure” (Judd et al., 1991: 30). According to Yin (1994: 33), establishing solid construct validity means “establishing correct operational measures for the concepts being studied”.

44

(RRSPs) and some engine control systems suppliers. The names of the companies have been disguised to respect confidentiality. The focus of the research is on civil aviation. Most companies in this sector produce engines for both military and civil applications. Companies keep military and civil divisions separate and at the product level the distinction is clear-cut. There are, however, certain caveats in drawing the line between military and civil propulsion. From a technological viewpoint, in fact, the relationships between military and civil are rather fuzzy and therefore difficult to single out. Technologies developed for military applications may be transferred to civil applications, and vice versa. Whether the direction of the relationship is mainly from military to civil or the other way round is debatable and actually debated. It almost certainly was one way from military to civil in the past. However, as Bill Gunston (Jane’s Aero Engines, 1996) argues the trend has been reversed, so that military engines are now derivatives of civil ones. The situation is further complicated since companies do pursue job rotation policies where engineers and managers who have been involved in military programmes are transferred to civil programmes and vice versa. The technological requirements of military and civil propulsion, however, still differ in several respects. For instance, military aviation is extremely responsive to ultimate performance, whereas reliability is a priority in civil engines. By the same token, environmental factors, such as noise and emissions that are virtually neglected in military propulsion, are of the greatest importance in the civil one (Jane’s Aero Engines, 1997). 4. Product and collaborative agreement data 4.1. Product data The data related to engine versions and families have been gathered from the section of Jane’s All the World Aircraft’s that is dedicated to aircraft engines. This section lists data related to engine versions and families in production. It also gives information in relation to entry into service, termination of production, and engine performance indicators such as weight, take-off thrust, and cruise thrust. The data collected relate to the engines produced by the Big Three between 1977 and 1996. The data have been entered manually using Microsoft’s Access software package. As a general rule the thesis has followed Jane’s All the World Aircraft’s publication method to identify new engine families or versions. New engine families are entered in the database when Jane’s All the World Aircraft’s publication has listed at least one version and its related performance. On the other hand, an engine version (or an engine family) no longer listed is discontinued in the database. The data on engine versions and families are employed to examine engine makers’ product strategies. In particular, the data are used to find out whether aircraft engine makers use modular design strategies. As explained in Chapter 4, a modular engine family enables engine manufacturers to employ the same engine architecture for different engine versions catering for different market niches as defined by thrust requirements. Therefore, the number of engine versions per family is used as a rough measure of engine modularity, since the greater the number of engine versions per family the more modular the engine architecture.

45

4.2. Collaborative agreement data The data related to collaborative agreements are collected by searching the SDC (Securities Data Corporation) database and the DIALOG database. The SDC database contains data of around 52,000 strategic alliances, joint ventures, long term contracts, and the like. The data refer to all industrial sectors between 1985 and 1998. Each entry contains information concerning partners, the technologies involved, the dates of signing and termination, and an abstract containing even more detailed information regarding the agreement. The names of the companies were used as search criteria to gather data related to collaborative agreements. For companies belonging to larger industrial groups both the name of the division, e.g. Pratt & Whitney, and the name of the parent company, e.g. United Technologies Corporation, were used. Using these criteria 135 entries were obtained. After reading the information related to each agreement, 12 agreements were deleted as they were considered either technology unrelated (e.g. landing gear) or business unrelated (because they had been signed by divisions involved in other businesses of the parent companies). Furthermore, a further 17 agreements were deleted because they were considered to be redundant. All in all, this thesis has analysed 106 agreements from the SDC database. This ‘cleaning-up process’ did not bias the data set, since both participants and activities involved in each collaborative agreement were carefully reviewed. The SDC data are complemented by data gathered from searching the DIALOG database. This much larger on-line database contains company-related information. The same search criteria (i.e. companies’ names) were used in order to retrieve a large amount of data related to each company. After a careful review of these data 23 collaborative agreements were identified. The information related to each agreement was classified according to the SDC database format. The two data sets were then compared and 10 of the agreements retrieved through DIALOG were added to those obtained through SDC. The difference in the number of entries obtained searching the two databases using the same criteria was due to the different sets of journals used during the database compilation. The number of collaborative agreements analysed was 116. Of these only the signed ones were considered, that is to say 95. These data are employed to assess the dynamics of firms’ external linkages. Whereas patent data give an account of firms’ internal technological capabilities (explained in Section 5), data on external linkages, such as alliances, joint ventures, licensing, measure the extent of inter-firm division of labour (Chapter 4). In order to have a more detailed understanding of the kinds of activities involved in each agreement, the content of each was reviewed and arranged accordingly. A two-tier classification was devised in order to investigate at which level the division of labour occurs in the industry. This classification is based on six types of activities, namely research, design, manufacturing, testing, maintenance, and other. This latter category includes contracts such as equity acquisition. Those agreements involving engine design and manufacturing were further classified in relation to the scope of the deal. Four types of agreements in relation to their scope were identified: (1) design and manufacturing of new engines, that is new engine development programmes; (2) design and manufacturing of components; (3) manufacturing of engines and components (this is the typical scope of agreements set up by the Big Three with Russian companies); (4) those involving airframers as participants. The latter type 46

of agreement regards airframers’ early involvement in new engine programmes. Table 3.1 describes the two-tier classification. Table 3.1. Collaborative agreement two-tier classification. Activity Scope Design/manufacturing Engine Design/manufacturing Component Manufacturing Engine/component Research Testing Maintenance Design/manufacturing Airframer Other 5. Patent data The technological capabilities of the companies in question were measured by a count of the patents granted in the US over the period 1977 to 1996. The technological profile of each company built using US patents is a ‘device’ employed to understand the boundaries of their in-house technological capabilities (Chapters 5 and 7). 5.1. The company dimension From US patent data it is possible to identify the assignee firm (i.e. the firm the patent has been granted to) as well as the type of the technological activity with which the patent is associated. At the company level the objective is to establish which patents were granted to the companies involved in the industry in question. For some companies, patents from the US Patent Office CD-ROM were collected and then doubled-checked against the legal department of the companies where these were mismatches between assignee names as attributed by the US Patent Office and company names. This is the case for those companies active only in the aircraft engine industry or when an aircraft engine division takes out patents in the US with its own assignee name and code. However, this is the exception, since most of the companies involved in this industry belong to bigger groups active in different businesses. Their legal departments are centralised so that they patent in the US (but also in Europe) under just one assignee name. To consider the patents granted to the entire group to assess their technological activities in only one sector, namely aircraft engines, would have been misleading. Likewise, analysing the patents belonging to the aircraft engine division using only those US patent classes deemed to be ‘typical’ of the sector in question would have led to misleading results. The study would have missed companies’ technological diversification strategies at the level of the aircraft engine division. In order to avoid this problem, the study singled out those patents of interest to the aircraft engine industry in different ways. For instance, it selected patents based on inventors’ addresses. However, companies usually have manufacturing plants and research centres dealing with different product markets in the same country, and in the US in the same state. Therefore, in order to compile a list of US patents related only to the aircraft engine business, the companies were contacted directly19. After lengthy

19

This was long and painstaking. Most companies’ contacts had to be built from scratch involving a lengthy process of letters of introduction (also via Italian Embassies) and personal recommendations.

47

complex negotiation they agreed to make available lists containing the US patents granted to the aircraft engine related companies from 1977 to 1996. In most cases the legal department in charge of the overall group patenting policy was contacted directly. However, in some cases the legal departments of the different divisions related to the aircraft engine business were contacted. In the former case, a list of US patents already consolidated was obtained, that is to say regardless of whether the patents had been taken out by the central research centre or by a company belonging to the group designing, for instance, electronic control systems for aircraft engines. In the latter case, the patents were consolidated afterwards. In both cases, however, the patents were double-checked with the US Patent Office CD-ROM and web site. The final database encompasses the US patents granted to the Big Three and the world’s two largest RRSPs. The study uses granted patents since it has not conducted a time-based analysis of research activities and possible related outcomes. However, the author is aware of the time lag (usually 18 months for US patents) between patent application and publication. 5.2. The technology dimension As regards the technology dimension, the aim was to organise the data into common types of technological fields to suit the level of analysis desired. US patents are grouped together in classes, subclasses and sub-subclasses, and there is a marked overlap between class definitions. Classes contain several subclasses and subsubclasses so that patents belonging to the same class can be very ‘distant’ in terms of their technological content and vice versa, patents belonging to different classes can be closely ‘related’. For example, the US patent class 60 entitled ‘Power Plants’ encompasses technologies ranging from engine control and fuel systems, to combustors and exhaust nozzles, and from fuel nozzles to tip clearance control systems. Therefore, as the aim of the study was to achieve a detailed understanding of the technological capabilities of the companies involved in one industry, using the US patent main classes as the primary level of aggregation would have led to rather broad and imprecise conclusions. This made the use of patent statistics for measuring and assessing firms’ technological capabilities more difficult. To tackle this issue, the research proceeded to break down and rearrange the US patent own classes according to a sector specific map. The idea to break down the US patent classes has drawn from a previous study, which carried out a first attempt to extract a more detailed set of information from US patent data in relation to companies’ technological capabilities (Prencipe, 1997). In this earlier study, a sector specific map according to the technologies required for aircraft engine components was built. A patent-by-patent analysis was required in order to assess the technological content of each and define the fourteen technological fields composing the map. The technological content of the patents was assessed analysing the titles and where possible the abstracts. The technological fields comprising the map were the result of the aggregation of the US Patent Office own classes. This first attempt turned out to be very useful for the thesis for two main reasons: (a) it revealed much about the US patent classification in general and more specifically about those classes related to the aircraft engine and the way they are organised, and (b) knowledge was gained about the technologies related to the design, development and manufacturing of the aircraft engine. 48

The rearrangement process of the US patent classes employed in this research proceeded in three steps, namely information-gathering, regrouping, and validity checking. Information gathering Building on the previous experience, this study first went through an extensive literature survey of technical publications on aircraft engine technology in order to gain a better understanding of the functioning principles of the aircraft engine. Bill Gunston’s publications (1995) and a publication by Rolls-Royce (1986) on the aircraft engine proved to be extremely useful. The section concerning aircraft engine technology in Jane’s All the World Aircraft publication also was very helpful. Relying on these publications, a component checklist encompassing the main sub-systems of the aircraft engine was constructed. This checklist included also those technologies that could not be ascribed to just one component or subsystem, notably manufacturing and material technologies. The checklist had two purposes. First, to inform the rearrangement of the US patent classes and, therefore, to form the basis of the sector specific map. Second, to rank the importance of each component/subsystem in terms of criticality for the engine system’s performance. Regrouping In order to regroup the patents, as a general rule the study relied on the titles and definitions of US classes and subclasses. The manual of US patent classification available on the web site was extremely useful. Each class and subclass definition specifies whether the patents relate to processes or products. This enabled a distinction to be made between product- and process-related patents as well as between patents related to material and testing technologies. Furthermore, three telephone interviews with two primary examiners of the US Patent Office were conducted. The interviews lasted between 30 and 45 minutes. The two interviewees were primary examiners of US patent classes 60, 415, and 416. They were specifically selected because more than 60% of the patents analysed fall into one of these three classes. The interviewees provided invaluable information in relation to class and subclass content. The regrouping process was an iterative one. It went from class definition through subclass definition to patent title and abstract and vice versa. In fact, even subclass definitions can be very broad and obscure. A patent-by-patent analysis was performed, taking into account title, abstract, and full text of every patent 20. In particular, the full text of US patents indicates the field of application of the invention. This analysis allowed a better understanding of the technological content of the patents classified in the subclasses, leading in turn to a confirmation and/or an understanding of the subclass content. This process was long and arduous since the study deals with more than 6,000 patents. During the regrouping process, companies’ legal departments were also consulted. As already mentioned for those companies where it was possible to identify patents related to the aircraft engine business, the procedure was to double-check with them the list of US patents analysed in case of any mismatch between assignee names and company names, or in case the results had missed (or included) suppliers that did (or did not) 20

The abstract and the full text of each patent are available at the US Patent Office web site.

49

belong to the company. Companies’ legal departments were contacted in some cases in order to ascertain the relationship of patents with any of the technologies underpinning the aircraft engine. For instance, some patents were associated with the use of computerised tomography, and were classified under class 378/4 of the US Patent Office. This class is specifically related to ‘living organisms’. These patents were difficult to relate to the aircraft engine. Once contacted, the legal department of the company holding these patents explained that the company was trying to apply this technology to scan turbine blades in order to identify flaws resulting from the manufacturing process. It is worth noting that US patent classification is continuously subject to change. This is due to the ever-increasing progress of knowledge bases including a technology fusion phenomenon, such as optoelectronics, and mechatronics. This then requires the addition of new classes and subclasses. As a consequence, patents can be reclassified. As a general rule for the thesis, rearrangements relied only on the latest classification. Furthermore, given this continuous reclassification by the US Patent Office itself, the regrouping process was concentrated between May and October 1997. The changes that took place during that period of time were identified but did not affect the final outcome of the process. Previous classifications proved useful when the current class and subclass definitions seemed rather obscure. Validity checking In order to check the accuracy of rearrangement, the final sector-specific map was independently checked by a company engineer who offered helpful suggestions on each technological field, the boundaries across them, and possible further aggregations (Section 5.3 and Table 3.3). Seventy patents randomly picked were independently checked by the same company engineer in order to further verify the accuracy of the rearrangement. 5.3. The sector-specific maps The use of technology mapping to study a firm’s technology base has been proposed by several authors (Patel and Pavitt, 1997). Goodman and Lawless (1994) suggest a technique for technology mapping. This technique comprises a number of phases, from the identification of the basic technologies underlying firms’ product and subsystem elements of the design, through the definition of key parameters and matching with relevant technologies, to the assessment of competitive forces in the industry and technology strategy formulation. The thesis does not go through all the phases put forward by Goodman and Lawless. Instead, based on the regrouping process explained above, it introduces a sector-specific map encompassing the technological fields underlying the engine components. By using this technology map, it was possible to build a technological profile of the companies under study. The companies’ technological profiles are understood to reflect their technology bases or, in other words, their set of in-house technological capabilities. Therefore, the analysis of companies’ technological profiles enables an evaluation of the dynamics of the boundaries of their technological capabilities over time. One of the characteristics of this technology map is its in-built flexibility, in the sense that it is possible to re-group the fields according to broader categories. In fact, the 58-technological field map has been aggregated to obtain the 23-technological field

50

and 7-technological field maps. Table 3.2 presents the three maps and how they relate to each other. The 23 technological fields were clustered to form five broader technological fields, that is ‘Product’, ‘Manufacturing’, ‘Material’, ‘Testing’, and ‘New Architectures’. Then, relying on information from interviews with company engineers and academics, those technological fields related to product-technologies were further grouped according to their impact on overall system performance. This process identified a two-level architecture in the engine system. The technological fields related to those subsystems that play a pivotal role in engine performance, such as fan, compressor, turbines system, combustion chamber, sealings, and control systems define the inner-core of the engine. By contrast, the outer-core of the engine is comprised of those subsystems playing a relatively marginal role in the economics of the engine system (e.g. couplings, nacelles, containment structures). The patent database covers a twenty-year period from 1977 to 1996. This period was divided into four five-year periods or two ten-year periods according to the specific research question and the statistical method used to answer it. Statistical methods encompass correlation and regression analysis. As regards regression analysis, the study uses the method proposed by Cantwell (1993). The method is fully described in Chapter 5. Listed below are the macro technological fields and their content. Product-related patents As regards product-related patents, the regrouping process followed the component checklist mentioned in Section 3.2. Each of the technological fields identifies a group of technologies underlying a set of components and subsystems constituting the aircraft engine (e.g. rotor assembly, stator assembly, sealings). The map distinguishes components as detailed such as vanes or blades, combustor or combustor liners, rotors or stators. Manufacturing-related patents As regards manufacturing-related patents, the distinction between ‘plants and processes’ stated in the class or subclass definition was taken into account in order to distinguish between methods (processes) of manufacturing from manufacturing equipment. This distinction led to different technological fields in the sector specific map, such as ‘Metal Working and Metal Treatment Processes’ and ‘Metal Working and Metal Treatment Equipment’. Likewise, electrical-, electrochemical-, and optics-related patents were grouped in separate technological fields. Patents related to chemical and coating technologies constitute two distinct technological fields. Material-related patents As regards material technologies, the map identifies the technological field referring to methods of manufacture of composite materials as distinguished from the technological field including composite materials. Metal powder composites and metal alloys and superalloys form different and separate technological fields. Testing This technological field comprises patents related to technologies employed to perform testing activities on engine components (e.g. forgings) as well as the entire engine system (e.g. test cell). 51

New architectures Patents falling within this technological field relate to new engine architectures (e.g. propfan and aft-fan) as well as to components and subsystems associated with them (e.g. pitch fan control systems). Control system-related patents Patents related to engine control systems were rearranged in relation to their underlying technology (hydromechanical, optical, electronics) as identified in US patent class or subclass definition. The engine control system was also selected as a case study. The patents related to this subsystem were rearranged in order to shed light on the issue of division of labour amongst engine makers and suppliers in terms of architectural and component knowledge. Thus, the patents were classified into four main groups, namely architecture, sub-architecture, functional, and physical. The break down into these four groups was suggested and carried out by the Head of Engine Control System of one of the world’s largest engine maker companies on the basis of the content of the abstract of each patent. Table 3.3 reports the definition of each of the four groups. 5.4. Limitations of the patent analysis The use of patent data as a proxy measure of companies’ technological capabilities has its own limitations as discussed in the literature (Scherer, 1988; Wyatt et al., 1988). It is worth noting here that: (a) Patents carry different importance as a means of protection of new products and processes across industries due to the nature of the technologies used (more or less patentable) and to the propensity to patent (Levin et al., 1995). This leads to interindustry and inter-sectoral differences in the likelihood of patenting new products and processes. (b) The propensity to patent can vary across companies within the same sector and, what is more, in this study as only US patents are used there are international differences as US-based companies are more likely to patent in the US than non-US companies. (c) Patents do not allow measurement of capabilities in software technologies, as copyrights are generally utilised as a means of protection. (d) Patent data measure only codified knowledge and therefore cannot offer any indication as to the importance of tacit knowledge in firm-specific competencies. A similar limitation is found in other potential indicators of a firm’s scientific and technological activities, such as publication in scientific journals. However, as Hicks (1995) argues, publications point to the existence of underlying tacit knowledge and skills possessed by the authors. This was confirmed by one of the Technical Directors interviewed during the fieldwork who argued “Our core capabilities lie in the knowledge behind design codes and patents” (Interview, 1997). (e) What is more, in relation to this study, it is worth mentioning two further limitations. First, as companies were asked to provide those patents related only to the aircraft engine business, there might be a subjective bias of the understanding of 52

the ‘relatedness’ to one business or another that in turn might have affected the patent data used. Second, although the focus of the thesis is on civil aviation, some of the technologies patented by the companies examined can be and are used in both civil and military applications. However, those patents explicitly related to military applications (e.g. afterburner and very-short-take-off technologies) were placed into the ‘Other’ technological field (see Table 3.2). On the other hand, viewing patents as indicators of a firm’s technological capabilities does present a number of advantages such as quantitative accuracy, considerable detail, and accessibility. Specifically, in using US patents as a proxy measure of companies’ technological activities, it is worth noting that companies usually patent in the US soon after patenting in their home country since the US is the single largest national market (Soete and Wyatt, 1983; Pavitt, 1988). Furthermore, inter-industry variation is not a problem in this study since the analysis is a comparison between firms in the same industry and not between firms from different industries. As far as inter-sectoral and (international) inter-firm variations in the propensity to patent are concerned, the index used in the thesis, namely the revealed technological advantage (RTA), normalises for both variations (Balassa, 1965; Cantwell, 1993; Soete, 1981). A more detailed description of this index is carried out in Chapter 5. In addition, as explained in the next section, the thesis uses qualitative data to overcome some of the limitations of the patent analysis. Qualitative data are used to validate the accuracy of the results of the patent analysis and to write case studies which, in turn, enrich and test the findings of the quantitative analysis. 6. Qualitative data and case study 6.1. Data sources Qualitative data were collected during fieldwork research conducted in Europe (England, France, Italy, Sweden, and Germany) and the US between 1996 and 1999. The main data sources are described below. (a) Technical literature. A systematic review of technical literature was carried out including trade publications (such as Flight International and Interavia), specialised engineering journals, and industry-specific publications (such as Jane’s Aero Engines). The data this yielded were employed to provide background information and to sketch an overall picture of the aircraft engine industry (Chapter 4). (b) Archival record. Several types of documentation were collected and analysed including companies’ annual reports, organisation charts, internal documents, and publications. These documents were employed to corroborate, enrich, or rectify information collected from other sources. (c) Attendance at conferences. The main topics of the thesis were discussed during attendance at various conferences with both industry observers and managers involved in the same industry or in related ones. These discussions greatly enriched and sharpened the research questions and provided feedback on findings during the research process. (d) Interviews. Interviews represent the main source of qualitative data in the thesis. Interviews were conducted using both structured and unstructured questionnaires to gather information on general and specific topics and to identify key issues relevant to the thesis. The outline of the questionnaires can be found in Appendices 1, 2, 3, and 4. The questionnaires were sent in advance to allow the interviewees to become familiar with the kind of information required. In total, 61 in-depth interviews were 53

conducted with different actors in the aircraft engine industry in order to gain different perspectives on the industry. Interviewees included company engineers, industry experts, academics, technical experts from the regulatory bodies, and also aircraft pilots. Interviews were also used to collect quantitative data in relation to firms’ make-buy strategies (Chapter 5). 6.2. Collecting information through interviews The aim of the firm-based interviews was twofold: (a) To obtain feedback on the validity and the accuracy of the results of the patent analysis and therefore overcome some of its limitations 21; (b) To gather both qualitative and quantitative company-related information for the case studies. The following steps were followed in order to achieve each objective. As regards the first, key informants were identified through the company Public Relations Division or Legal Department 22. A questionnaire was developed in order to discover to what extent the findings of the patent analysis reflected firms’ in-house technological capabilities and to identify the key issues in relation to firms’ make or buy strategies. Up to three people per company were interviewed with each interview lasting between one and three hours. The interviews were carried out on-site. Where possible, company visits were made in order to gather further information. Interviewees were all senior staff, such as Vice Presidents of Technology, Technical Directors of the Advanced Engineering Department, Head of Design Technology, and alike. All of them were either currently or had been involved in engine design activities. In some cases managers in the Strategy or Marketing Departments (but always with a technical background) were interviewed. All in all, 18 in-depth face-to-face interviews were carried out in three engine manufacturers and two RRSPs. In some cases, a confidentiality agreement had to be signed, in others it was guaranteed that the company name would be kept anonymous. Further nine interviews (either face-to-face or by phone) were carried out with two engine manufacturers and four component suppliers involved in the industry that were not included in the patent analysis. Information gathered through these interviews was used as background. The second aim of the firm-based interviews was to gather information related to specific cases. As mentioned in Section 2, the analysis of patent data identified a number of cases related to topics relevant to the thesis worth further investigation. Relying on insights gained during the interviews carried out for validating purposes 23 three cases were selected. The selection process took into account (a) the relevance of each case to the research topic of the thesis, (b) accessibility to and availability of companies’ information, (c) broad focus (i.e. many cases and more aspects) vs. narrow focus (i.e. a few in-depth cases), and (d) time available for fieldwork research. Therefore, a compromise was reached between these different dimensions, and it was decided to carry out three in-depth cases at the expense of a larger set of cases. The cases selected are engine makers’ all-engine capabilities (Chapter 6), the fan subsystem (Chapter 6), and the aircraft engine control system (Chapter 7). Ad-hoc questionnaires 21

In this way interview data are employed to understand why and how the patterns observed in the quantitative analysis occurred (Yin, 1994). 22 A copy of the letter can be found in Appendix 5. 23 The questionnaire employed in these interviews also used open-ended questions in order to gather information and insights in relation to key issues.

54

were developed for each of the three cases. In addition, key informants were also asked to review, comment on, and check the drafts of the three case studies. The use of key informants to assess the drafts of the case study is another tactic suggested by Yin (1994) to establish solid construct validity 24. Their comments and suggestions were then discussed over the phone. For the case study of engine manufacturers’ all-engine capabilities, three face-to-face interviews and two telephone follow-up interviews were carried out with two engine manufacturers. Interviewees included the Head of Performance and the Chief Engineer from Fluid Systems. They were asked to give detailed information in relation to technological (e.g. background of scientist and engineers working in each group) and organisational (e.g. relationships with other firm’s units) issues related to firms’ allengine capabilities. The follow-up interviews were used to correct any discrepancies that emerged during the analysis of the evidence. The case study on the fan subsystem involved two face-to-face interviews, one with the Chief Fan Engineer of Rolls-Royce and one with a practitioner. The interviews lasted respectively two hours and half and 30 minutes. The aim of these interviews was to understand the key issues (including information and communication technologies) influencing firm’s make or buy decisions related to the fan subsystem. In the case of the aircraft engine control system, 10 (including phone and face-to-face) interviews were carried out with one engine manufacturer and four engine control system suppliers. Interviewees included the Chief Control System Engineer, the Head of the Control System Division of both engine maker and suppliers, and a Chief Software Engineer. They were asked questions about the extent of the division of labour between engine maker, and first- and second-tier suppliers. A further three phone interviews were carried out in order to clarify misunderstandings and provide information relevant to the case. Since this case study relies also on patent data, interviewees were asked to comment upon the findings of the patent analysis 25. Another set of 14 interviews was carried out in order to collect general information on the aircraft engine industry. In particular, other companies not directly involved in the production of engines, but part of the multi-actor industry environment, notably aircraft manufacturers (two) and airlines (one) were interviewed by phone 26. A representative of a certification agency was also interviewed by phone. This last interview led to additional interviews in one engine manufacturer to find out its standpoint on the certification process (4 face-to-face interviews). Information gathered during these interviews was used to draw the introductory picture of the industry (Chapter 4). Industry experts and academics were also consulted in order to obtain general information on the industry and share the findings of the analysis (four face-to-face and two phone interviews, lasting from 30 minutes to three hours). Some of them were extremely helpful for the construction of the component checklist and the ranking of the components included in it (Section 5).

24

See footnote 18. As Yin (1994) argues case study methodology is wrongly associated with qualitative empirical evidence. 26 As regards airlines, whenever possible pilots were interviewed during the fieldwork research. 25

55

6.3. Analysing case study evidence Interview data, companies’ internal documentation, and technical publications (also patent data in the case study of the engine control system) were the basis on which the case studies were built. The unit of analysis for each case study was the engine manufacturer. In particular, in each case study the engine manufacturers’ technological capabilities were examined in relation to one specific aspect (e.g. the engine control system). Case study research is always criticised because there is no received and codified method to develop the raw data into a study’s conclusions. In analysing the empirical evidence collected through interviews the study employed some of the tactics suggested by Yin (1994) to improve the quality of the research 27. Hand written notes were taken during the interviews in order to collect as much evidence as possible. These notes were transcribed soon after conducting the interviews. This enabled a preliminary scan of the interview data. The overall analysis used two tactics. (a) The first was a case by case interpretative analysis to examine the key issues of the single case (within case analysis, Eisenhardt, 1989). (b) The second was the pattern-matching tactic suggested by Yin (1994: 106), “Such a logic (Trochim, 1989) compares an empirically based pattern with a predicted one (or with several alternative predictions)”. The use of this tactic assumes the development of rival explanations involving mutually exclusive independent variables. As Yin argues (1994: 108), “the presence of certain independent variables (predicted by one explanation) precludes the presence of other independent variables (predicted by a rival explanation)”. The use of the pattern-matching tactic strengthens the internal validity of the research 28. Furthermore, since similar results were found, literal replication 29 has been achieved. According to Yin (1994), the replication logic underlying multiple case study design increases the external validity of the research 30. 6.4. Writing the case studies The case studies were written using a linear-analytic structure (Yin, 1994). The findings from the data collected with reference to the specific aspect analysed and the implications and conclusions from the findings are presented in sequence paying particular attention to the key analytical issues of the thesis (Chapters 6 and 7). 6.5. Drawbacks of interview data and case study method Although the thesis employed certain tactics to improve the quality of the study (as discussed in Sections 6.2 and 6.3), it is worth recalling the drawbacks to interview data 27

As discussed in Sections 2 and 6.2, the study has used some tactics also during data collection. The internal validity of the study (see footnote 18) refers to the “extent to which we can infer causal connections from a relationship between two variables” (Judd et al., 1991: 32). According to Yin (1994: 33), establishing internal validity means “establishing a causal relationship, whereby certain conditions are shown to lead to other conditions, as distinguished from spurious relationships”. 29 Replication is the logic underlying multiple-case study design. If the cases are selected to predict similar results, then the replication logic is literal (Yin, 1994). 30 The external validity of the study (see footnote 18) refers to the “domain to which a study’s findings can be generalised” (Yin, 1994: 33). The procedures designed to increase research’s external validity aim “to increase our ability to generalise the research results to the populations and settings of theoretical interest” (Judd et al., 1991: 35). 28

56

and the case study method. As is widely recognised, interviews do not provide the same quality or quantity of data. Therefore, data obtained from interviews should be carefully interpreted then. As mentioned in Section 6.2, three case studies were selected in relation to a number of dimensions. The cases were selected and not sampled. The cases were selected in the same way a laboratory investigator selects experiments (Yin, 1994). This selection process impinges on the generalisibility of the findings of the case studies. As Yin (1994: 31) suggests, the method of generalisation from case studies is ‘analytic generalisation’, “in which a previously developed theory is used as a template with which to compare the empirical results of the case study”. This method of generalisation differs from ‘statistical generalisation’; where data related to a sample form the basis for inferences about a population. 7. Conclusions and summary This chapter has discussed the research method of the thesis. The discussion included the research process and the research design, forms of data employed, database compilation methods, and data collection and analysis. The thesis employs two methods, comparative quantitative analysis (based on patent statistics) and case study. Furthermore, the thesis is based on a combination of two distinct types of data, namely qualitative and quantitative. These data are collected from multiple sources. The combination of two methods and the use of multiple data sources have several advantages. First, employing two methods allows methodological triangulation. Case studies are used to overcome some of the limitations of patent statistics. Second, using multiple data sources enables the development of converging lines of inquiry (Yin, 1994), a process of triangulation of data sources. That is, the same phenomenon is investigated with multiple lenses following a corroboratory mode and leading to more accurate conclusions. Table 3.2. Relationships among the three taxonomies Macro-technical fields (7) MATERIALS

MANUFACTURING

Technical fields (23) MATERIALS

Sub-technical fields (58) Metal powder composites Metal alloys & superalloys Composite materials (structures and stock materials, including ceramics) COATING AND Coating processes and CHEMICAL PROCESSES apparatus AND APPARATUS Chemical processes METAL WORKING AND Metal working and metal METAL TREATMENT deforming methods PROCESSES Bonding Metal treatment processes Powder metallurgical processes Casting processes (and moulds) 57

TESTING

PRODUCT INNER

Composite materials method of manufacturing METAL WORKING AND Metal working and metal METAL TREATMENT treatment equipment EQUIPMENT ELECTRICAL Electrical machinery MACHINERY ELECTROCHEMICAL Electrochemical apparatus MACHINERY and processes ELECTRONIC AND Electronic and optics OPTICS SYSTEMS FOR systems for manufacturing MANUFACTURING TESTING AND Testing and inspection INSPECTION apparatus Image analysis CONTROL SYSTEMS Control systems Optical components Thermal probes Ignition and starting systems Fuel systems Hydromechanical components Electronic components ROTOR ASSEMBLY Rotor assembly Rotor subassembly Cooled blades Blades Blade mountings STATOR ASSEMBLY Stator assembly Selectable adjustable vanes SHAFTS AND Shafts and bearings BEARINGS CASINGS Casings Shrouds SEALINGS Turbine clearance control arrangements Seals COMBUSTION Combustion chambers CHAMBERS Combustion chamber supports Combustion chamber liners Fuel injectors and fuel mixers Water and steam injectors COOLING SYSTEMS Fluid handling systems Heat exchangers Cooling systems for rotor assembly 58

PRODUCT OUTER

EXHAUST SYSTEMS LUBRICATION SYSTEMS CONTAINMENT STRUCTURES COUPLINGS AND JOINTS

Cooling systems for auxiliary components Exhaust systems Lubrication systems Containment structures Couplings and joints

Support and mountings Nacelles Air intakes Ice preventer Thrust reversers By-pass duct arrangement NEW ARCHITECTURES New architectures (including geared-fan, prop-fan, counter-rotaing fan, aft-fan, engine modularity, rotary pitch change mechanisms) Rotary pitch change mechanisms OTHER (including Other auxiliary power units, rockets, afterburners, and VTOL) AERONAUTICS

NEW ARCHITECTURES

OTHER

Table 3.3. Definition of the patent groups (source: author’s elaboration on interview data) Group Definition Main content describes how components are arranged in a system Architecture Sub-architecture Main content describes how components are arranged in a subsystem Main content describes functional behaviour of a system Functional Main content concerns physical description of an item or a Physical subsystem

59

CHAPTER 4: AN OVERVIEW OF THE AIRCRAFT ENGINE INDUSTRY AND TECHNOLOGY 1. Introduction The aim of this chapter is to provide an overview of the aircraft engine industry and technology. By illustrating the key characteristics of these main issues, the chapter provides a background for the discussion on the boundaries of engine makers’ conducted in the following chapters. The aircraft engine industry shares some of the characteristics of complex product systems industries (CoPS) as identified by Miller et al. (1995) and Hobday (1998). CoPS are high cost capital goods that differ from mass-produced goods in terms of dynamics of innovation process, firms’ strategies, and industry structure. Using the CoPS framework, this chapter sets the scene for the ensuing discussion on engine makers’ technological capabilities. The chapter builds on different types of data, namely interviews, technical literature, companies’ annual reports and publications, and industry data. Technical literature and articles from specialised journal (particularly Flight International, Interavia, and World Aerospace Technology) provide the basis for the discussion of the technological drivers and developments of the aircraft engine. Information was also collected via face-to-face and telephone interviews with engine makers, airframers, airlines, industry experts, and regulatory bodies. This information was used to discern and describe the role of the main actors of the industry in the innovation process. A draft of the chapter has been reviewed and commented upon by two key informants (two of the company engineers interviewed during the field work research). According to Yin (1994) this technique helps establish construct validity 31. The chapter brings to light the managerial challenges that engine manufacturers face. The aircraft engine is characterised by a large and increasing number of interacting components belonging to different technological fields. New engine development programmes involve a large number of actors (suppliers, regulatory bodies, airframers, and airlines) that require an effort of co-ordination from both an organisational and a technological viewpoint. In order to co-ordinate, manage, and integrate the roles of the actors involved in the industry, engine makers need to span their capabilities over a wide range of scientific and technological fields and they are required to develop specific organisational (e.g. project management) and relational (e.g. marketing) capabilities. The chapter also illustrates the heavily regulated character of the industry. This requires engine makers to have a clear understanding of the impact of the rules imposed by certification bodies on new engine designs and to develop appropriate testing techniques to comply with such rules. Engine makers must also have a clear understanding of customer requirements. In the civil aircraft engine industry, the customer is the airline placing the order. However, it is the airframer that passes the requirements to the engine maker. Therefore, engine makers must be able to translate such requirements into technical specifications that can be met from their technological capabilities.

31

See Chapter 3 for a definition of construct validity.

60

The chapter is organised as follows. Using the interpretative framework proposed by Miller et al. (1995) to study CoPS industries, Section 2 illustrates the structure of the innovation process and the role of the industry’s major actors. Section 3 presents a brief history of the aircraft engine and the characteristics and suitability of the major aircraft engine types. Section 4 offers a discussion of the major technological developments that characterise the aircraft engine. Section 5 concludes the chapter. 2. The structural context of the innovation process in the aircraft engine industry This section illustrates the main characteristics of the structural context of the innovation process in the aircraft engine industry. The discussion opens with a summary examination of the main characteristics of CoPS industries as identified by Miller et al. (1995) and Hobday (1998). Based on this work, an interpretative scheme is proposed to identify the roles of the main actors involved in the innovation process and to highlight the factors influencing sources, pace and direction of technical change in the aircraft engine industry 32. 2.1. Mass-manufactured products vs. complex product systems industries CoPS have been defined as “high cost, engineering-intensive products, sub-systems, or constructs [i.e. capital goods] supplied by a unit of production” (Hobday, 1998: 690). Hobday has identified a number of dimensions to compare and contrast mass-produced products and CoPS 33. Accordingly, CoPS differ from simpler, mass-produced products in terms of product and production characteristics, dynamics of the innovation process, competitive strategies, managerial constraints, industrial co-ordination, and market characteristics. Miller et al. (1995) relying on a study of the flight simulator industry have shown that the industry life cycle model as proposed by Abernathy and Utterback (1975) is unable to explain the determinants of the life cycle in CoPS industries. According to the Abernathy and Utterback model, industries evolve through two phases characterised by different rates of product and process innovations. In the initial or fluid phase, the rate of product innovation is high, stimulated by a large number of small competing firms. Product innovations are performance maximising and processes are uncoordinated and customised. The fluid phase is terminated by the selection by the market of a particular product configuration, labelled dominant design. The emergence of a dominant design has two main effects. First, it ushers in an industrial shakeout as the industry becomes dominated by a small number of large firms. Second, due to the standardisation of product design, the rate of product innovation slows down, whereas the pace of process innovations increases to exploit economies of scales. By contrast, Miller et al. (1995) show that the flight simulator industry tends to remain in the fluid stage. The emergence of a dominant design does not pave the way to an industrial shake out. Due to high barriers to entry and despite radical technological shifts, the flight simulator

32

The interpretative scheme is based on Miller et al. (1995) and Hobday (1998) since the framework they propose enable a better understanding of the roles of and the relationships among the different actors involved in the industry under study. In comparison, the frameworks developed by Carlsson and Jacoboson (1994) and Porter (1990) to analyse the dynamics of industrial sectors make a more specific reference to the local or national dimension, which is not within the scope of this study. 33 For a comprehensive discussion of the main features of CoPS industries see Hobday (1998). For a description of a CoPS product life cycle see Davies (1997).

61

industry is characterised by high stability of flight simulator suppliers (systems integrators) and high turbulence in the supply chain (component suppliers). 2.2. The aircraft engine industry as a CoPS industry The aircraft engine industry shares some of the characteristics of CoPS industries as identified by Miller et al. (1995) and Hobday (1998). An aircraft engine is a high cost capital good composed of many interacting and often customised elements that belong to different technological fields. The number of components varies according to the size and, therefore, thrust of the engine. The powerful engine powering such aeroplanes as the Boeing 747 and the Airbus A340 may encompass up to 40,000 components (Interview, 1999). In addition, although the same basic engine powers these two aeroplanes, certain engine parts need to be customised to each application. The software governing the engine control unit is a case in point since it has to be fine-tuned with the avionics system of the aeroplane. Aircraft engines are batch produced. The design, development, and production of a new aircraft engine involve several firms and the degree of user involvement is very high. Airframers, airlines, regulatory bodies, specialised suppliers, and engine makers closely collaborate during new engine development programmes. The development costs of a new engine are extremely high, notably between US $ 500 million and US $ 2 billion 34. These figures highlight that any failure of a programme may badly affect the financial situation of an engine maker and/or push it towards receivership as in the case of RollsRoyce during the development of the RB211 engine in the 1970s. Such huge financial outlay has led to financial collaborative agreements between engine makers and suppliers for the development of new engines 35. With regard to the dynamics of the innovation process in the aircraft engine industry, the Abernathy and Utterback (1975) life cycle model does not seem to capture its salient characteristics. Although the turbofan engine configuration can be regarded as the industry dominant design, its emergence does not seem to have slowed down the rate of product innovation. In the aircraft engine industry, product and process innovations are closely intertwined and product innovations are always performance-maximising. This is supported by the findings from the patent analysis carried out in Chapter 5. 2.3. The structure of the innovation process in the aircraft engine industry Miller et al. (1995) argue that the appropriate unit of analysis for studying the process of innovation in CoPS industries is the network of actors involved in the process as well as the single supplier companies. They propose that the innovation parameters of CoPS industries can be described in terms of a meso-system composed of three main groups, notably the innovation superstructure, the systems integrators (in this case the engine makers), and the innovation infrastructure. Drawing on such framework, this section illustrates the structure of the innovation process in the aircraft engine industry. The scheme depicted in Figure 4.1 is organised as follows. The engine manufacturers make up the core of the meso-system. They co-ordinate the functioning of the innovation process by organising the roles of the different actors involved. Suppliers are part of the innovation infrastructure. They provide materials, components, 34

Interviewees underlined that new engine programmes break even after 15-20 years (Interviews, 1997, 1998, 1999). 35 See Section 2.5.

62

subsystems, machine tools, and software to the engine makers. Suppliers are increasingly involved in new engine programmes both financially as risk and revenue sharing partners, and technologically as they take on larger chunks of engine design and manufacturing tasks. Government-funded laboratories and universities are also part of the innovation infrastructure. They provide the research infrastructure as well as the technologies and training. Engine makers and the innovation infrastructure are the supply side of the industry. They can be regarded as the sources of technical change. The innovation superstructure represents the ‘market’ for aircraft engines. Airlines, airframers, certification agencies, and professional bodies that comprise it, heavily influence the pace and direction of technical change. The airlines are the buyers of the engines. Airframers define engine requirements in relation to aircraft characteristics and pass them on to the engine manufacturers. Certification agencies and professional bodies form the remaining part of the innovation superstructure. The former set engine certification requirements, whereas the latter act in the context of information exchange between the different parties involved in the process. A substantial role in directing technical change is also played by the high level of regulations imposed by national governments and international organisations. It is worth analysing the role of each group and how they interact in the innovation process in order to have a better understanding of the functioning of the process itself. Each group therefore will be examined in turn. Figure 4.1. The aircraft engine industry meso-system

Airframers

Airlines

Certification agencies Professional bodies

Innovation superstructure Engine makers Innovation infrastructure

Government-funded laboratories Universities

Risk and revenue sharing partners

Suppliers

63

2.4. Engine makers As reported in Figure 4.1 the engine makers represent the core of the innovation mesosystem of the industry. Engine makers design, develop, and manufacture engines according to the airlines’, airframers’, and regulator’s requirements. They co-ordinate the activities of a large number of specialised suppliers, integrate components and add value through their systems integration capabilities.

At the systems integration level, the industry is highly concentrated. The large turbofan market (over 35,000lb) is dominated by the so-called Big Three or Primes, notably General Electric Aircraft Engines, Pratt & Whitney, and Rolls-Royce. In the mediumand small-sized engines market (below 35,000lb), the main actors are AlliedSignal Engines, Rolls-Royce Allison, General Electric Aircraft Engines, Pratt & Whitney Canada, Williams International, and two international joint-ventures, CFM International, and International Aero Engines. CFM International was set up by General Electric and Snecma, whereas International Aero Engine (IAE) is composed of Pratt & Whitney, Rolls-Royce, Motoren und Turbinen Union (MTU), Fiat Avio, and Japanese Aero Engine (in turn a consortium of Japanese aerospace companies). In the turboshaft/turboprop market, in addition to the previously mentioned engine makers, it is worth mentioning the French company Turbomeca. In the aircraft engine industry, East Asian firms play a relatively minor role. They are in fact suppliers or at the most joint venture partners, such as Mitsubishi Heavy Industries and Kawasaki Heavy Industries from Japan and Samsung Aerospace from Korea (Nakamoto, 1997). In the last ten years a number of mergers and acquisitions (M&A) has further concentrated the industry. AlliedSignal Engines acquired Textron Lycoming and Garret Engines, and Allison Engines was acquired by Rolls-Royce. M&A have also taken place at the supplier level within the larger M&A movement of the entire aerospace sector. 2.5. The innovation infrastructure Risk and revenue sharing partners and specialised suppliers The industry is characterised by an ever-increasing number of international collaborative agreements for the design, development and manufacture of new engines. This practice has been borrowed from the military side of the industry where collaborations amongst companies from different countries have been taking place since the early 1970s. The reasons why the development of new engines is shared amongst different companies lie in the increasing development cost and related risk of failure of the programme. Competition in the large turbofan market is in fact severe and pricebased. Industry trade journals report that engine list prices are heavily conceded by engine makers in order that they can secure a foothold in the spares market business considered to be the ‘gold mine’ of the industry.

Thus, new engines are developed using a new form of contractual relationship, labelled a risk and revenue sharing partnership (RRSP). Accordingly, suppliers, typically firsttier ones, are invited to join the engine programme early on and to buy a stake in it (usually one or more components or an entire subsystem) in order to share the risks and future revenues (if any) of the programme. The shares held by suppliers have been increasing over time. On the one hand, engine makers want to ‘split’ risks and revenues across several suppliers to reduce their own stakes and to gain customers (i.e. airlines) via the involvement of suppliers of the same nationality as the customers. According to 64

some industry experts, airlines (usually state-funded) are more likely to place engine orders when national component suppliers have been involved in the engine programme. On the other hand, component suppliers, especially from developing countries, push to get bigger engine programme shares in order to get experience with more engine parts. The industry therefore is characterised by a three-tier structure made up of engine makers, RRSPs, and suppliers. The world’s largest RRSPs are MTU, Snecma, and Fiat Avio. It is worth noting that the boundary between the RRSP and supplier categories is fuzzy. A supplier can be invited to buy a stake in an engine programme and become a RRSP, and at the same time be a mere supplier in another programme. According to the component suppliers interviewed, RRSP is a risky business as it links the revenues of the supplier to the success of the programme. The fierce competition between engine makers has drastically reduced the prices of the engines. In this way RRSPs end up relying on spare part sales. However, in a supplier situation, suppliers’ revenues are linked to the extent of the supply regardless of the success of the engine programme (Interviews, 1996, 1997,1998). The boundary between engine makers and RRSPs is very well defined, however. Although, engine makers may decide to act as RRSPs in a specific engine programme, where they are in charge of the design and development of one or more engine parts, RRSPs do not and cannot take on integration responsibilities at will. They lack the required systems integration capability. These capabilities are built over a long period of time and require large investment in multiple technical fields and across several knowledge domains, such as concept and detailed design, and development. The boundary between systems integrators and RRSPs therefore is permeable but only oneway. Government-funded laboratories: the role of national governments Mowery and Rosenberg (1982) argue that government policy has influenced both the demand and supply of the US commercial aircraft industry. On the supply side, the industry has received government support via research on commercial application technologies undertaken at national laboratories such as NASA, direct funding of firm internal research programmes, and spill-over from the military side. The demand side has been influenced via the imposition of heavy regulations for safety and subsequently environmental concerns 36.

The support of national governments for the development of new engine technologies via direct and indirect funding is still strong. Government in both developed and developing countries fund research as well as development programmes. The former to strengthen the technological leadership of their national champions, the latter to improve their technological capabilities in such a value-added industry 37. As regards 36

See Section 2.6 for an analysis of the national governments’ influence on the demand side. As confirmed by an industry expert “the reasons government [in developing countries] will do that [i.e. ask engine makers to involve national companies in the development of new engines] are very many, but essentially two. The first is the aero gas turbines are closely related to defence and therefore a country has to have the ambition in the medium or long term to have a defence capability equal to the best. The other issue is the view, totally correct view, that if you invest in very high value added technologies like the aero gas turbine, you get a spreading of the capabilities which are research, design and even management, because we are managing a very complex technology. (…) So many developing countries 37

65

Western countries, there are several differences between the organisation of supports for technical progress between Europe and US. In the US, the support of Federal Government still continues via direct and indirect initiatives. The Glenn and Langley Research Centres at NASA and the Wright Patterson Laboratory of the Air Force are considered centres of excellence for civil and military engine-related technologies respectively. Cuts in the defence budget have to an extent reduced the role of military-related research programmes used to have in the development of new technologies. The focus is now on the development of dual-use technologies. The latest major programme, the Integrated High Performance Turbine Engine Technology (IHPTET), for instance, “addresses critical defence technology objectives” but also “develops dual-use technologies” (IHPTET, 1999: 1). IHPTET is a US $ 4.5 billion programme funded by the US Government (65%) and industry (35%). It involves the US Army, Navy, Air Force, Defense Advanced Research Project Agency (DARPA), and NASA as well General Electric Aircraft Engines, Pratt & Whitney, Allison Engines, AlliedSignal Engines, Williams International, and Teledyne Ryan Aeronautical. IHPTET is considered a major programme due to its time span (25 years), the areas it covers, and the magnitude of its goals (Sweetman, 1989). It covers the major jet engine component technologies, such as turbine and combustors, as well as the pervasive technologies, such as materials and processes. The underlying objective is not to develop engines, but to develop technologies that can be incorporated into demonstration and validation programmes or full scale development programmes (Sweetman, 1989). The achievement (in terms of completion dates) of each objective per engine technology have been scheduled into three stages, notably in 1991, 1997, and 2003. The support of European Governments for the technological developments in the aircraft engine industry is less visible than in US. Europe does not have organisations that can match the shear size of the NASA Research Centres and the Wright Patterson Laboratory. Nonetheless, it is well known that European Governments support national engine companies via direct funding of research programmes and new engine development programmes. Likewise, the European Union has funded and continues to fund important research programmes where industry and universities collaborate within the Framework Programmes. Special initiatives cover the specific subprogramme on the aircraft industry labelled ‘Aeronautics’ within the Brite – Euram Programme, the work of the ‘Aeronautics Task Force’ to co-ordinate the various aeronautics research programmes, and the broad and heavily funded subprogram on ‘Aeronautics’ within the recently launched 5th Framework Programme. The role of Universities Universities are also part of the innovation infrastructure. Their role is not confined to being mere providers of general knowledge, however. Universities are heavily involved in research projects with a much larger scope. In fact, the scope of academic research projects ranges from design of components, verification of company’s design, and development and verification of design codes, through the population of experimental databases and exploration of component physical behaviour, to the training of researchers and technical process improvements (Interviews, 1997, 1998). Companies will be opened to say why don’t you invest in my organisation to develop gas turbine technologies for this country, because there will be spin off effects for many other people in the country.” (Interview, 1996).

66

interviewed fund studentships on specific topics and contemplate using in-house professors for trouble-shooting and problem-solving activities as well as for longer term research guidelines (Interview, 1997). The importance of the role played by academic units is demonstrated by the fact that the companies interviewed consider the technological areas investigated in collaboration with them to be highly confidential 38. Table 4.1. RRSP research projects in collaboration with Universities (source: author’s elaboration on interview data) Development of method for analysis and optimisation of engine system Through-flow analysis of transonic blade rows and compressors Aerothermal vibration in cascades Flow with high temperature in rotating part Advanced control concept for fuel Method for analysis and design of safety system for FADEC Efficient development of design parameters Frictional damping of blade vibration Titanium and nickel-based alloys Engine life analysis Joining of nickel-based super-alloys Joining of titanium alluminade, nickel-based alloys and steel Product simulation of a jet engine It is worth noting that academic collaboration is a phenomenon that interests engine makers as well as first and second tier suppliers (Interviews, 1996-1999). Table 4.1 reports the research projects undertaken during 1997 by one supplier with a number of national universities. This supplier is one of the major providers of engine casings in the civil business. It also manufactures military engines under licence agreements. 2.6. The innovation superstructure The regulatory network imposed by national governments The effects of regulation on aircraft engine demand are both direct and indirect. Direct effects concerns the issuing of rules that regulate engine noise and emissions and the certification of new engines. These effects are analysed in the following sections through a discussion of the role of certification bodies. The indirect effects are discussed below.

The indirect effects result from the heavy regulatory network imposed by national governments and international bilateral and multilateral agreements to regulate the exploitation of the air space over each national state by air carriers. These rules influence air carriers’ competitiveness in terms of price and route structuring. According to such rules, air carriers are granted licences to exploit the air space 39. The 38

In fact the request of the author for a list of the company research projects performed in collaboration with universities was very often denied. 39 The Chicago Convention in 1944 has regulated most of the issues related to air traffic control. The socalled ‘freedoms of air’ have been instead left to international agreements. As reported by Rolls-Royce (1997), there are seven ‘freedoms of air’ “1st Overfly foreign territory, for example en-route from one country to another. 2nd Make a non-traffic stop in another country, for example to refuel. 3rd Carry passengers from the home country to another country. 4th Carry passengers to the home country from another country. 5th Carry passengers between two countries by an airline of a third country, with the route beginning or ending in the home country. 6th Carry passenger between two foreign countries by

67

International Air Transport Association (IATA) is the regulatory body with regard to fares. Bilateral air carriers’ agreements (for instance code sharing) overlap this heavy regulatory network. This dense regulatory structure has shrunk dramatically in the last 20 years. In the US, the tight control of the Civil Aeronautics Boards (CAB) that since 1938 had controlled pricing policies and entry and exit from air transportation, came to an end in 1978 (Mowery and Rosenberg, 1982). The deregulation had a profound impact on the US air carriers’ route structures. Air carriers have in fact adopted a hub-and-spoke strategy. Airlines choose an airport as hub where they concentrate passengers coming from other airports (spokes) to redirect them to yet others (spokes). In Europe, the European Union’s Third Package of aviation liberalisation became effective in April 1997 giving airlines the right to cabotage in another country 40 (RollsRoyce, 1997). However, according to industry sources this last stage in European liberalisation was only a symbolic move for the European Governments since national carriers still enjoy big cost advantages in their respective domestic market (Boeing, 1999; Rolls-Royce, 1997). At the end of 1998, a new Japan-US bilateral agreement was signed. This represents a significant step towards the liberalisation of the transpacific market (Boeing, 1999). The certification agencies and the professional bodies Certification agencies are part of the innovation superstructure. Engines have to comply with rules defined by regulatory agencies, such as the Federal Aviation Administration (FAA) in the US and the Joint Aviation Authorities (JAA) in Europe. There is a close, dialectic, and ongoing dialogue between engine makers and regulatory agencies during the design and the development of new engines (Interview, 1998). Safety requirements have an important and constraining effect on the application of new technologies. The introduction of new technologies is in fact always extensively discussed between engine makers and regulators. New technologies often attract specific new regulations and stringent testing procedures. In this way, the severe testing procedures imposed via new rules act as targeting devices for innovation (Miller et al., 1995). The first phase in the dialogue between engine maker and regulator ends with the formal certification of the engine after long and arduous testing carried out by the engine maker and overseen by the regulator. According to Miller et al. (1995) this focusing role played by the regulator for the introduction of the innovation is a salient characteristic of many CoPS industries as compared to mass-productions where innovation is predominantly mediated by the market.

The fact that the introduction of new engine parts based on innovative technologies is discussed between engine manufacturers and the certification authority does not lessen the importance of feedback as mediated by the market. Information related to engine behaviour gathered by engine operators (i.e. airlines) and maintenance engineers is deemed extremely valuable by the engine makers during the innovation process. One of our interviewees emphasised that the acquisition of maintenance companies by engine makers in the last five years has realised as a by-product an important source of stopping or connecting in the home country. 7th Carry passengers between two foreign countries, without extending the route to the home country. 8th (cabotage) The right to carry traffic wholly within a foreign country”. 40 See previous footnote for a definition of cabotage.

68

information about customer requirements 41 (Interview, 1997). The acquisition of information related to engine behaviour is now facilitated by digital engine control technologies that have replaced the hydromechanical control system. Digital engine control units, also labelled FADEC (i.e. full authority digital engine control unit) are able in fact to monitor and store engine performance data throughout the life of the engine 42. Two other points on certification are worth mentioning. The first is related to its scope. Certification procedures cover design, operations, maintenance, and licensing standards. The second is related to the recent move towards the global harmonisation of certification procedures. The lack of common worldwide certification requirements requires engine makers and airframers to certify and validate their products according to the standards defined by the major national regulatory bodies. Meeting these different standards inevitably results in additional costs for the engine manufacturers. In Europe, a body representing the civil aviation authorities of 29 European countries was created in 1970 (Interview, 1999). This body, the JAA, ensures common high levels of safety standards within the member countries. The JAA has also set up a Certification Group that is responsible for certification of new aircraft and engines. After the completion of a JAA certification programme, a Type Certificate can be issued by all member countries. The JAA and the FAA are currently working to harmonise the aircraft and engine certification procedures. It is worth noting that the discussion of amendments to current rules occurs within study groups and committees, which include representatives of national or international industry associations such as the Society of British Aerospace Companies (SBAC), the European Association of Aerospace Companies (AECMA), and the Aerospace Industries of America (AIA). The airlines Airlines are another group part of the innovation superstructure of the aircraft engine industry. As they are the final customers for the engines, they are in a position of a considerable power to influence the performance characteristics of an engine. Airlines demand reliable engines with low operating costs for a variety of reasons but ultimately to boost their profit margins 43. In addition, they demand quieter and less polluting engines to comply with the rules set by airports and regulatory agencies.

Airlines’ strategies heavily influence the rate and direction of technological change in the aircraft engine industry. An airline’s demand for aircraft in terms of size and range depends on its route structures. Route structure in turn depends on firms’ individual strategies but it is also influenced by the rules imposed by national governments. As mentioned earlier, since deregulation in the US, airlines have adopted a hub-and-spoke strategy, which requires a change in the composition of airline fleets in terms of long41

It is worth noting, however, that the main reason why engine makers have entered the maintenance business is to be found in the increasing profit margins characterising this business as opposed to the nearly negative cash flow deriving from the sale of aero engines. By strengthening their position in the maintenance business, engine makers claim to be able to provide a total engine service, from physical product to engine life maintenance. In some cases the acquisition of engine maintenance companies by engine makers was part of the financial deal struck with airlines. 42 See Chapter 7 for an in-depth case study on the engine control system. 43 An engine’s fuel consumption and reliability influence around 60% of the total airline direct operating cost (DOC). The remaining 40% is related to airframe costs (Todd, 1992).

69

and medium-range aircraft. Long-range or medium- and short-range aircraft and twin or quad require the optimisation of different engine design parameters such as specific fuel consumption and noise. The preference of airlines for particular engine brands is due to historical and regional factors. Historically, US air carriers have preferred US engines, whereas, for instance, British Airways has been a loyal customer of Rolls-Royce. However, complicated financial deals, privatisation of national flag carriers, and offsetting practices have caused profound changes in the relationships between airlines and engine makers. Airlines who remain loyal to a particular engine manufacturer still exist mainly due to the heavy investments in engine support infrastructure but also to consolidated pricing policies. In the last five years, major airlines have teamed up to extend their market coverage via global alliances 44. In forming these alliances, airlines have combined their marketing efforts in order to provide capacity more efficiently to passengers. According to industry sources, however, the impact of these alliances on the structure of the competitive environment and on aircraft and engine purchases has been negligible so far (Boeing, 1999; Rolls-Royce, 1999) In relation to airline involvement in new engine development programmes, it should be noted that in the last decade the airlines have been involved far more heavily. For instance, during the development of the Boeing 777 and the GE90, engine teams involving airframers, engine makers, and airlines were set up. These teams were labelled ‘Working Together Teams’, WTT (Todd, 1992). Boeing has set up WTT with the other engine suppliers also and subsequently applied the same principles to other aircraft programmes. The airframers The airframers are also part of the innovation superstructure. Though they are no longer the final customers of the engine manufacturers, airframers play a central and active role in future (and futuristic) engine configurations as well as during new engine programmes within an established engine configuration. Airframe/engine integration is of paramount importance for efficient and safe air transport. Radical changes concerning the engine design configuration involving new airframe/engine installation solutions require the joint and close co-ordination effort of both airframers and engine makers 45. 44

The world’s 30 largest airlines have formed four major alliances, notably oneworld (American Airlines, British Airways, Canadian, Cathay Pacific, and Qantas), Star Alliance (Air Canada, SAS, Lufthansa, United, Varig, and Thai), KLM & Northwest (including also Alitalia and Continental), and Atlantic Excellence (Swissair, Delta, Austrian, and Sabena). These four major groupings accounted for around 60% of the world’s passenger traffic in 1997 (Rolls-Royce, 1999). 45 Radically new airframes involving a step change in the airframe/engine integration configurations require the joint effort of expertise in airframe and propulsion related technologies. A case in point is represented by a 3-year NASA technology development program whose aim is to assess the technical and commercial viability of an advanced, unconventional aircraft configuration, namely the Blended-WingBody (BWB). This airframe configuration is a flying wing with embedded engines. It is a very large subsonic transport with a design payload of 800 passengers, a 7000-n.mi. range, and a cruise Mach number of 0.85. As reported by NASA (1999) “Because the BWB configuration is such an extremely integrated design, a multidisciplinary optimisation process will be utilised extensively to address technical issues in configuration design, aerodynamics, structures, propulsion, and flight mechanics. An initial evaluation of this configuration indicates significant cost and performance benefits over conventional configurations: a 56-percent increase in lift-drag ratio, a 20-percent decrease in fuel burn, and a 10percent decrease in the operating-empty weight. The research team consists of McDonnell Douglas

70

Even within established engine design configurations, such as the turbofan, airframers heavily influence aircraft engine characteristics. The choice of engine cycle, size, and thrust is in fact weighted by its intended application. Similarly, the main engine parameters, such as fuel consumption, noise, weight, cost, and emissions are of primary importance for the aircraft builder. Airframers’ involvement during new engine programme During new engine programmes the airframers’ involvement is highly interactive. The dialogue between airframers and engine manufacturers starts when a potential business opportunity arises (a new or derivative engine for a new aircraft or a stretched version of an existing aircraft). If the discussion is fruitful and the business opportunity seems to have some grounds, a technical and/or business agreement is signed by both parties. This agreement is called a memorandum of understanding (MoU). It contains detailed engine specifications such as thrust, fuel consumption, noise, weight, stability, and vibration. Relying on its internal technological capabilities, the aircraft manufacturer performs a detailed technical audit to assess engine technological characteristics (thermodynamic cycle, component efficiency, and installation losses). The technical audit also covers the engine’s overall structural/mechanical design in terms of installation (accessibility, maintainability), materials used, component durability. The aircraft builder also evaluates the engine test programme. The technical audit is the beginning of the ongoing dialogue between the airframer and the engine manufacturer. The actual involvement of the airframer goes through engine development, flight tests, and certification of the airframe/engine combination (Nordstrom et al., 1980).

Airframers have acquired substantial in-house expertise in engine technology to cope with its increasing complexity (Mowery and Rosenberg, 1982). Functional and aerodynamic relationships between aircraft and engine render their integration critical task 46. According to Nordstrom et al. (1980), Boeing has strengthened its in-house engine technology expertise following the severe difficulties encountered in the installation of the all-new high by-pass engines in the all-new wide body 747 aircraft in

Aerospace, Stanford University, the University of Southern California, the University of Florida, ClarkAtlanta University, NASA Glenn Research Centre, and NASA Langley Research Centre the Technology Study funded by NASA is to assess.” 46 The aircraft/engine interface is a major area of concern for airframers and engine makers. There are around 300 mechanical interfaces between engine and airframe that are critical to the operation of the aircraft (Todd, 1992). These interfaces encompass air, hydraulics, electrics, fuel, electronic and mechanical controls, health monitoring, fire sensing and protection (Interview, 1998). The airframe/engine integration necessitates close collaboration throughout the design, development and assembly of the aircraft/engine combination. At the airframer level, it is co-ordinated by the Propulsion Department. A number of specialist departments of the airframer are involved in specific issues of the engineering integration of the engine. These are Structures, Aerodynamics, Nacelle Design, Performance, Noise, and Systems (e.g. avionics, electronics and hydraulics). In a large aircraft builder, the aircraft/engine integration can involve up to 300 people (Interview, 1998). The aerodynamic integration of aircraft-propulsion system combination is another important activity where the airframer and engine maker roles are closely co-ordinated. The engine has to be integrated with the nacelle and the nacelle with the aircraft. The location of the propulsion system combination is of paramount importance for aerodynamic reasons. In long range aircraft the engine underwing configuration has become the dominant location. However, the trend towards higher by-pass ratio and ensuing larger fan diameter may render the underwing configuration unsuitable due to the penalty it would entail on the landing gear length. In short-range aircraft the alternative mounting from the rear fuselage is preferred (Howe, 1992).

71

the early 1970s. Similarly, the shift from sole sourcing to dual or even triple sourcing has required that aircraft manufacturers become competent transactionists. It is worth noting that the use of information and communication technology-based tools has enormously improved the management of such a complex interface. 3D CAD design systems, such as CATIA, allow airframers, engine makers, and suppliers to electronically define their products and make changes to the design on the screen. This has substantially reduced the need for costly physical mock-up. Computer modelling also helps to design maintainability, which represents a critical factor for airline operation (Todd, 1992). 2.7. Concluding remarks Using the framework proposed by Miller et al. (1995) to study CoPS industries, this section has illustrated the structural context of the innovation process in the aircraft engine industry. The analysis has shed light on the role of each actor in the industry and how they influence source, rate and direction of technical change. The involvement of a large number of actors in the new engine development process requires an intense organisational and technological effort with regard to engine makers. Engine makers need to develop organisational capabilities, such as project management, and relational capabilities, such as marketing, in order to manage and integrate the role of each actor. Engine makers are also required to co-ordinate the different actors from a technological viewpoint in order to synchronise the technological developments undertaken by suppliers and make sure that they comply with regulations while meeting the customer requirements.

The discussion has paid particular attention to the heavily regulated character of the industry. Engines have to comply with ever-tighter regulations regarding noise, emissions, and certification. Engine manufacturers must have a clear understanding of these rules and their impact on engine design parameters. As explained in the next section, certification procedures require engine makers to master testing technologies. Engine manufacturers must also have a clear understanding of the customer requirements. The aircraft engine is a subsystem of a larger system, namely the aircraft. Engine manufacturers need to understand how aircraft design parameters affect engines. Aircraft design parameters (e.g. size) are in turn influenced by the strategy of the airline. As detailed in the next section, long-range and short-range aircraft pose different requirements for engine manufacturers. The following sections analyse in depth aircraft engine technology. Section 3 presents a brief history of the aero gas turbine and describes the characteristics of the major types of aircraft engine. Section 4 provides an analysis of the engine product characteristics and illustrates the major technological developments of the aero gas turbine. 3. Introduction to the aircraft engine 3.1. A brief history of the gas turbine The underlying principles of the gas turbine engine are described in the British Patent number 1833 that was granted to John Barber in 1791. The patent title is “A method for rising inflammable air for the purpose of producing motion and facilitating metallurgical operations” (as reported in Singh, 1996). Numerous attempts to develop an engine prototype using those principles took place at the beginning of the 20th century about 150 years after John Barber’s patent. Elling from Norway, Stolze and 72

Holzwarth from Germany, the French duo Armenguard and Lemale, Moss from US, and Lorin from France were amongst the pioneers of gas turbine technology (Singh, 1996). The internationally-recognised inventor of the jet engine, a variant of the gas turbine, is the British Sir Frank Whittle. He took out a British Patent in 1930 and ran the first jet engine in 1937 (Singh, 1996). The Whittle engine completed its first flight in 1941 (Rolls-Royce, 1986). Whittle’s great breakthrough can be better appreciated by noting that his W2/700 engine was the basis for the first engines launched by the so-called Big Three, namely General Electric Aircraft Engines, Pratt & Whitney, and Rolls-Royce. In fact, the W2/700 formed the basis of the Rolls-Royce Welland, Derwent, Tay, and Nene. In turn, Pratt & Whitney used the Rolls-Royce Nene engine for their J42. General Electric Aircraft Engines used the Whittle engine for their 1-A (Singh, 1996). 3.2. Some hints on the basic principles of the gas turbine The gas turbine is a heat engine that works by accelerating a working fluid to provide thrust. To accelerate the working fluid, first the pressure is increased, then heat is added, and finally converted to kinetic energy (Rolls-Royce, 1986).

The working cycle of the gas turbine is similar to that of a piston engine. Both the gas turbine working cycle (also called the Brayton cycle) and the piston engine cycle (the Otto cycle) show four phases, notably induction, compression, combustion, and exhaust. However, there are two main differences between the Brayton cycle and the Otto cycle. First, in the Otto cycle combustion occurs at constant volume, whereas in the gas turbine it occurs at constant pressure. Second, the four phases are intermittent in the piston engine, whereas they are continuous in the gas turbine. This latter difference enables the gas turbine to burn more fuel in a given time and therefore provide more power for a given size of engine (Rolls-Royce, 1986). In technical terms, the gas turbine has a higher power-to-weight ratio than the piston engine. The simplest gas turbine is the turbojet engine. It consists of a compressor to compress the working fluid, a combustion chamber to burn fuel, and a turbine to drive the compressor. Once the power needed for compression has been extracted, the hot gas stream coming from the turbine is expelled to the atmosphere via an exhaust nozzle to create propulsive thrust (Gunston, 1995). 3.3. Engine performance and efficiency Early turbojets were characterised by very low thrust and efficiency, “ [they] looked too light and flimsy even to contain their noise, let alone the power they generated, but they burned fuel at a daunting rate” (Gunston, 1995: 10). Advances in the technologies underlying the gas turbine have substantially increased engine thrust and improved its efficiency. Today’s turbofan engines are rated up to around 100,000lb showing an overall efficiency of about 35% (Interviews, 1997, 1998, 1999).

There are different kinds of efficiency. Following Gunston (1995: 10-14), thermal efficiency is defined as “the useful power generated divided by the rate at which chemical energy is produced by burning fuel”; propulsive efficiency is “the percentage of the power produced by the engine that is actually put to use in moving the aircraft”; cycle efficiency is “the ratio of useful work obtained divided by the useful work obtained from an engine with an ideal working cycle”. Overall efficiency is the product 73

of thermal efficiency and propulsive efficiency; polytropic efficiency is the component efficiency 47 (Singh, 1996). Specific fuel consumption is the ratio of fuel consumption to thrust and is defined in terms of pounds of fuel per hour per pound of net thrust. It is worth noting that increases in cycle efficiency stem from improvements in overall pressure ratio and turbine entry temperature 48. These improvements demand large investments in high strength and high temperature materials and cooling technologies. Increases in propulsive efficiency are mainly the result of different engine configurations. In particular, a dramatic improvement in propulsive efficiency occurred in the 1970s after the introduction of high by-pass ratio turbofan engines (Singh, 1996). Therefore the quest for higher power and higher efficiency has not been limited to the introduction of new technologies within the same engine configuration. Indeed, higher engine efficiency (and particularly propulsive efficiency) has been achieved through the introduction of new engine configurations. The following sections illustrate in detail the operating characteristics of turbojet, turboprop, and turbofan engines and discuss their suitability for particular thrust applications. 3.4. Turbojet and turboprop: technical characteristics and ‘selected’ applications Jet engines come in different design configurations, notably turboprop and turboshaft, turbojet, and turbofan. These different design configurations give them peculiar technical features and performances making them more or less appropriate according to the application. Although all jet engines use the same operational principle (i.e. they accelerate a flow of working fluid to provide thrust) turboprop, turbofan, and turbojet make different ‘use’ of the working fluid (Gunston, 1995; Rolls-Royce, 1986). As already mentioned, in a turbojet the hot gas stream coming from the high-pressure turbine is expelled to the atmosphere via an exhaust nozzle to create propulsive thrust. In turboprops and turboshafts the hot gas energy is converted via internal shafts into mechanical power to drive the propeller. As a consequence only a small amount of hot energy is exhausted.

These different design configurations show different propulsive efficiency according to the aircraft speed. The propulsive efficiency of the turboprop is approximately 80% up to an aircraft speed of 400mph (Gunston, 1995; Rolls-Royce, 1986). At higher aircraft speeds the turboprop’s blade tips become supersonic causing the propulsive efficiency to fall (Figure 4.2). The case is reversed in the turbojet. At relatively low aircraft speeds the turbojet’s propulsive efficiency is approximately 45% (Rolls-Royce, 1986). This is due to its high jet velocity 49. At higher aircraft speeds the turbojet achieves 47

For a detailed analysis of an example of improvements in component efficiency see the case of the fan discussed in Chapter 6. 48 The overall pressure ratio “is the ratio of the outlet pressure from the compression process to the inlet pressure to the compression process”. The turbine entry temperature “is the temperature of the gas at entry to the first turbine rotor” (Jane’s Aero Engines, 1996). 49 The propulsive efficiency is affected by the “amount of kinetic energy wasted by the propelling system. Waste energy dissipated in the jet wake, which represents a loss, can be expressed as W (vj – V)/2g where (vj – V) is the waste velocity. It is apparent that at the aircraft lower speed range the pure jet stream wastes considerably more energy than a propeller system and consequently is less efficient over the range. However, this factor changes as aircraft speed increases, because although the jet stream continues to issue at high velocity from the engine its velocity relative to the surrounding atmosphere is reduced and, in consequence, the waste energy loss is reduced” (Rolls-Royce, 1995: 223). W = mass flow in pounds per second, V = velocity of aircraft in feet per second, vj = velocity of flow in feet per second, and g = gravitational constant.

74

greater propulsive efficiency than a turboprop. In the case of Concorde the turbojet propulsion efficiency is about 90%. From a technical viewpoint, therefore, turboprops are more appropriate for short-haul aircraft, whereas turbojets are more appropriate for long-haul aircraft. However, regulation has heavily affected the selection of engine configurations. According to Mowery and Rosenberg (1982) during the 1950s and 1960s the development of turboprop engines was somewhat hindered by the heavily regulated industry environment. In the US airlines were not competing on fares, so that there were no incentives for airlines to adopt the economical turboprop for short-range uses (Caves, 1962). Mowery and Rosenberg (1982) argue that the turboprop engine represents a missed opportunity. The fuel crisis of the 1970s and the legislation imposing more severe restrictions on aircraft noise have re-tipped the balance towards turboprop engines for short-haul aircraft (Jane’s Aero Engines, 1996). Figure 4.2. Broad-brush comparative propulsive efficiency (source: adapted from Gunston, 1995)

Propulsive efficiency

80

60

40

20

0

400

600

800

1000

Airspeed mph

Turboprop Contrarotating fan Propfan High by-pass turbofan Low by-pass turbofan Pure turbojet Turboprop engines were ‘rediscovered’ in the early 1980s also for long-haul aircraft. Improved turboprop technology has made possible the design of multi-bladed propellers capable of turning at high speed without loss of propeller efficiency. This 75

configuration, also labelled prop-fan, is still under study. However, turboprop engines power a relatively small and declining part of aircraft as compared to turbofan engines (Flight International, 3-9 November, 1999). In fact, according to fieldwork interviewees, airframers are increasingly converting their turboprop aircraft into turbofans (Interviews, 1998, 1999). 3.5. The turbofan engine A turbofan is a jet engine configuration lying between a turboprop and a turbojet. In the turbofan the thrust is generated by the fan mounted at the front of the engine and by the core. Figure 4.3 is a cutaway drawing of the Rolls-Royce Trent 800 turbofan engine. The proportion of thrust generated by the fan depends on the amount of mass flow bypassing the core engine. This is usually expressed by a ratio, the by-pass ratio (BPR), that is the “numerical ratio of the mass flow entering the fan duct divided by that entering the core” (Jane’s Aero Engines, 1996). The fan is driven by a low-pressure turbine that uses some of the hot gas downstream of the high-pressure turbine. The bypass air is used via an exhaust nozzle to lower the mean jet temperature and velocity. As a result, the turbofan engine deals with larger airflow and lower jet velocities as compared with a turbojet (Rolls-Royce, 1986). The bypass concept applied in turbofan engines has filled the gap between turbojet and turboprop in terms of propulsive efficiency. Figure 4.2 highlights this point. The turbofan also shows a better specific fuel consumption than the turbojet engine (Gunston, 1995; Rolls-Royce, 1986).

Figure 4.3. The Rolls-Royce Trent 800 turbofan engine (source: Rolls-Royce internal documentation)

Frank Whittle invented the turbofan in 1936. Examples of early turbofan engines are the Rolls-Royce Spey and Rolls-Royce Conway. They were both launched in the early 1960s. Their by-pass ratio was less than 1 and their thrust range varied between 10,000lb and 22,000lb 50. According to Gunston, two factors limited the by-pass ratio of the early turbofan engines. The first was that these engines had to be fitted in military aircraft. This constrained the diameter of the fan. The second factor was that 50

The Spey has a BPR of 0.71, the Conway 0.42 (Jane’s Aero Engines, 1996).

76

aerodynamicists wrongly calculated the drag of an engine installed under a wing or by the rear fuselage. Current turbofan engines such as the GE90 engine are characterised by higher by-pass ratio, up to 9. Due to their low specific fuel consumption, low noise, and high propulsive efficiency, turbofan engines are the most common technological solutions in civil applications. This holds true over a wide thrust range. Current high by-pass turbofan engines show overall efficiencies of 35 per cent compared to about 25 per cent efficiency achieved by the low by-pass turbofans of the early 1960s. The turbofan engine is characterised by two design layouts, namely two-shaft and threeshaft. 3.6. Engine efficiency and related design parameters In a jet engine the two most important design parameters are the overall pressure ratio and the turbine entry temperature. As mentioned in Section 3.3, improvements in cycle efficiency call for higher pressure ratio and higher turbine entry temperature. Early turbojets were characterised by a pressure ratio of 4:1. Today’s engines show a pressure ratio of 40:1 as in the Rolls-Royce Trent and GE90 engines. Turbine entry temperatures have been increased from about 1000K in the Whittle engine to today’s 2100K (Singh, 1996). As is explained later, increases in pressure ratio and turbine entry temperature have been achieved as a result of large investments in heat resistant material, encompassing both new materials and cooling technologies.

The turbofan has two extra design parameters, by-pass ratio and fan pressure ratio. If these design parameters “are varied in order to maintain a constant specific thrust, then it is advantageous […] to increase turbine entry temperature and overall pressure ratio together, which not only improves specific fuel consumption but also results in a small, light core” (Jane’s Aero Engines, 1996). There are, however, some technical considerations that impose an upper limit to increases to these design parameters. For instance, a high by-pass ratio requires a large-diameter fan, which amongst other things, introduces installation problems, increased nacelle drag, and variable pitch-fan 51 that may well offset the improvements deriving from the higher by-pass ratio. Likewise, pressure ratio and turbine entry temperature cannot be increased ad infinitum due to limitations related to material properties and cooling technologies. 3.7. Factors influencing engine design As mentioned in Section 2.6, airline strategies heavily influence technical change in the aircraft engine industry. The demand of aircraft in terms of size and range depends on an airline’s route structure. Changes in route structures are triggered by changes in the regulatory network imposed by national governments. Long- and short-range aircraft require the optimisation of different engine design parameters. This section offers comments on the relationships between aircraft and engine design parameters.

Aircraft design characteristics are affected by a number of factors such as payload, range, runway length, airport infrastructure and operational factors, and operating costs, which in turn heavily influence the key characteristics and performance of engine design. Engine design parameters are optimised according to the intended airframe application (Interview, 1999). Short- and medium-haul aircraft are exposed to a larger number of take-off and landing cycles than long-range aircraft. In technical jargon, they accumulate more flight cycles than long-range ones (Condom, 1998). As a 51

It should be noted that variable pitch-fan is not a current technology.

77

consequence, their engines are subjected to higher temperatures more often and cyclically lifed parts will require more frequent replacement. There is thus a requirement for a larger volume of spare parts. Reliability is paramount 52. In longrange aircraft, fuel consumption is the critical engine design parameter to be optimised. Table 4.2 compares amongst other parameters the fuel requirements for the Boeing 747400, the current largest long-range aircraft, and a new large aeroplane, possibly the Airbus A3XX or Boeing 747-500/600. It is apparent that the larger the aircraft and the longer the range, the higher the fuel requirements (i.e. its weight). As a consequence, a 1% improvement in engine specific fuel consumption would result in an increase in equivalent payload (Cumpsty, 1997). Table 4.2. Comparison of some salient aircraft parameters (source: adapted from Cumpsty, 1997) New Large Aeroplane Boeing 747-400 No of passengers 620 400 Range (nautical miles) 8000 7300 Payload at this range (a) 58.8 tonne 38.5 tonne Max take-off weight (d = a + 635.6 tonne 395.0 tonne b + c) Empty weight (b) 298.7 tonne 185.7 tonne Fuel capacity (c) 275.4 tonne 174.4 tonne Cruise Mach number 0.85 0.85 Initial cruise altitude 31000 ft 31000 ft Cruise Lift/Drag 20 17.5 2 511 m2 Wing area 790 m Howse (1998) explains clearly how aircraft requirements influence engine requirements. In particular, he compares a short-range wide bodied twin aircraft with a long-range four-engine aircraft. He shows that although both aircraft require the same take-off thrust 53, payload and range requirements it is not appropriate to use the identical engines in both aircraft. In fact, “the payload and range of twin engine aircraft tend to be dominated by engine take-off thrust and engine weight … engine sfc and noise play their part in the engine configuration but are less critical if reasonably competitive.… For a long-range four-engine aircraft, fuel burn and noise tend to be the engine design determining factors.” (Howse, 1998: 2). Therefore, the same engine rated at the required thrust does not represent the optimal solution for a twin or for a quad aircraft. “The engine for the twin aircraft will have a lower bypass ratio with a smaller diameter fan and bigger core, the latter to keep the engine temperature manageable for reliability at the highest thrusts and to meet the top of climb thrust condition. The engine for the four-engine aircraft needs a higher bypass ratio to give the best sfc, and so fuel burn, and noise. Thus 52

However, this does not mean that reliability is unimportant in longer-range aircraft. The engine take-off thrust is determined from the runway length, range, payload and number of engines (Howse, 1998). 53

78

the fan will be larger and the core smaller, the size of the core being determined by physical constraints or top-of-climb thrust requirements.” (Howse, 1998: 2). 4. Product characteristics and technological requirements in the aircraft engine industry Early turbojet engines, such as that designed by Frank Whittle in the 1930s, were relatively simple machines. The only moving part was the compressor-turbine combination. Today’s engines are much more complicated machines. Engine designers have added a number of subsystems in order to achieve higher and better performance as demanded by the regulatory bodies, airframers and airlines. The large turbofan engines can encompass up to 40,000 components (Interviews, 1997, 1998, 1999). These components may differ in kind and variety. They belong to different and often distant technological fields. To name but a few, thermodynamics, aerodynamics, fluid dynamics, tribology, heat transfer, combustion, structures, materials, manufacturing processes, instrumentation, and controls (Mattingly et al., 1987). The number of components can increase further over time, as firms have to cope with customers’ evolving needs as well as ever-tighter regulations. The next sections offer comment on some developments in aircraft engine technology. 4.1. Combustion technology Progress in the gas turbine combustion process has stemmed from improvements in aerodynamics, chemistry, physics and mechanical design. The combustion chamber (combustor in the US) has registered the most substantial reductions in weight and size compared to other units in the engine (Gunston, 1995).

The mechanical design of the combustion chamber has gone through several arrangements of the combustion process. The Whittle engine was characterised by a giant curved chamber which was surrounded by relatively small rotating parts. Today’s engines show relatively small chambers surrounded by giant rotating parts. The most common design is the fully annular chamber. Compared to other designs, such as the can-annular chamber, a fully annular chamber makes the best use of the space between the compressor and turbine sections. It delivers the same power in three-quarters of the length of a can-annular chamber of the same diameter. The main objective in combustion chamber design is to deliver a flow of gas at the maximum temperature that can be withstood by the high-pressure turbine. Aerodynamicists have improved the design such that the airflow downstream of the compressor is slowed (air velocity is turned into pressure) early on in the combustion chamber so that the combustion flame is not continually blown out. Around 60% of the airflow is used to cool the combustion gas (from 2,000° C to 1,400° C) and to form a barrier of film-cooling between the hot gas and the inner wall of the combustion chamber. The remainder of the airflow is used to mix the fuel and to stabilise the flame. The combustion chamber produces four types of emission: hydrocarbons and carbon monoxides at low power, smoke and oxides of nitrogen (NOx) at high power. The progressive tightening of environmental legislation has driven developments in combustion technology to reduce emissions. The refinement of the mixing process in the combustion chamber and changes to the fuel injectors to improve fuel atomisation have positively affected the reduction of low-power emissions (Ruffles, 1992). In 79

addition, engine manufacturers have increased combustion temperature to boost fuel efficiency, which has resulted in lower carbon monoxide and hydrocarbon emissions 54 (Keller, 1992). 4.2. Control technologies In the last 30 years control systems have been characterised by relentless technological change. Digital electronics have superseded hydromechanical technology. Chapter 7 documents the technological and ensuing functional evolution of control systems in more detail. Suffice to say here that full authority digital engine controls (FADECs) optimise steady state and transient engine performance by controlling all engine functions, such as tip clearance, bleed valves, and thrust reverser. This has reduced pilot workload and above all allowed improvements in engine efficiency and reliability. Further benefits are expected from linking the FADEC with the engine health monitoring system (Interview, 1999). 4.3. Material technologies Material technologies are and will be a key technology in gas turbine engines. The ultimate efficiency of the engine is in fact limited by material the operating temperature. As mentioned earlier, two fundamental design parameters in engine design are overall pressure ratio and turbine entry temperature. The introduction of high temperature materials and cooling technologies has enabled improvements in both design parameters. Materials play a major role also in engine weight reduction. High-strength materials such as titanium and nickel have gradually replaced steel, reducing weight and achieving a higher thrust-to-weight ratio. In addition, materials can be considered to be a pervasive technology. All engine components can be potentially designed for enhanced performance by application of improved material technology.

Materials are chosen according to the maximum temperature they can sustain. Aluminium or aluminium alloy is used for cool parts such as inlet, casings, gearboxes, and accessories (Gunston, 1995). Highly stressed parts require more capable materials. Titanium and nickel alloys have gradually replaced steel, which until 1960 accounted for about 60% by weight of an aero engine, achieving significant improvement in temperature capacity, operating life and weight saving. Titanium alloys are used for compressor and fan blades, drums and structures, given their high strength-to-weight ratio up to 500°-600° C. Nickel-based alloy applications are used in the high-pressure combustion and turbine areas. These two alloys account for about 65% by weight of the 1990s’ gas turbine engine (Howse, 1998). Thermal barrier coatings, such as ceramic coatings, are applied to high-pressure turbine blades and vanes to insulate the metal from the hot gas (Jane’s Aero Engines, 1996). Further increases in temperature will require the development of even lighter and stronger materials than those of today’s, capable of operating at much higher temperature. Research is underway for new materials such as composites. “Composites are materials made up of more than two dissimilar components” (Gunston, 1995: 91). They show several advantages over metal alloys. The major advantage is their high strength-to-weight ratio, which could lead to lighter structural engine parts. For instance, resin-based composites have found applications in non-load bearing, low 54

Chemists have made it possible to achieve a combustion efficiency of about 100% in a very small burning length (Gunston, 1995).

80

temperature applications, such as nose cone, fairings, and cowl doors (Ruffles, 1992). The most impressive achievement for composite materials has been the introduction of a hybrid composite-titanium fan blade by General Electric Aircraft Engines in the GE90 engine. The fan is made of intermediate-modulus carbon fibre in a matrix of toughened epoxy. Composite materials are the way forward also in the hot section of the gas turbine engine. Some composite materials such as Silicon Carbide (SiC) can withstand temperatures of up to 1000° C. However, they are characterised by a rigid crystalline structure that makes them more brittle than metal alloys. More promising materials are the so-called fibre-reinforced materials, such as metal matrix composites (MMCs) and ceramic matrix composites (CMCs). An example of a MMC is titanium matrix composite (TMC), whose high specific strength and stiffness may make it suitable for fan parts and compressor applications (Hutchinson, 1995). However, the application of these materials to gas turbine engine components has some drawbacks. Because of the titanium alloy used as matrices, these materials would suffer at high temperatures. Furthermore, these materials are characterised by high manufacturing costs and irrespective of the use of powerful computational modelling techniques to better understand their behaviour, the mismatch in thermal expansion coefficients between matrix metal and reinforcing fibre is not fully understood (Piellisch, 1991). CMCs may be instead more suitable for high-temperature applications. Silicon carbide fibres in a silicon carbide matrix (SiC/SiC) have the potential to operate uncooled 55 to about 1200° C (Howse, 1998). As already mentioned, materials can be considered a pervasive technology in gas turbine engines as they cut across the entire engine. Other than being critical in improving the engine’s thermodynamic cycle, material technologies are also important for the engine’s external components. For instance, in large turbofan engines, higher by-pass ratios require light, high-strength materials for the nacelle otherwise the increased engine weight would cancel out the benefits of increased by-pass ratio (Jane’s Aero Engines, 1996). Polymer matrix composites have already been introduced in the nacelle structure, engine doors, and thrust reverser (Hutchinson, 1995). 4.4. Manufacturing techniques Progress in high-pressure blades has been and is being achieved as a result of large investments in new alloys, cooling techniques, and new methods of castings. The combined effect of these three-pronged investments has enabled substantial increases in turbine entry temperatures and, as a consequence in the engine thermodynamic cycle. Having already discussed improvements in the form of new alloys and thermal barrier coatings, this section briefly illustrates some major improvements in casting techniques and internal airflow, highlighting the relevance of new manufacturing techniques (particularly in metallurgy) to achieve higher product performance. In some cases, entirely new manufacturing equipment is required. As shown later in Chapters 5, the patent analysis supports the point that innovation in product technologies calls for innovation in process technologies. Interviewees also confirmed this point.

“Casting techniques have been revolutionised by controlling the way the metal solidifies from the melt” (Gunston, 1995: 96). When bulk metals solidify discrete crystals form 55

It is worth noting that a reduction in blade internal cooling would benefit engine performance.

81

and grow with a random orientation. This phenomenon renders metals weak since the strength of the joints between crystals is weaker than that within them. The first progress in solidifying processes was the introduction of directionally solidified casting at the beginning of the 1970s. According to this casting technique, the solidification process of the bulk metal is carefully controlled in such a way that the crystal grows along the entire length of the casting rather than randomly. In this way, hightemperature resistance and blade life can be achieved. A further advance was made with the introduction of single-crystal casting in the 1980s. In this casting technique, through careful controls, only one crystal throughout the entire casting is enabled to grow (Gunston, 1995). Single-crystal alloys have a temperature advantage of about 30° C over directionally solidified castings (King, 1995). Improved internal cooling airflow has also increased turbine entry temperature. Cooling techniques have demanded the use of precision ceramic cores that enable the manufacture of blades completely enveloped in a film of cooling air. Chapter 6 illustrates the introduction of the wide-chord fan blade by Rolls-Royce, highlighting that a product innovation has demanded new manufacturing techniques, such as superplastic forming and diffusion bonding. Other important manufacturing techniques used for engine components are machining (milling, turning, broaching), electro-chemical machining, electro-discharge machining, chemical etching, spot welding, seam welding, electron-beam welding (Gunston, 1995). 4.5. Computation fluid dynamics Engine performance is also improved by augmenting the efficiency of the engine components via a reduction in aerodynamic and mechanical losses. This has been achieved through the introduction of advanced theoretical methods of computing the airflow and advanced stress and vibration analysis. Computational fluid dynamics (CFD) codes enable aerodynamicists to decompose “the airflow in millions of small elements whose flow properties can then be easily calculated using the laws of fluid mechanics” (Jane’s Aero Engines, 1996). CFD has enabled designers to optimise the shape of compressor and turbine blades to improve efficiency. Further benefits have been achieved in relation to increases in stage loading without increasing loss and blade matching of multistage compressors that allow a reduction in blade and stage number (Howse, 1998). All in all, CFD has enabled higher pressure ratios to be achieved. Finite element analysis (FEA) techniques enable design analysts to “divide structures [e.g. blades] into millions of small elements whose mechanical and thermal properties can then be easily calculated using the laws of solid mechanics and heat flow” (Jane’s Aero Engines, 1996). 4.6. Testing technologies Safety and reliability issues require engine manufacturers to carry out severe and extensive tests on engines. Tests are performed for different reasons. For development and certification purposes, to test the validity of the engine design and identify areas that may require further scrutiny and/or resources in the development stage. During production, testing in terms of quality control is performed on raw materials, parts, assemblies, and the entire engine. Engines in service are continually tested by air carriers or specialised maintenance service companies to meet safety regulations issued by the regulatory agencies (Dawson, 1991).

82

Non-destructive inspection encompasses a large set of testing techniques used for development, quality control, and field-service purposes. At the component level, such as casting and forging, these techniques enable the detection of flaws that have occurred during the production of the part or during service. It is worth mentioning ultrasonic testing, radiographic testing, magnetic particle testing, and eddy current testing (Davis, 1990). Other than for single parts or components, non-invasive techniques are also used to test complete engines. In particular, high-energy X-ray techniques are used to measure the movements of metal parts and fluid flows while the engine is running (Stewart, 1987). Non-invasive techniques can allow problems to be identified and action to be taken without always stripping the engine. As a consequence, non-invasive techniques reduce time and costs of engine development. Application of computer tomography is currently under study for similar purposes (Interview, 1997). Computer simulation is being extensively implemented and used for development purposes. Simulation enables design and development engineers to assess a larger set of design alternatives in a relatively shorter time than by physical testing. Taking into account the initial investments in hardware and software, computer simulation allows substantial cuts in the costs of the testing programmes. The testing of physical prototypes is, by definition, expensive because of the tests that involve significant life of engine parts, which may be also destroyed. Computer simulation provides a more precise visualisation of the dynamics of the test that allows identifying the weaknesses of the design alternatives. However, though substantial improvements have been made in modelling and meshing techniques, computer simulation has not superseded physical testing completely. Computer simulation is used in conjunction with physical testing 56. Further, computer models are based on experimental data gained by physical testing. 5. Conclusions The aim of this chapter was to set the scene for the discussion on the engine maker’s technological capabilities carried out in the following chapters. The chapter analysed the key characteristics of the aircraft engine industry and technology. Using the framework proposed by Miller et al. (1995) to study CoPS industries, the chapter analysed the role of each actor in the industry highlighting their influence on source, rate, and direction of technical change. Particular attention was paid to the dense regulatory network imposed by national governments and certification bodies and how it stimulates and affects the introduction of new technologies. Engine makers are required to have a clear understanding of such rules as well as of customer requirements. The analysis has highlighted the organisational and technological coordination endeavours that engine manufacturers undertake throughout new engine development programmes. The chapter delved into the nature of engine product characteristics and technological requirements. It illustrated the different types and suitability of aircraft engines and the factors influencing their design. The analysis of the major technological developments highlighted how engine performance improvements have stemmed from a variety of technological fields. The multicomponent and multitechnology nature of the aircraft

56

During one company visit, an engineer presented his views on the use of computer simulation for development purposes. As the engineer was arguing convincingly that computer simulation was about to displace physical testing, we were passing by an enormous building. It was the recently built testing facility for the new engine family introduced by the company.

83

engine requires that engine makers be active in multiple technological fields in order to design, develop, integrate, and manufacture engines. According to the analysis carried out in this chapter, engine makers’ activities span many different knowledge domains: (a) the scientific and technological fields underpinning the high variety of components and subsystems, (b) organisational (e.g. project management) and relational (e.g. marketing) capabilities required to manage and integrate the roles of the actors involved in the industry; (c) knowledge about client requirements, and (d) knowledge about rules and regulations for engine certification. Figure 4.4. The drivers of technological and organisational capabilities of engine manufacturers

Regulatory bodies • •

Certification Deregulation

Engine maker •

Multiple technological fields •



Engine design parameters

Project management and relational capabilities

Airlines •

Route structure

Aiframers •

Aircraft design parameters

84

The challenging task for engine manufacturers is to integrate these different knowledge domains in order to be able to identify business opportunities, and translate regulatory and customer requirements into technical specifications that can be met relying on its own and suppliers’ technological capabilities. The scheme pictured in Figure 4.4 provides a summary of the main findings of the chapter. The figure illustrates the major drivers of engine makers’ technological and organisational capabilities. Engine manufacturers must develop capabilities in multiple technological fields to be able to integrate a variety of components from different technological fields. They also need project management and marketing capabilities to manage the role of each actor during new engine programmes and across them. The regulatory bodies heavily influence technological developments via certification, noise, and emissions rules. Airlines’ route structure strategies (in turn affected by deregulation) affects their fleet composition in terms of aircraft size. Since the engine is a subsystem of the aircraft, changes in aircraft design parameters also affect changes in engine design parameters.

85

CHAPTER 5: INTRAMURAL CAPABILITIES AND EXTERNAL LINKAGES IN THE AIRCRAFT ENGINE INDUSTRY 1. Introduction Chapter 4 provided an overview of the key characteristics of aircraft engine technology and industry structure. The purpose was to identify the managerial challenges that engine makers face in terms of the different types of capabilities required to compete, namely technological, organisational, and relational. Relying on the analysis in Chapter 4, the purpose of this chapter is to investigate the dynamics of the boundaries of the technological capabilities of engine manufacturers in the light of: (a) the product characteristics and technological requirements of the aircraft engine, and (b) the driving forces at work in the industry that impinge on the technological capabilities of engine makers. In particular, this chapter sets out to analyse the dynamics of the breadth and depth of engine makers’ technological capabilities. Breadth is understood in terms of number of technological fields (e.g. combustion chamber technology, control systems technology) maintained in-house by engine makers. Depth is understood in terms of the different stages of new engine development (e.g. research, design, manufacturing, testing). The analysis of the dynamics of the breadth and depth of engine makers’ technological capabilities directly addresses the central argument of this thesis since it reveals the extent of the division of labour between engine makers and suppliers in terms of its direction and constraints. The chapter addresses these issues using a combination of qualitative and quantitative data. The detailed analysis of the nature and the direction of the drivers impinging on engine makers’ capabilities is based on (a) qualitative data collected via interviews with company engineers and industry experts and (b) quantitative data on companies’ collaborative agreements and product portfolios gathered through a database search. US patent statistics are used to illustrate firms’ technological profiles as a proxy measure of their technological capabilities. These data permit an assessment of the impact of the driving forces on companies’ technological capabilities. In order to present a fine-grade analysis of the boundaries of engine makers’ technological capabilities, the chapter relies on a sector-specific map obtained by breaking down the US Patent Office classification. The method is fully described in Chapter 3. This chapter shows that the technological boundaries of the firm differ from the boundaries of the firm as defined by make-buy decisions. Although engine manufacturers make extensive use of collaborative agreements to develop new engines and to conduct research on new technologies, they maintain in-house a broad spectrum of capabilities spanning different technological fields. The scope for technology outsourcing in the aircraft engines is limited due to the compelling technological requirements involved in the integration of the engine. In other words, systems integration shapes the scope of technology outsourcing. The time-based analysis of engine maker’s technological capabilities shows that their breadth has been increasing over time. Engine makers capabilities must span over a wide range of technological fields to co-ordinate the work of, and to deal with, the other actors in the industry, namely suppliers, airlines, airframers, governments, and regulatory bodies. The chapter is organised as follows. Section 2 details the driving forces impinging on engine makers’ technology bases. In the light of this, Sections 3 discusses the pattern of 86

division of labour in the industry. Sections 4 and 5 present the results of the analysis of engine makers’ technological profiles as reflected in US patent statistics. Section 6 presents the conclusions to the chapter. 2. The driving forces As highlighted in Chapter 4 the specific product characteristics of the aircraft engine and its increasing complexity call for engine makers with all-round knowledge. There are, however, some driving forces that play a counteracting role that affects engine makers’ technological capabilities (Table 5.1). As argued below, the combined overall effect of these factors ‘enables and pushes’ engine makers towards greater externalisation of activities. The first set of driving forces described below has been labelled enablers. Internal accumulated knowledge of components interfaces, for example, leading to the modularization of the engine enables engine makers to specify components and subsystems fully and, as a consequence, to outsource them to first-tier suppliers. Enablers affect directly the nature of engine makers’ technological capabilities. At the same time another set of driving forces labelled pushers literally pushes engine manufacturers to hive off larger chunks of activities related to the design, development, and manufacture of the engine system. For instance, in order to secure orders in developing countries engine manufacturers are often forced to outsource the development and/or the manufacture of some components to local suppliers. 2.1. The enablers Internal accumulated technological knowledge related to the engine system enables engine makers to understand better the behaviours of components within the engine’s operating environment. The accumulation of this ability rests upon advances in scientific and technological disciplines, complex mathematical models, and design knowledge embodied in people. The use of powerful computers and sophisticated computer models also underpins the progress of knowledge in this industry. In particular, as underlined in Chapter 4, computational fluid dynamics (CFD) enables engine designers to optimise the shape of compressor and turbine blades to improve their efficiency. The use of increasingly powerful computers does not, however, downplay the importance of empirical data gathered via experimental activity and during engine in-flight service, and the engineers’ embodied tacit knowledge. CFD codes are, in fact, always validated using empirical data related to engine behaviour and performance stored in the databases held by engine makers. These databases represent an important and extremely valuable part of the memory of these companies. In addition, engineers’ accumulated experience and rules of thumb are still prominent in the choice of new design routes.

A better understanding of the technological principle governing the engine system enables engine manufacturers to conceive engines in terms of basic modules. In particular, engine makers can specify component interfaces in terms of their functional and spatial specifications as well as their physical properties. In this way most complex interactions take place within rather than across the same modules. Modular aero engines have been launched since the early 1970s. The section on aircraft engines contained in Jane’s All the World Aircraft has included ‘modular engines’ and ‘engines composed of modules’ since 1970. From an operational viewpoint, the concept of modular engines has its origin in the efforts of engine makers to standardise component 87

parts to ease the maintenance of engines in use. Gardiner and Rothwell (1990) neatly illustrated the case of the RB211 engine for which Rolls-Royce was able to scale-down or scale-up (or more precisely de-rate and up-rate) the original design to cater for a variety of market requirements and power outputs. Such modular design enabled RollsRoyce to exploit economies of scale and scope across a substantial number of engines and over time. Table 5.1. The driving forces affecting engine makers’ technological capabilities. (Source: author’s elaboration on interview data) Enablers Pushers • Internal accumulated technological • Spiralling costs of development knowledge of the engine system behaviour • Accumulated technological knowledge of • Pressure from developing countries the engine components suppliers • Increasing use of ever more powerful • Cut in defence budget (entailing computational capacity personnel reduction) • Knowledge codification process • Advantages of specialisation • Modularization of the engine system This design philosophy is common in the industry. By adopting it, manufacturers can launch engine families rather than single engines and take advantage of the associated cost benefits. Using a common engine core manufacturers spread the recovery of nonrecurring design costs over several market segments and benefit in terms of cost from larger scale production. Thrust ranges can be targeted by the addition to the engine of a tailored low-pressure system. Modularity in this industry also paved the way for the practice of ‘retrofitting’ and ‘design commonality’ whereby engine makers improve existing engine versions by introducing technological modifications, which, in turn, permit performance improvements within the same thrust range. The Chief Design Engineers interviewed confirmed that modularity is a powerful tool to reap economies of scale, scope, and knowledge in design, production, and use. Similarly, they confirmed that there is strong trend towards greater modularity. “Engines are defined (i.e. configured) by their duty, then you try to get as much modularity as you can” (Interview, 1999). Table 5.2 depicts the number of engine families developed by the Big Three between 1977 and 1996 57. The number of engine versions per family is also presented. The data in the table confirm the point made above, that manufacturers develop engine families rather than single engine in order to exploit both economies of scale and scope in design and production tasks. Further, the number of engine versions per family can be used as a rough measure of the modularity of engine architectures. The table shows that this measure increases over time confirming that there is a trend towards greater modularization, since the greater the number of versions the more modular the engine architecture. For example, engine families A4, A7, and A10 display an increasing number of versions over time. Those families (e.g. B2 and B3) that show a steady number of engine versions over time are characterised by less modular architectures.

57

The method of compilation of the database on which Table 5.2 is based is fully described in Chapter 3.

88

Table 5.2. Engine manufacturers: number of versions per engine family 1977-1996. (Source author’s elaboration on Jane’s All World Aircraft data) Company A 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 Family A1 1 1 1 1 1 1 1 1 1 1 1 2 3 3 4 6 6 Family A2 11 14 14 13 13 13 13 13 13 13 13 2 2 2 2 2 2 2 2 2 Family A3 3 4 5 4 4 4 4 4 4 4 4 4 4 4 4 4 4 Family A4 1 1 1 1 10 14 14 13 13 13 15 15 21 21 Family A5 3 3 3 3 3 4 4 Family A6 3 3 3 3 3 3 3 3 3 3 Family A7 1 2 3 9 10 15 12 14 15 17 19 20 20 25 26 26 Family A8 7 7 7 7 7 7 7 7 7 7 Family A9 2 2 2 2 2 2 2 2 2 2 2 2 Family A10 1 1 1 1 3 3 3 5 4 5 5 6 8 8 8 8 9 9 8 8 Family A11 1 1 1 1 2 2 Company B 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 Family B1 1 1 1 1 1 1 1 1 1 1 1 Family B2 12 12 12 12 12 9 9 9 9 9 9 9 9 9 9 9 9 9 Family B3 2 2 2 2 2 3 4 4 6 5 5 5 5 5 5 6 5 5 5 5 Family B4 8 9 11 11 11 9 9 9 9 9 10 10 10 10 10 10 11 11 11 11 Family B5 1 8 12 6 6 6 6 6 7 7 7 7 7 7 7 7 7 7 Family B6 2 2 2 1 2 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 Family B7 1 1 1 1 7 8 8 9 11 13 13 15 15 Company C 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 Family C1 2 2 2 2 2 3 3 4 4 4 1 1 Family C2 2 2 2 Family C3 11 10 9 9 9 9 9 1 1 1 1 1 1 1 1 1 Family C4 1 1 1 1 4 4 4 4 4 4 4 4 4 4 4 4 Family C5 4 5 5 6 8 7 8 8 8 11 10 12 12 12 12 12 8 8 8 8 Family C6 1 3 3 2 3 3 2 2 2 2 2 2 2 3 3 3 3 3 3 Family C7 1 1 1 1 1 1 1 1 Family C8 1 1 2 2 3 4 4 5 6 6 5 5 5 5 Family C9 1 6 7 8 10 10 10 Family C10 9 9 9 9 9 9 9 9 9 10 10 10 10 10 10 10 10 10 10 10

89

90 The table also shows that other engine families such as A2 and C3 display a decreasing number of engine versions during the period examined. These engine families were launched in the early 1970s and the decreasing number of engine versions reflects the policy of engine manufacturers to reduce the number of engine versions or those families based on old architectures. Chapter 8 discusses this issue further. 2.2. The pushers As mentioned in Chapter 4, new engines are increasingly developed via international collaborative agreements amongst engine makers and suppliers. The development of new engines is split amongst different companies because of the increasing development cost and related risk of failure of the programme. Competition in the large turbofan market is fierce and price-based. Tables 5.3-5.8 report on companies’ collaborative agreements between 1985 and 1996 58. The relatively low number does not permit a statistical analysis of the portfolio of agreements of each company. However, a close scrutiny of the content of each of them proved to be extremely useful. The content of each agreement was reviewed and arranged according to the kind of activity involved in the contractual agreement. A two-tier classification was devised in order to investigate at which level in the industry the division of labour occurs. Six types of activities are identified, namely research, development, manufacturing, testing, maintenance, and ‘other’. Also, those agreements involving engine development and manufacturing were further classified in relation to the scope of the deal.

The tables show that the number of agreements displays a sharp rise in 1992-1996 (from 30 to 65). Analysis of the content of each agreement reveals that they are set up for several reasons. The activities concerned include suppliers’ involvement in new engine programmes, research on new engine technologies, manufacturing of engine components, and engine maintenance and repairing. Table 5.3. Total number of collaborative agreements (Source: author’s elaboration on SDC data; number in brackets indicate agreements that have been terminated) Activity Scope 1987-91 1992-96 Subtotal Design/manufacturing Engine 12 27 (1) 39(1) Design/manufacturing Component 4 4 Manufacturing Engine/component 3 7 10 Research 7 11 18 Testing 1 1 Maintenance 2 11 13 Design/manufacturing Airframer 3 2 5 Other 3(1) 2 5(1) Subtotal 30(1) 65(1) 95(2) It is worth noting that whereas agreement portfolios of RRSPs are focused almost exclusively on new engine programmes, engine makers are involved in agreements of all sorts. In particular, alongside new engine programmes, manufacturers’ agreements involve research, development, and manufacturing of specific components and testing equipment. The broader scope of the agreement portfolios of engine makers indicates that they resort to a larger number of external sources than RRSP companies. Not only 58

The method of compilation of the database on which Tables 5.3-5.7 are based is fully described in Chapter 3.

91 do engine manufacturers enter collaborative agreements to develop new engines, but they also team up with research centres and suppliers to acquire and develop new technologies for future engine concepts. Using US patents Section 4 assesses the impact of the increasing use of external sources of technologies on engine manufactures’ technology bases 59. Table 5.4. Company A’s number of collaborative agreements (Source: author’s elaboration on SDC data) Activity Scope 1987-91 1992-96 Subtotal Design/manufacturing Engine 2 7 9 Design/manufacturing Component Manufacturing Engine/component 1 2 3 Research 1 2 3 Testing Maintenance 4 4 Design/manufacturing Airframer 1 1 Other Subtotal 4 16 20

Table 5.5. Company B’s number of collaborative agreements (Source: author’s elaboration on SDC data; number in brackets indicate agreements that have been terminated) Activity Scope 1987-91 1992-96 Subtotal Design/manufacturing Engine 2 10(1) 12(1) Design/manufacturing Component 1 1 Manufacturing Engine/component 2 4 6 Research 2 4 6 Testing Maintenance 1 5 6 Design/manufacturing Airframer 2 2 Other 3(1) 2 5(1) Subtotal 12 (1) 26(1) 38(2) The largest number of agreements (39 out of 95) falls in the category related to new engine development programmes. This result reflects the increasing use of partnership agreements for the development of new engines. CFM International, for instance, is a long-standing joint venture between General Electric Aircraft Engines and SNECMA. The CFM56 has become the world’s best selling engine. International Aero Engine is the response to this joint venture by Rolls-Royce and Pratt & Whitney teaming up with MTU, Fiat Avio, and Japan Aero Engine. Furthermore, new turbofan engines, such as the GE90, the Rolls-Royce Trent, and the Pratt & Whitney PW4000, were developed using RRSP agreements. As mentioned at the beginning of the section, the flourishing

59

Nearly all agreements concerning maintenance (12 out 13) were set up by engine manufacturers. This confirms the entry of engine manufacturers into the maintenance business via joint ventures with airlines and/or acquisition of airlines’ maintenance divisions and specialised suppliers. As confirmed by the interviewees, a long-term objective of engine makers is to provide total system solutions (i.e. engine + maintenance service). Engine makers have also entered agreements with airframers in order to become sole suppliers of a particular aircraft version.

92 use of RRSP is due to the rising development costs of new engines and high risks of programme failures. Table 5.6. Company C’s number of collaborative agreements (Source: author’s elaboration on SDC data) Activity Scope 1987-91 1992-96 Subtotal Design/manufacturing Engine 5 2 7 Design/manufacturing Component 1 1 Manufacturing Engine/component 1 1 Research 1 4 5 Testing 1 1 Maintenance 2 2 Design/manufacturing Airframer 1 1 2 Other Subtotal 7 12 19 Table 5.7. RRSP A’s number of collaborative agreements (Source: author’s elaboration on SDC data) Activity Scope 1987-91 1992-96 Subtotal Design/manufacturing Engine 2 4 6 Design/manufacturing Component 1 1 Manufacturing Engine/component Research 1 1 2 Testing Maintenance Design/manufacturing Airframer Other Subtotal 3 6 9 Table 5.8. RRSP B’s number of collaborative agreements (Source: author’s elaboration on SDC data) Activity Scope 1987-91 1992-96 Subtotal Design/manufacturing Engine 1 4 5 Design/manufacturing Component 1 1 Manufacturing Engine/component Research 2 2 Testing Maintenance 1 1 Design/manufacturing Airframer Other Subtotal 4 5 9 The shares held by suppliers have been increasing over time. Table 5.9 reports on the financial split of some of the most recent new engine programmes. On the one hand, engine makers want to share risks and revenues across several suppliers to reduce their own stakes and to gain customers (i.e. airlines) via the involvement of suppliers of the same nationality as the customers. According to some industry experts, airlines (usually state-funded) are more likely to place engine orders when national component suppliers have been involved in the engine programme. On the other hand, component suppliers,

93 especially from developing countries, push to get bigger shares of engine programmes in order to be able to learn about more engine parts. Table 5.9. Financial breakdown of new engine programmes. Figures relate to the financial stake (share) maintained by companies and RRSPs in each programme (Source: author’s elaboration on Jane’s Aeroengine data) Company A RRSP RRSP RRSP RRSP RRSP 59 27 10 4 Engine 1 100 Engine 2 61 27 7 5 Engine 3 71 10 9 5 5 Engine 4 61 20 9 5 5 Engine 5 59.09 25.25 7 8.66 Engine 6 50 50 Engine 7 Company Engine 1 Engine 2 Engine 3 Engine 4 Engine 5

B 100 76 70.8 79 66

RRSP

RRSP

Company Engine 1 Engine 2

C 91 84

RRSP

RRSP

3.5

5

12.5 21.2 3 3

13

RRSP

RRSP

4 2 2 RRSP 5 2.5

RRSP

RRSP

9 4

RRSP

RRSP

4 4

3 3

2.5 1 1 RRSP 4

10 10 RRSP 5

3. The changing pattern of division of labour The previous section illustrated the driving forces at work in the aircraft engine industry. The aim of this section is to present a preliminary assessment of the impact of these driving forces on the pattern of division of labour in the aircraft engine industry. Table 5.10 reports on the shares of in-house design and manufacturing of engine makers and RRSPs. In the case of engine makers the shares refer to the entire engine by design and manufacturing tasks performed. As regards RRSP companies, the shares refer to the specific part of engine in which they specialise. As a general result, the shares of in-house design and manufacturing decrease over time reflecting an overall trend towards a greater externalisation of both design and manufacturing tasks. As regards engine makers, the shares of the design responsibility ranged from 100% to 70% in 1986 and from 80% to 50% in 1996. Manufacture is less ‘internal’ since its shares were about 50% on average ten years ago, whereas they now range from 40% to less than 10%. According to Table 5.10, engine makers reduced their manufacturing responsibility by half or by at least one third. Company A is an exception as its responsibility decreased by nearly ten times, from 50% to 5-10% during the observed time period. This trend towards greater externalisation holds true also for RRSP companies. Their design responsibility shares range from 100% to 50% in the mid-‘80s, and from 75% to 20% in the mid-‘90s. Likewise, the shares related to manufacturing dropped from 90% on average in 1986 to 70% in 1996. In the case of RRSP C, manufacturing dived from

94 80% to 30%. The decreasing share of design and especially manufacturing responsibility affecting both engine makers and RRSPs reflect the increased ability of suppliers (first- and second-tier) to take on design and manufacturing responsibility. Engine makers and RRSPs take advantage of more capable suppliers to subcontract larger parts of the engine and pass on to them the burden of some fixed costs. Therefore, there is increased specialisation of the supplier base leading to a greater division of labour across the different levels of the industry supply chain. Table 5.10. Shares of in-house design and manufacturing of aircraft engine companies. (Source: author’s elaboration on interview data) Company Design Manufacturing Engine manufacturers 1986 1996 1986 1996 Company A 70% 50% 50% 5-10% Company B 85% 80% 50% 25% Company C 100% 75% 60% 30% Company D 95% 80% 60% 40% Company E 90% 70% 50% 30% RRSPs RRSP C RRSP D RRSP E

1986 50% 90% 100%

1996 20% 75% 75%

1986 80% 90% 100%

1996 30% 75% 65%

Table 5.10 does not tell the whole story, however. It is worth stressing that as far as engine makers are concerned: (a) There is a gap between design and manufacturing shares, reflecting the fact that engine makers (but also RRSPs) subcontract a larger proportion of manufacturing than design. In other words, they outsource the production of engine components and subsystems whose design has been performed in-house. This practice is also known in the industry as ‘make-to-print’ 60. This first result reveals that choices about outsourcing are not simple and binary (make or buy). Sections 4 and 5 in this chapter and the following chapters further investigate this issue to unearth the intermediate types of technological capabilities (lying between full integration, the pure ‘make’ option, and full disintegration, the pure ‘buy’ option) that engine manufacturers maintain in-house. In addition, Section 4 (in this chapter) analyses engine makers’ technology bases as reflected in US patent statistics to assess how the continuous outsourcing of component production has impacted on their manufacturing capabilities. (b) The shares in Table 5.10 refer to the entire engine system. The interviewees clearly stated that the design and sometimes manufacture of the most critical engine components is jealously kept in-house by engine makers. Chapters 6 presents the case study of the fan, considered one the most critical engine subsystems, to discuss this issue further.

60

RRSP C represents an example of a ‘make-to-print’ company, since its manufacturing shares are always higher than the design ones, suggesting that it manufactures according to the drawings of engine manufacturers (Interview, 1997).

95 (c) The shares related to engine manufacturers’ design responsibility refer to overall design activities, as they do not make a distinction between concept and detailed design. Within their design activities, engine manufacturers focus more on concept design, leaving detailed design to RRSPs and suppliers (Interviews, 1997, 1999). As argued by one company engineer “It is in the advanced design project [i.e. concept design] that you get the most clever stuff, such as how you can scale down this new version without impairing engine performance? This is and will remain 100% our responsibility” (Interview, 1998). This issue will be discussed in Chapter 6, which documents the case of the so-called all-engine activities (e.g. concept design, performance), and in Chapter 7, which presents an in-depth case study of the engine control system. (d) The shares cut across all new engine development programmes in the two time periods analysed. In other words, the degree of involvement of engine manufacturers in engine programmes and within the same programmes changes over time due to resource (time and financial) constraints (Interviews, 1997, 1999). The role of collaborative agreements for development and research purposes will be analysed in Sections 4 and 5 in this chapter, Chapters 7 and 8 (in relation to systems integration capabilities). 4. The patent analysis As explained in Chapter 4, the aircraft engine is composed of a large number of components from different technological fields. As a result, the integration of the engine requires that systems integrators span their capabilities over a wide range of technological fields. On the other hand, Section 2 illustrated that a set of forces is at work in the industry that enables and pushes engine manufacturers to spin off larger parts of their technological capabilities and focus on just a few of them. This section aims to illustrate the result of this tension through patent analysis. US patents are used to build the technological profiles of these companies. The method used to create the sector-specific map on which each company’s technological profile is built was described in Chapter 3. The analysis is presented in different steps according to different levels of aggregation. This section is a rough description of the technological profiles of the companies examined using two technological maps (7- and 23-technological fields). These two technological maps summarise the general characteristics of engine makers’ technological profiles. Section 5 analyses the boundaries of the breadth and depth of engine makers’ technological capabilities by performing a finer-grade assessment of the companies’ technological profiles in terms of changes in the number of fields covered over time (entry-exit analysis). This is complemented by an investigation of companies’ technology diversification strategy using the Herfindal index. A more detailed map (58-technological field) is used to perform these analyses. Correlation analysis is then used to assess how companies’ technological profiles relate to each other. Building on Cantwell (1989, 1993), Section 5.3 analyses the stability and variation in companies’ technological profiles. 4.1. The Big Three A macro overview Tables 5.11-5.12 report on the technological profiles of the Big Three. Their patenting activity is presented over the 1977-1996 period. The period has been divided into four

96 five-year periods. Table 5.11 reports on the absolute patent number in the identified technological fields. To capture variations in the relative importance over time of the technological fields, the number of patents granted to a company in a given technological field is divided by the total number of patents assigned in the time period in question. Table 5.12 reports on the relative patent shares (RPSs) in the identified technological fields in each period. Descriptive statistics of each technological profile are summarised for comparison. Table 5.11. Big Three’s technological profiles on the 7-technological fields: total patent number 1977-1996. (Source: author’s elaboration on US patent data). Company A Sub-technological fields 7781 8286 8791 9296 Subtot Av StDv Materials 26 15 61 106 208 52 41 Manufacturing 74 56 184 226 540 135 83 Testing 9 18 63 78 168 42 33.7 Product-inner core 119 76 130 455 780 195 175 Product outer core 63 20 45 152 280 70 57.4 New architectures 1 3 16 16 36 9 8.1 Other 5 1 12 17 35 8.7 7.1 Subtotal 297 189 511 1050 2047 512 383 Company B Sub-technological fields Materials Manufacturing Testing Product-inner core Product outer core New architectures Other Subtotal

7781 25 122 18 137 50 1 19 372

8286 36 135 24 212 51 10 16 484

8791 45 126 25 302 82 22 40 642

9296 36 177 15 317 94 10 39 688

Subtot 142 560 82 968 277 43 114 2186

Av 35.5 140 20.5 242 69.3 10.8 28.5 547

StDv 8.2 25.3 4.8 84 22.2 8.6 12.8 145

Company C Sub-technological fields Materials Manufacturing Testing Product-inner core Product outer core New architectures Other Subtotal

7781 9 48 11 84 33 2 7 194

8286 7 53 9 120 49 1 5 244

8791 2 46 14 90 52 11 6 221

9296 6 52 8 121 45 4 7 243

Subtot 24 199 42 415 179 18 25 902

Av 6 49.8 10.5 104 44.8 4.5 6.2 226

StDv 2.94 3.3 2.6 19.5 8.3 4.5 0.9 23.5

Companies’ technological profiles are broken down according to seven main technological fields, that is to say ‘Other’, ‘Product Inner-Core’, ‘Product Outer-Core’, ‘Testing’, ‘Materials’, ‘Manufacturing’, ‘New Architectures’. This 7-technological field map is the result of a series of aggregations based on the technological fields derived from the US patent classification as described in detail in Chapter 3. Suffice to say here, that the product-related technological fields have been divided into ‘Product

97 Inner-Core’ and ‘Product Outer-Core’ according to the impact of the underlying technologies on the engine system’s performance. Table 5.12. Big Three’s technological profiles on the 7-technological fields: RPS 19771996. (Source: author’s elaboration on US patent data). Company A Sub-technological fields 7781 8286 8791 9296 Av StDv 7796 Materials 8.8 7.9 11.9 10.1 9.7 1.7 10.2 Manufacturing 24.9 29.6 36 21.5 28 6.3 26.4 Testing 3 9.5 12.3 7.4 8.1 3.9 8.2 Product-inner core 40.1 40.2 25.4 43.3 37.3 8 38.1 Product outer core 21.2 10.6 8.8 14.5 13.8 5.5 13.7 New architectures 0.3 1.6 3.1 1.5 1.6 1.1 1.8 Other 1.7 0.5 2.3 1.6 1.5 0.8 1.7 Subtotal 100 100 100 100 100 Company B Sub-technological fields Materials Manufacturing Testing Product-inner core Product outer core New architectures Other Subtotal

7781 6.7 32.8 4.8 36.8 13.4 0.3 5.1 100

8286 7.4 27.9 5 43.8 10.5 2.1 3.3 100

8791 7 19.6 3.9 47 12.8 3.4 6.2 100

9296 5.2 25.7 2.2 46.1 13.7 1.5 5.7 100

Av 6.6 26.5 4 43.4 12.6 1.8 5.1 0

StDv 1 5.5 1.3 4.6 1.4 1.3 1.3 0

7796 6.5 25.6 3.8 44.3 12.7 2 5.2 100

Company C Sub-technological fields Materials Manufacturing Testing Product-inner core Product outer core New architectures Other Subtotal

7781 4.6 24.7 5.7 43.3 17 1 3.6 100

8286 2.9 21.7 3.7 49.2 20.1 0.4 2 100

8791 0.9 20.8 6.3 40.7 23.5 5 2.7 100

9296 2.5 21.4 3.3 49.8 18.5 1.6 2.9 100

Av 2.7 22.2 4.7 45.7 19.8 2 2.8 0

StDv 1.5 1.8 1.5 4.4 2.8 2 0.6 0

7796 2.7 22.1 4.7 46 19.8 2 2.8 100

In general, the distribution of patents presents an upward trend for the firms examined over the period in question. Company A shows the most impressive growth. The number of US patents taken out in the last period is twice as many as was granted in the third period. The largest number of patents granted to the companies in question falls in the product-related technological fields. Specifically, ‘Product Inner-Core’ is on average the most important technological field followed by ‘Manufacturing’ and ‘Product Outer-Core’. ‘Product Inner-Core’ accounts on average for about 40% in the technological profile of each company over the period in question. Patent shares concerning ‘Product Outer-Core’ range from 7% to 20% and on average account for about 15% in the Big Three’s technological profile. Moreover, it is worth underlining that in all companies the RPS for ‘Manufacturing’ is never smaller than 20% throughout

98 the period in question. ‘Testing’ and ‘Materials’ follow different patterns between 1977 and 1996 and their importance in terms of RPS varies greatly across the different companies’ technological profiles. A more detailed picture A more detailed view of the technological profiles of the companies in question is presented in Tables 5.13-5.15. The map used is that encompassing 23 technological fields. Although this level of analysis is still rather broad, it summarises concisely some important characteristics of companies’ technological capabilities. Company patenting activity is presented in terms of RPSs over the 1977-1996 period, in turn, divided into four five-year periods. Descriptive statistics are also summarised. Considering the companies’ technological profiles over a twenty-year time period enables the dynamics of the boundaries of companies’ technological capabilities to be observed.

Table 5.13. Company A’s technological profile on 23-technological fields: RPSs 19771996. (Source: author’s elaboration on US patent data). Sub-technological fields 7781 8286 8791 9296 Av StDv 7796 8.8 7.9 11.9 10.1 9.7 1.7 10.2 Materials 5.1 2.1 2 1.8 2.7 1.5 2.3 Coating and chemical processes and apparatus 14.8 8.5 12.9 9.3 11.4 3 10.9 Metal working and metal treatment proc. 0.7 1.6 3.1 2.3 1.9 1 2.2 Metal working and metal treatment equips. 2 7.9 6.1 2.1 4.5 3 3.6 Electrical machinery 1.7 1.1 0.4 0.7 0.9 0.6 0.8 Electrochemical apparatus and processes 0.7 8.5 11.5 5.3 6.5 4.6 6.5 Electronic and optics systems for manufacturing 3 9.5 12.3 7.4 8.1 3.9 8.2 Testing and inspection apparatus 11.4 12.2 7.2 8.7 9.9 2.3 9 Control systems 7.7 9.5 5.3 10.3 8.2 2.2 8.6 Rotor assembly 1.7 1.1 0.6 2.3 1.4 0.7 1.7 Stator assembly 2.7 1.1 2.3 2.6 2.2 0.8 2.4 Shafts and bearings 1.3 1.1 0.4 1.1 1 0.4 1 Casings 5.1 4.8 2.9 6 4.7 1.3 5 Sealings 5.1 6.3 5.3 7.8 6.1 1.3 6.6 Combustion chambers 8.1 1.6 2 2.7 3.6 3 3.2 Exhaust systems 2 0 0.2 0.7 0.7 0.9 0.7 Lubrication systems 0.3 0.5 0.2 0.9 0.5 0.3 0.6 Containment structures 2 2.6 2 4.7 2.8 1.3 3.4 Couplings and joints 5.1 4.2 1.4 4.6 3.8 1.7 3.8 Cooling systems 8.8 5.8 4.5 5.6 6.2 1.8 5.8 Aeronautics 0.3 1.6 3.1 1.5 1.6 1.1 1.8 New architectures 1.7 0.5 2.3 1.6 1.5 0.8 1.7 Other 100 100 100 100 100 0 100 Subtotal

99 The distribution of patents reported in Tables 5.13-5.15 shows that the firms examined can be considered to be multitechnology: the patenting activity between 1977 and 1996 covers numerous and highly varied technological fields. A closer look at the technological profiles of these three firms over the period under consideration indicates that their capabilities were not restricted to a spectrum of technologies encompassing only those fields that could be regarded as strictly core to this sector. Instead, these three firms were also active in technologies that could be dubbed as marginal. This exceptional array of technologies derives from the extensive range of technological knowledge necessary for the design, development, production, and integration of the aircraft engine (Mattingly et al., 1987). From this perspective, it is worth recalling that the aircraft engine is an extremely complicated product that can include as many as 40,000 components. As discussed in Chapter 4, these components may belong to numerous and frequently radically divergent technological paradigms. The high number of technologies involved in the design, development, production and integration of aircraft engines therefore requires in-depth knowledge of the entire system, embracing knowledge of the parts, of the interaction among the parts, and of the whole. Despite the increasing trend of externalisation of both design and manufacturing tasks discussed in Section 2, engine manufacturers maintain technological capabilities related to outsourced components. For instance, the patent analysis highlights that the firms in question take out patents, albeit with somewhat erratically in the various cases examined, in the technological field of ‘Aeronautics’. Patents falling into this technological field are concerned with the aircraft-engine interface (engine mounting and coupling systems, nacelles, and thrust reversers). This finding shows that despite the fact of having surrendered the design and the manufacture of these components to third parties, engine manufacturers do not discard the technological capabilities related to them (Chapter 4). Additionally, and more surprisingly, the broad spectrum of technologies identified includes not only product technologies but also process technologies. The Big Three show an exceptional interest in manufacturing technologies. Consider, for instance, the field defined as ‘Metal Working and Treatment Processes’, referring to patents in metal deforming and working methods and metal treatment processes. It actually represents one of the most (and in some periods thee most) important fields in terms of RPSs in each of the cases examined over the period 1977-1996. Therefore, although engine makers continuously delegate manufacturing tasks to suppliers, they still maintain inhouse capabilities in relation to manufacturing technologies in order to use and to innovate manufacturing equipment to fine tune it to their scope 61,62.

61

The protection of the manufacturing technologies exerted by the big engine makers has been confirmed also by our interviewees indicating that “Although the Big Three may continuously resort to external sources for make-to-print situations, they will retain a minimum threshold of technological capabilities [in manufacturing-related technologies]” (Interview, 1997). 62 It is worth noting, however, that another explanation underlies the increasing number of patents related to process technologies. Engine makers have started developing new manufacturing facilities in partnership with suppliers, providing financial and technological resources. They then retain the intellectual property rights on the outcome of the partnership.

100

Table 5.14. Company B’s technological profile on 23-technological fields: RPSs 19771996. (Source: author’s elaboration on US patent data). Sub-technological fields 7781 8286 8791 9296 Av StDv 7796 6.7 7.4 7 5.2 6.6 1 6.5 Materials 3.2 5.6 3 7.3 4.8 2 4.9 Coating and chemical processes and apparatus 17.5 12.4 9 11.2 12.5 3.6 11.9 Metal working and metal treatment proc. 1.3 3.1 1.7 1.5 1.9 0.8 1.9 Metal working and metal treatment equip. 1.9 1.9 1.1 0.6 1.4 0.6 1.2 Electrical machinery 4.3 1.9 2 4.4 3.1 1.4 3.1 Electrochemical apparatus and processes 4.6 3.1 2.8 0.9 2.8 1.5 2.6 Electronic and optics systems for manufacturing 4.8 5 3.9 2.2 4 1.3 3.8 Testing and inspection apparatus 10.8 12.4 16.4 9.4 12.2 3 12.4 Control systems 5.6 8.7 9.2 10.5 8.5 2 8.9 Rotor assembly 2.4 1.4 3.3 4.4 2.9 1.2 3.1 Stator assembly 3.2 5.4 4.2 2.9 3.9 1.1 3.9 Shafts and bearings 0.3 1.2 0.9 1.5 1 0.5 1.1 Casings 4 6.2 4.5 5.1 5 0.9 5 Sealings 7.3 5.4 4.7 8.3 6.4 1.7 6.4 Combustion chambers 8.3 2.7 6.5 6.4 6 2.4 5.9 Exhaust systems 0.3 0.6 1.1 0.4 0.6 0.4 0.6 Lubrication systems 0 1.2 0.3 0.9 0.6 0.6 0.6 Containment structures 1.3 2.3 3 2.9 2.4 0.8 2.5 Couplings and joints 3.2 3.1 3.9 4.1 3.6 0.5 3.7 Cooling systems 3.5 3.7 1.9 3.1 3 0.8 2.9 Aeronautics 0.3 2.1 3.4 1.5 1.8 1.3 2 New architectures 5.1 3.3 6.2 5.7 5.1 1.3 5.2 Other 100 100 100 100 100 0 100 Subtotal Furthermore, not only do engine makers control the manufacturing technologies of contracted out components, but they also maintain in-house manufacturing-related capabilities for those components deemed to be critical. Chapter 6 describes the case of a radically new fan blade design, namely the wide-chord blade, showing that RollsRoyce developed in-house new manufacturing processes (i.e. superplastic forming and diffusion bonding techniques) to produce the new blade concept. This leads in turn to another point: in aircraft engine innovation in product technologies calls for innovation in process technologies.

101

Table 5.15. Company C’s technological profile on 23-technological fields: RPSs 19771996. (Source: author’s elaboration on US patent data). Sub-technological fields 7781 8286 8791 9296 Av StDv 7796 4.6 2.9 0.9 2.5 2.7 1.5 2.7 Materials 5.2 2 1.8 2.1 2.8 1.6 2.7 Coating and chemical processes and apparatus 4.1 13.5 11.3 12.8 10.4 4.3 10.8 Metal working and metal treatment proc. 0.5 0.4 3.2 2.5 1.6 1.4 1.7 Metal working and metal treatment equip. 5.2 0 2.3 2.5 2.5 2.1 2.3 Electrical machinery 2.6 2 0.9 1.6 1.8 0.7 1.8 Electrochemical apparatus and processes 7.2 3.7 1.4 0 3.1 3.2 2.9 Electronic and optics systems for manufacturing 5.7 3.7 6.3 3.3 4.7 1.5 4.7 Testing and inspection apparatus 5.7 7.4 11.3 9.9 8.6 2.5 8.6 Control systems 14.4 8.6 7.7 9.5 10 3 9.9 Rotor assembly 1.5 0.8 5 2.9 2.6 1.8 2.5 Stator assembly 5.2 4.9 1.4 0.8 3.1 2.3 3 Shafts and bearings 0.5 2 1.4 1.2 1.3 0.6 1.3 Casings 3.6 10.7 5 10.3 7.4 3.6 7.6 Sealings 8.2 12.3 5.4 10.3 9.1 2.9 9.2 Combustion chambers 3.6 7.4 9.5 4.9 6.4 2.6 6.4 Exhaust systems 0.5 3.3 0.5 1.6 1.5 1.3 1.6 Lubrication systems 0.5 0 0.5 1.6 0.7 0.7 0.7 Containment structures 1 1.2 2.7 1.6 1.7 0.8 1.7 Couplings and joints 4.1 2.5 3.6 4.9 3.8 1 3.8 Cooling systems 11.3 8.2 10.4 8.6 9.6 1.5 9.5 Aeronautics 1 0.4 5 1.6 2 2 2 New architectures 3.6 2 2.7 2.9 2.8 0.6 2.8 Other 100 100 100 100 100 0 100 Subtotal It is also worth noting the long-standing and ‘deep’ interest of the Big Three in ‘Measuring and Testing’ technologies. This technological field shows high RPSs in all three companies. In this technological field patents related to those technologies used during the testing phase of new engine components are clustered, for instance X-ray system devices, non-intrusive analysis, image analysis, and related data processing systems. As discussed in Chapter 4, engine makers perform, in-house, several and different tests to assess the feasibility of new engine designs. These testing activities encompass both those demanded by the certification authorities and those carried out by engine makers to provide evidence to the customers on particular engine performance, such as reliability. The comparative patent analysis discussed above shows that the capabilities of engine manufacturers span a broad range of technological fields, encompassing both product and process technologies. On the basis of this result, we can attempt a preliminary

102 assessment of the impact of the enablers and pushers discussed in Section 2 on the boundaries of firms’ technological capabilities. First, despite outsourcing design and manufacturing tasks to external suppliers, the integration of the engine requires that manufacturers maintain in-house deep technological knowledge of the engine system in terms of its parts, the interactions between them, and the whole. Engine manufacturers must be knowledgeable in a variety of technological fields to be able to specify, buy, and eventually integrate components into the engine. Their in-house technological capabilities enable them also to co-ordinate and benefit from changes in outsourced components 63. This result squares with and supports Granstrand et al. (1997) and Patel and Pavitt (1997) analyses of multitechnology firms. Second, engine manufacturers enter collaborative agreements for two main purposes (a) development of new engines and (b) acquisition of new technologies. As regards (a), it is argued that engine manufacturers must be knowledgeable about a variety of technological fields in order to delegate design and manufacturing tasks to suppliers and eventually integrate components back into the engine. Engine manufacturers see development programmes as learning opportunities to maintain their systems integration capabilities over time 64 (Interviews, 1997, 1999). As regards (b), collaborative agreements are the means by which engine manufacturers can tap into the technological knowledge of suppliers, universities, and research centres for the development of new technologies. Collaborative agreements are decentralised modes of acquiring new knowledge. They constitute sources of diversity and learning (Cohendet et al., 1998). However, in order to take full advantage of collaborative relations, firms need to be equipped with an adequate and independent set of technological capabilities (Colombo and Garrone, 1999; Mowery, 1983). In this light, collaborative agreements should be considered as a complement to engine manufacturers’ technology bases. Thus, it can be argued that engine manufacturers perceive collaborative agreements as learning mechanisms to maintain their systems integrator status (homeostatic mechanisms). Chapters 7 and 8 discuss further the role of collaborative agreements in relation to systems integration capabilities. Third, the patent analysis above supports the view that in order to modularise engines, manufacturers require in-depth technological knowledge of the entire engine system, including knowledge of the parts and of the interactions among the parts (Baldwin and Clark, 1997; Parnas, 1972). A note on some relevant differences The foregoing analysis has illustrated the similarities displayed by the technological profiles of the Big Three. Examination of their technological profiles also highlights some differences across companies reflecting different technological effort towards some technological fields that may indicate correspondingly a particular strength or weakness. For instance, Company A is widely recognised as a leader in the industry as regards material technologies and the thesis results confirm this view. Company A 63

Following Granstrand et al. (1997), co-ordination is primarily understood as technological coordination (i.e. co-ordination of technological change in the supply chain). 64 As underlined by a Technical Director from one of the Big Three “We enter [lead or take part into] new engine development programmes to refresh and nurture our systems integration capabilities: we remember by doing” (Interview, 1999).

103 displays higher RPSs in material-related technologies throughout the period under investigation as than its two competitors. As a consequence, ‘Materials’ counts on average for about 10% as compared with 2% and 5% in Company B and C, respectively. It is also worth noting that Company A is the only engine maker that has introduced a hybrid metal-composite fan blade. The analysis also provides a vantage-point from which to identify cases that are worth investigating further. For example, the case of manufacturing capabilities in relation to critical components. This point will be furthered analysed in Chapter 6 in relation to the case of the Fan Key System developed in-house by Rolls-Royce. The ‘Control System’ technological field represents another interesting case. The three companies display comparable RPSs in relation to this technological field. This is a surprising result since only two of the companies, Companies A and B, have internal suppliers that design, develop, manufacture, and in some cases sell (also to competitors), engine control systems. Company C instead has always resorted to external suppliers. Despite that, Company C shows comparable RPSs in ‘Control Systems’ between 1982 and 1996. This result finds its rationale in the growing importance of control system technologies for the ultimate performance of the engine. Chapter 7 explains how the control system has become a critical engine component due to a shift in its underlying technologies. The case of Company C will be analysed in detail to understand the change in terms of the breadth and depth of its technological capabilities in ‘Control Systems’ as a result of this technological shift. 4.2. The RRSP companies A macro overview Tables 5.16-5.17 summarise the technological profiles of the two RRSPs between 1977 and 1996 broken down according to the 7 technological fields. Companies’ patenting activity is presented in terms of absolute number and RPS. Descriptive statistics are also summarised for comparison.

The two RRSPs exhibit two completely different patterns in terms of patents granted during the period under consideration and the corresponding RPSs. RRSP A shows impressive growth between 1977 and 1996. The number of patents granted to this company in the final period is five times as large as that in the first. On average the percentage shares related to the engine inner-core rank highest. ‘Manufacturing’ displays an upward trend, while ‘Testing’ shows a rather erratic pattern and ‘Materials’ plays a relatively minor role. As discussed later in this chapter, RRSP A pursues a technology strategy of full systems integration. The present analysis reveals that RRSP A has deepened its capabilities in those technologies associated with manufacturing. The growth in this technological field points to an interesting result, that is to say manufacturing-related technologies play an important role in the development of the technological profile of a would-be systems integrator. This supports the point made in the previous section on the close link between innovations in product technologies and innovations in production technologies. According to Tables 5.16-5.17, RRSP B is following a different technology strategy. It shows a dramatic fall in terms of total number of patents in the fourth period. As regards its RPSs, ‘Product inner-core’ is on average the most important followed by ‘Manufacturing’.

104

Table 5.16. RRSP A’s technological profiles on the 7-technological fields: total patent number 1977-1996. (Source: author’s elaboration on US patent data). RRSP A Sub-technological fields 7781 8286 8791 9296 Subtot Av StDv Materials 0 1 1 4 6 1.5 1.7 Manufacturing 7 12 31 59 109 27.3 23.6 Testing 2 7 5 18 32 8 6.9 Product-inner core 36 68 86 118 308 77 34.3 Product outer core 8 26 36 37 107 26.8 13.5 New architectures 0 0 1 2 3 0.7 0.9 Other 7 6 13 17 43 10.8 5.2 Subtotal 60 120 173 255 608 152 82.7 RRSP B Sub-technological fields Materials Manufacturing Testing Product-inner core Product outer core New architectures Other Subtotal

7781 0 16 7 38 8 0 4 73

8286 3 27 1 41 6 0 4 82

8791 8 28 6 77 11 3 13 146

9296 9 32 4 33 7 4 7 96

Subtot 20 103 18 189 32 7 28 397

Av 5 25.8 4.5 47.3 8 1.7 7 99.3

StDv 4.2 6.8 2.6 20.1 2.1 2 4.2 32.6

Table 5.17. RRSP B’s technological profiles on the 7-technological fields: RPS 19771996. (Source: author’s elaboration on US patent data). RRSP A Sub-technological fields 7781 8286 8791 9296 Av StDv 7796 Materials 0 0.8 0.6 1.6 0.7 0.7 1 Manufacturing 11.7 10 17.9 23.1 15.7 6 17.9 Testing 3.3 5.8 2.9 7.1 4.8 2 5.3 Product-inner core 60 56.7 49.7 46.3 53.2 6.3 50.7 Product outer core 13.3 21.7 20.8 14.5 17.6 4.3 17.6 New architectures 0 0 0.6 0.8 0.3 0.4 0.5 Other 11.7 5 7.5 6.7 7.7 2.8 7.1 Subtotal 100 100 100 100 100 0 100 RRSP B Sub-technological fields Materials Manufacturing Testing Product-inner core Product outer core New architectures Other Subtotal

7781 0 21.9 9.6 52.1 11 0 5.5 100

8286 3.7 32.9 1.2 50 7.3 0 4.9 100

8791 5.5 19.2 4.1 52.7 7.5 2.1 8.9 100

9296 9.4 33.3 4.2 34.4 7.3 4.2 7.3 100

Av 4.6 26.8 4.8 47.3 8.3 1.6 6.6 100

StDv 3.9 7.4 3.5 8.7 1.8 2 1.8 0

7796 5 25.9 4.5 47.6 8.1 1.8 7.1 100

105 A closer look at the table reveals that in the final period ‘Manufacturing’ displays a marked rise, whereas ‘Product inner-core’ a dive. ‘Materials’ also displays an escalating growth of RPSs, while patent shares in ‘Testing’ after decreasing in the first two periods show a constant pattern. These results may well be due to the stated strategy of RRSP B to maintain its status of mere supplier of engine components and subsystems for large systems integrator companies. In contrast to the previous case, RRSP B is focusing down its technology strategy. A more detailed picture Tables 5.18-5.19 report on the technological profiles of the two RRSPs according to the 23-technological field map. The previous section argued that RRSP A represents an interesting case for analysis because of its pursued strategy of full systems integration reflected in its technological profile.

Table 5.18. RRSP A’s technological profile on 23-technological fields: RPSs 1977-1996. (Source: author’s elaboration on US patent data). Sub-technological fields 7781 8286 8791 9296 Av StDv 7796 0 0.8 0.6 1.6 0.7 0.7 1 Materials 0 2.5 2.3 1.2 1.5 1.2 1.6 Coating and chemical processes and apparatus 10 4.2 8.1 11 8.3 3 8.7 Metal working and metal treatment proc. 0 2.5 4.6 3.5 2.7 2 3.3 Metal working and metal treatment equip. 1.7 0.8 0.6 3.5 1.7 1.3 2 Electrical machinery 0 0 1.2 3.5 1.2 1.7 1.8 Electrochemical apparatus and processes 0 0 1.2 0.4 0.4 0.5 0.5 Electronic and optics systems for manufacturing 3.3 5.8 2.9 7.1 4.8 2 5.3 Testing and inspection apparatus 10 18.3 9.2 8.6 11.6 4.6 10.9 Control systems 13.3 19.2 11.6 9.8 13.5 4.1 12.5 Rotor assembly 0 0.8 5.2 8.2 3.6 3.9 5.1 Stator assembly 6.7 4.2 3.5 0.8 3.8 2.4 2.8 Shafts and bearings 3.3 0.8 2.9 0.8 2 1.3 1.6 Casings 11.7 9.2 9.2 3.1 8.3 3.6 6.9 Sealings 13.3 2.5 6.4 12.5 8.7 5.2 8.9 Combustion chambers 3.3 5.8 5.8 3.5 4.6 1.4 4.6 Exhaust systems 0 3.3 2.3 1.6 1.8 1.4 2 Lubrication systems 0 1.7 0 0 0.4 0.8 0.3 Containment structures 3.3 4.2 1.2 2.7 2.9 1.3 2.6 Couplings and joints 1.7 1.7 1.7 2.4 1.9 0.3 2 Cooling systems 6.7 6.7 11.6 6.7 7.9 2.4 8.1 Aeronautics 0 0 0.6 0.8 0.3 0.4 0.5 New architectures 11.7 5 7.5 6.7 7.7 2.8 7.1 Other 100 100 100 100 100 0 100 Subtotal

106 The topmost technological fields in terms of cumulative patent shares are ‘Rotor assembly’, ‘Combustion Chambers’, ‘Control Systems’, ‘Metal Working and Metal Treatment Processes’, and ‘Aeronautics’. The most impressive and surely interesting expansion is that registered by ‘Combustion Chambers’. After reducing its efforts in the first period, RRSP A has deepened its capabilities in combustion technologies over the next three periods. It is worth noting that RRSP A has always been responsible for the so-called cold (or low pressure) parts in the engine programmes in which it has been involved. Recently, it has attempted to re-negotiate its stake in one of these engine programmes to take on also the hot (or high pressure) parts with uncertain results (Interview, 1998). This is reflected in the patent analysis. Although some of the patents falling in this technological field may be related to military technologies, this growing concern towards combustion technologies is an indication of RRSP A’s effort to become a full systems integrator. This result was confirmed by the company engineers interviewed (Interview, 1998). Section 5.3 further investigates these results by testing statistically the variation of RRSP A’s technological profile. Table 5.19. RRSP B’s technological profile on 23-technological fields: RPSs 1977-1996. (Source: author’s elaboration on US patent data). Sub-technological fields 7781 8286 8791 9296 Av StDv 7796 0 3.7 5.5 9.4 4.6 3.9 5 Materials 2.7 8.5 2.7 13.5 6.9 5.2 6.5 Coating and chemical processes and apparatus 11 14.6 12.3 14.6 13.1 1.8 13.1 Metal working and metal treatment proc. 2.7 2.4 1.4 2.1 2.2 0.6 2 Metal working and metal treatment equip. 1.4 2.4 0.7 1 1.4 0.8 1.3 Electrical machinery 4.1 4.9 2.1 1 3 1.8 2.8 Electrochemical apparatus and processes 0 0 0 1 0.3 0.5 0.3 Electronic and optics systems for manufacturing 9.6 1.2 4.1 4.2 4.8 3.5 4.5 Testing and inspection apparatus 12.3 2.4 4.1 1 5 5.1 4.5 Control systems 11 11 11 7.3 10 1.8 10.1 Rotor assembly 5.5 1.2 4.8 4.2 3.9 1.9 4 Stator assembly 4.1 3.7 2.7 2.1 3.1 0.9 3 Shafts and bearings 2.7 1.2 2.1 0 1.5 1.2 1.5 Casings 2.7 19.5 9.6 5.2 9.3 7.4 9.3 Sealings 11 1.2 2.7 7.3 5.6 4.4 5 Combustion chambers 4.1 0 0.7 4.2 2.2 2.2 2 Exhaust systems 2.7 2.4 2.1 0 1.8 1.2 1.8 Lubrication systems 0 0 1.4 0 0.3 0.7 0.5 Containment structures 2.7 1.2 1.4 1 1.6 0.8 1.5 Couplings and joints 2.7 9.8 15.8 7.3 8.9 5.4 10.1 Cooling systems 1.4 3.7 2.1 2.1 2.3 1 2.3 Aeronautics 0 0 2.1 4.2 1.6 2 1.8 New architectures 5.5 4.9 8.9 7.3 6.6 1.8 7.1 Other 100 100 100 100 100 0 100 Subtotal

107

The case of RRSP A also sheds light on the role of collaborative agreements as learning mechanisms. RRSP A has been involved in new engine development programmes since the 1970s. Its involvement in terms of design and manufacturing responsibility of engine parts has continually expanded. RRSP A has perceived new engine development programmes as learning mechanisms. That is to say, development programmes were used not only as a means to acquire new technological capabilities, but also as a means to change its status of supplier into that of full systems integrator. As compared to engine manufacturers that use engine development programmes to retain their status of systems integrators (homeostatic mechanisms), RRSPs can employ them to acquire systems integration capability (i.e. to change their status of suppliers into full systems integrators, heterostatic mechanisms). As mentioned in the previous section, RRSP B has followed a different strategic route to RRSP A. The technological fields that rank highest are ‘Metal Working and Metal Treatment Processes’, ‘Rotor Assembly’, ‘Cooling Systems’, ‘Sealings’, and ‘Coating and Chemical Processes’. It is worth noting that the two manufacturing-related technological fields are the only fields in the top five that register an upward trend in the last period in terms of RPSs. Unlike the previous case, according to our data RRSP B has been strengthening its technological efforts in production technologies, while reducing them in product-related ones. This result finds its rationale in the fact that RRSP B pursues a supplier technology strategy. This point is further scrutinised in Section 5.3. 5. Stability, persistency, and variation of aircraft engine makers’ technological profiles The previous section presented a first, coarse account of the boundaries of engine makers’ technological capabilities using two synthetic sector-specific technological maps (7 and 23 fields). It showed that the technological profiles of engine makers are not focused on the few technological fields related to the inner-core of the engine. Rather, engine makers are also active in other technological fields, such as testing and manufacturing. The purpose of this section is to go a step further and provide a more in-depth analysis of the breadth and depth of engine makers’ technological capabilities, using a more detailed technological map (58-technological fields) to assess three main points: (a) Whether and how the internal structures of companies’ technological profile are related. Correlation analysis is employed to assess such relationship. On the basis of the discussions conducted in the earlier sections, a positive relationship amongst companies within each time period is expected. Furthermore, the analysis of the size of the coefficients in the two periods examined enables us also to comment on changes that might have occurred in the relationships between companies’ technological profiles reflecting changes in their technology strategies. (b) The number of technological fields covered by the companies over the four periods examined in order to understand whether companies are ‘broadening out’ their technological capabilities. This entry-exit analysis will be complemented by using Herfindal indices. (c) The degree of stability of companies’ technological profile, using Cantwell’s (1989, 1993) methodology to analyse variations in such profiles over time.

108 5.1. The relationships across companies’ technological profiles: what correlation analysis suggests The correlation analyses have been separately run in two ten-year time periods, namely 1977-86 and 1987-96, using the SPSS software package. For each period the correlation coefficients give an idea of how closely related (or unrelated) are the internal structures of company technological profiles. A positive outcome indicates that companies’ technological profiles ‘move together’, whereas a negative one indicates the reverse. The magnitude of the correlation coefficients reveals the strength of these relationships, where the higher magnitude indicates a stronger relationship.

Tables 5.20 and 5.21 show the correlation coefficients calculated using the 58technological field map in 1977-86 and 1987-96 respectively. As expected, the coefficients are all positive indicating that within the same industrial sector companies’ technological profiles are related. This is true in both the periods examined. The strength of the relationships varies between 0.270 and 0.769 in the first period and between 0.272 and 0.672 in the second. In the first period the correlation coefficients are significant at the 0.01 level, with the exception of that between Company A and RRSP B, which is significant at the 0.05 level. In the second period all coefficients are significant at the 0.01 level, with the exception of those between Company A and the two RRSPs that are significant at the 0.005 level. The magnitude of the correlation coefficient should identify those companies with similar technological missions. Consequently, any variation in its size would indicate changes in a company’s technological mission. Such variations should be understood as deriving from changes in the relative importance of the ranking of the technological fields. Tables 5.20 and 5.21 highlight this interesting link. In general, the majority of the correlation coefficients show a decrease in 1987-96 as compared with 1977-86. This first finding would suggest that these companies are moving in different directions as regards their technology strategy. A closer look at the tables reveals some interesting industry trends. Table 5.20. Correlation coefficients derived from companies’ technological profile in terms of RPSs in 1977-86. (Source: author’s elaboration on US patent data). Company Company A Company B Company C RRSP A RRSP B Company A

1.000

Company B

0.769**

1.000

Company C

0.617**

0.681**

1.000

RRSP A

0.475**

0.607**

0.521**

1.000

RRSP B

0.270*

0.532**

0.475**

0.502**

Note

1.000

** Correlation is significant at 0.01 level (2-tailed) * Correlation is significant at 0.05 level (2-tailed) Number of observations = 58

In 1977-86 the highest correlation coefficients are displayed by companies with similar technological missions. Thus, the big engine makers’ technological profiles correlate higher with each other than with RRSP companies. The only coefficient amongst the Big Three that has increased is that between Company B and C, suggesting that their

109 technological profiles are moving closer. Company A is moving further away from its competitors indicating a change in the relative importance of each technological field within its technological profiles. This is due to the increased importance in terms of RPS of such technologies as ‘Materials’, and ‘Testing and Inspection’ in this company’s technological profile, which instead display stable dynamics in the other companies’, as discussed in Section 4.1. Table 5.21. Correlation coefficients derived from companies’ technological profile in terms of RPSs in 1987-96. (Source: author’s elaboration on US patent data). Company Company A Company B Company C RRSP A RRSP B Company A

1.000

Company B

0.451**

1.000

Company C

0.411**

0.672**

1.000

RRSP A

0.327*

0.584**

0.603**

1.000

RRSP B

0.272*

0.520**

0.365**

0.353**

Note

1.000

** Correlation is significant at 0.01 level (2-tailed) * Correlation is significant at 0.05 level (2-tailed) Number of observations = 58

The coefficients of the two RRSPs display fuzzier dynamics. In the first period, the technological profile of RRSP A correlates more closely with the Big Three’s than with RRSP B. In the second period the correlation coefficient between the two RRSPs ranks low. The correlation coefficient of RRSP A follows a general decreasing pattern between the two periods. However, the correlation coefficient between RRSP A and Company C shows a sharp increase. This result squares with the description of the technological profiles discussed in Section 4.2. RRSP A’s full systems integration strategy is reflected in the correlation results since it is now moving closer to the Big Three’s in terms of the relative importance of each technological field. On the other hand, RRSP B’s technological profile is moving apart from the Big Three’s, confirming that it is striving to maintain its status of specialised supplier. 5.2. The breadth of technological capabilities Table 5.22 reports on the number of technological fields covered by the companies analysed over the four periods examined using the 58-technological field map. The most detailed technological map was used since the idea was to assess whether these companies are focusing or diversifying their technological capabilities at a finer-grade level. The technological fields are arranged according to the 7-technological field map.

The entry-exit analysis shows that the Big Three have increased the number of technologies covered between 1977 and 1996. A closer look at the table shows that the number of those technological fields related to the so-called inner-core displays a steady pattern for all the companies. Manufacturing-related technological fields represent a field of intense technological interest to these companies too. These results are consistent with the findings in Section 4. The expansion of companies’ patenting activity in the product- and manufacturing-related technologies is actually reflected and accompanied by an increasing number of technological fields covered. This outcome is in stark contrast with prescriptions deriving from simple notions of core capability,

110 according to which efforts should focus instead on a few capabilities (Prahalad and Hamel, 1990). Table 5.22. The distribution of engine makers and RRSPs technological capabilities according to the 7-technological field map (Source: author’s elaboration on US patent data). Company A 1977-81 1982-86 1987-91 1992-96 Materials 2 3 3 3 Manufacturing 11 11 12 12 Testing 1 2 2 2 Product-inner core 27 25 25 27 Product outer core 10 6 9 10 New architectures 1 1 2 2 Other 1 1 1 1 Subtotal 53 49 54 57 Company B Materials 3 3 3 3 Manufacturing 12 12 12 12 Testing 2 2 2 2 Product-inner core 23 26 25 28 Product outer core 5 9 10 10 New architectures 1 2 2 1 Other 1 1 1 1 Subtotal 47 55 55 57 Company C Materials 1 2 2 3 Manufacturing 10 10 10 11 Testing 2 1 2 2 Product-inner core 21 22 25 25 Product outer core 9 8 10 10 New architectures 1 1 2 2 Other 1 1 1 1 Subtotal 45 45 52 54 RRSP A Materials 0 1 1 2 Manufacturing 4 6 10 12 Testing 1 1 2 2 Product-inner core 16 17 20 21 Product outer core 5 7 8 9 New architectures 0 0 1 1 Other 1 1 1 1 Subtotal 27 33 43 48 RRSP B Materials 0 1 1 1 Manufacturing 9 10 11 10 Testing 1 1 1 1 Product-inner core 14 14 17 15 Product outer core 4 5 6 4 New architectures 0 0 1 1 Other 1 1 1 1 Subtotal 29 32 38 33

111

The technological profiles of the RRSPs are more focused. As regards RRSP A, the number of technological fields increases continuously throughout the period under consideration, to the extent that it has nearly doubled in the last period as compared to the first. It can be argued, therefore, that such a pattern mirrors the company’s stated strategy. RRSP B shows a different pattern. After a steady increase during the first three periods, the number of technological fields covered registers a nosedive in the last period. This contraction affects all the technological fields shown, with the exception of manufacturing. Such a sharp decrease may well reflect the company’s stated strategy of maintaining its role of supplier, as found in Section 4. As a matter of fact, in a recent government-funded programme for the next generation aircraft engine, the company has contributed to only the so-called low-pressure parts (Interview, 1998). From the foregoing discussion it can be argued that in considering the mix of technological capabilities of these firms in terms of the spectrum of technologies covered, the different technological mission of each company is clearly brought to light. Moreover, a declared change in a company’s product strategy is mirrored in their technological profile. Consequently, the technological profile of each of these firms appears to reflect their divergent nature. Thus, the patenting activity of the big systems integrators covers the broadest spectrum as compared to those of RRSPs. This difference is due to the more diverse and varied technological requirements that engine makers have to meet to develop and maintain over time an appropriate level of technological knowledge spanning the entire engine system. Likewise, the case of RRSP A has shown that in order to pursue a full systems integrator strategy, the company’s technological profile has branched out into more and different technological fields, possibly reaching for an industry-specific minimum threshold in terms of the spectrum of technological capabilities required to compete. The breadth of companies’ technological capabilities: The Herfindal Index The breadth of the technological capabilities of the companies analysed can also be assessed using concentration indices, notably the Herfindal index and the concentration ratio. The Herfindal index is one of the most commonly used measures of concentration. As is well known, it is defined as the sum over all classes of the squares of the shares of the variable. As regards this study, the Herfindal index for company j in one time period is:

Hj = ∑i=1, 58 (xij/Xj)2 where: i = technological field, j = company, x = number of patents granted to company j in the technological field i during the period in question, X = total number of patents granted to company j during the period in question. In this case there are 58 classes corresponding to the 58 identified technological fields. The Herfindal index then varies between 0.017 (i.e. 1/58, each technological field has the same share) corresponding to max diversification and 1 corresponding to max specialisation (share of the highest technological field is 1). Therefore, the lower the index, the greater the breadth of the technological capabilities of the company at issue.

112

Table 5.23 shows the Herfindal index for the companies under consideration. The degree of technological specialisation of the companies analysed as reflected in the Herfindal is very low. In other words, all companies are rather diversified in terms of the technologies listed in the 58-technological field map. As expected, the Big Three display the lowest Herfindal indices as compared to the RRSP companies. Although the dynamics of the Herfindal indices of the Big Three display a rather erratic pattern over time, the differences in each index across the time period are rather small and overall the index is very low throughout the period examined. These low values and hardly little change in the magnitude of the indices suggest that the companies are not pursuing a strategy of specialisation at the technological level. This confirms the results found in Section 4. RRSP A displays a decreasing Herfindal Index all along the period examined. This latter outcome supports the previous argument related to the full systems integration strategy pursued by this company 65. Table 5.23. Herfindal index distribution (minimum: 0,017). (Source: author’s elaboration on US patent data). Company 1977-81 1982-86 1987-91 1992-96 0.036 0.038 0.045 0.03 Company A 0.036 0.028 0.034 0.032 Company B 0.037 0.039 0.034 0.036 Company C 0.055 0.05 0.038 0.035 RRSP A 0.045 0.053 0.049 0.056 RRSP B

5.3. Stability and dynamics of change in the long term: some statistical evidence This section assesses statistically two propositions: (a) the ‘stability’ or ‘persistency’ over time of companies’ technological profiles in the industry in question and (b) the ‘variations’ in such profiles. The ‘stability’ or ‘persistence’ of a firm’s technological profile over time finds its place in the broader view that technological learning within business organisations follows a cumulative path. This learning process, in other words, is not random. This view also squares with Nelson’s (1991) point that firms in the same industry differ in that each firm has its own field of distinctive technological capabilities. These capabilities are firm-specific and the underlying learning processes are not readily imitable by other firms (Pavitt, 1990). In the context of this study, the argument of persistency relates to the idea that at the system integration level companies continue to maintain in-house a broad spectrum of the technologies relating to a wide range of components composing the aircraft engine. This proposition can be examined statistically by testing the revealed technological advantage (RTA) distributions in the two different time periods.

The second proposition refers to the changing degree of technological specialisation of the firms in the industry in question. As highlighted earlier in Section 5.2, not only are these companies active in a wide range of technological fields, but they are also spreading their capabilities in others. The variation of the degree of technological specialisation can be also assessed statistically.

65

The Herfindal index distribution was also calculated using the other two taxonomies (23-technological field and 7-technological field). The calculation produced results similar to those presented in Table 5.23.

113 The methodology proposed by Cantwell To test the two propositions the revealed technology advantage (RTA) distributions of each firm in two-ten year periods, namely 1977-1986 and 1987-1996, are compared. As described below, the RTA index reflects the firms’ degree of technological advantage with respect to other companies in the same industry. Following Cantwell (1989, 1991, 1993) the Galtonian regression model is used. This statistical methodology was first used in economics to study firms’ size distribution (Hart and Prais, 1956) and income distribution (Hart 1976). Recently Cantwell has applied the same methodology to analyse cross-sectional distribution of countries’ and companies’ technological activities.

The RTA index reflects the relative technological advantage of one firm with respect to others across a range of identified technological fields. This index has been used to study inter-country comparisons (Balassa, 1965; Soete, 1987; Cantwell, 1989; Laursen, 1998) as well as intra-industry and inter-firm comparisons (Patel and Pavitt, 1997; Cantwell, 1991, 1993). The RTA of a company in a specific technological field is given by its US patent share in that field, relative to the firm’s overall US patent share in all fields. In other words, the numerator of the RTA index is given by the number of US patents held by a firm in a technological field divided by the number of US patents in that technological field granted to companies in the same industry. The denominator is given by the firms’ total number of US patents in all technological fields divided by the number of US patents assigned to firms in all technological fields in the industry in question. In symbols: RTAij = (Pij/∑j Pij)/(∑i Pij/ ∑ij Pij) Where: Pij = number of US patents granted to firm j in sector i. This index varies around unity so that values greater than 1 reflect a comparative technological advantage of the firm in a particular technological field in respect to other firms in the same industry, whereas values less than 1 suggest a relative technological disadvantage. The use of this index overcomes some of the limitations of the use of patent statistics as a proxy measure of companies’ technological capabilities. As regards the different propensity to patent that characterises technological fields and firms, the RTA is normalised in the numerator for inter-field variation and for international inter-firm differences in the denominator. According to Cantwell (1991, 1993), there may still be inter-field variations in inter-firm propensity to patent. Some firms may be more likely to patent in some fields than others, and vice versa. However, Cantwell suggests that this intra-firm variance is lower than the inter-field difference. It can therefore be “hypothesised that on relatively large numbers the propensity to patent of a given firm in any sector [i.e. technological field] cannot be expected to have any systematic bias as compared to that firm’s notional average propensity to patent and the notional average propensity to patent of all firms in that sector [i.e. technological field]” (Cantwell, 1993: 221). The regression model used to compare the RTA indices in the two time periods ((t-1) =1977-1986 and t = 1987-1996) is the following:

114

RTA it = α + β RTA i(t-1) + ε it where i = technological field i t = time period t α, β = linear regression parameters The underlying assumption is that the regression is linear and the residual ε it is independent of RTA i(t-1). Further, such a regression model is valid if the RTA distribution conforms to a bivariate normal distribution 66. Also the RTA indices have been normalised in order to render them symmetric around unity, so that each RTA index varies between -1 and +1. Laursen (1998) argues that in regression analysis the RTA index must be always adjusted in a way that it becomes symmetric since an unadjusted RTA would lead to misleading results. Following Cantwell (1989, 1993), the temporal analysis of the distribution of the RTA index revolves around the values that the β coefficient takes. To put it in another way, the magnitude of β is a measure of the stability of the technological profile between the two periods in question. If β0. As explained later, however, a positive value of β still allows for some gradual evolution of the firm technological profile. When β=1 the ranking of the technological fields remain stable. Each technological field maintains the same proportional position in the sense that the relatively more important remain more important, and the relatively less important remain less important. Where β>1, instead, the relatively more important fields become even more important. The same trend holds for the less important fields. Where 0